Research ArticleInfectious diseases

Restoring glucose uptake rescues neutrophil dysfunction and protects against systemic fungal infection in mouse models of kidney disease

See allHide authors and affiliations

Science Translational Medicine  17 Jun 2020:
Vol. 12, Issue 548, eaay5691
DOI: 10.1126/scitranslmed.aay5691

Uremia undermines antifungal immunity

Systemic fungal infections are more prevalent and difficult to treat in patients with kidney disease. Jawale et al. set out to discern the mechanisms responsible for this enhanced risk and identify a way to mitigate disseminated fungal infections in kidney disease. Using multiple mouse models of kidney disease, they observed that uremia specifically conferred enhanced susceptibility to Candida albicans, which is typically controlled by neutrophils. Uremia blunted reactive oxygen species generation by neutrophils through glucose uptake perturbation. Accordingly, neutrophils isolated from patients with kidney disease were more capable of controlling fungal growth in vitro after dialysis. GSK3β inhibition restored the defect in mice or in human neutrophils and represents a potential intervention for patients with chronic kidney disease.


Disseminated candidiasis caused by the fungus Candida albicans is a major clinical problem in individuals with kidney disease and accompanying uremia; disseminated candidiasis fatality is twice as common in patients with uremia as those with normal kidney function. Many antifungal drugs are nephrotoxic, making treatment of these patients particularly challenging. The underlying basis for this impaired capacity to control infections in uremic individuals is poorly understood. Here, we show in multiple models that uremic mice exhibit an increased susceptibility to systemic fungal infection. Uremia inhibits Glut1-mediated uptake of glucose in neutrophils by causing aberrant activation of GSK3β, resulting in reduced ROS generation and hence impaired killing of C. albicans in mice. Consequently, pharmacological inhibition of GSK3β restored glucose uptake and rescued ROS production and candidacidal function of neutrophils in uremic mice. Similarly, neutrophils isolated from patients with kidney disease and undergoing hemodialysis showed similar defect in the fungal killing activity, a phenotype rescued in the presence of a GSK3β inhibitor. These findings reveal a mechanism of neutrophil dysfunction during uremia and suggest a potentially translatable therapeutic avenue for treatment of disseminated candidiasis.


Disseminated candidiasis, the third most common nosocomial infection, is caused by bloodstream infection of the opportunistic fungus Candida albicans (1, 2). Infections in general are a major cause of mortality in patients with kidney disease (responsible for 20% of deaths), especially those with end-stage kidney disease (ESKD) undergoing hemodialysis. Patients with ESKD have a higher risk of disseminated candidiasis (relative risk factor of 4.2) (3, 4). Moreover, the mortality rate from systemic fungal infection is twice as high in patients with ESKD than those with normal kidney function (3, 5, 6). Epidemiological data indicate that the higher mortality from disseminated candidiasis in patients with ESKD is not merely due to an increased fungal access to the bloodstream, i.e., from catheters. Rather, patients with ESKD have intrinsic immune defects, which are poorly understood. With an alarming rise in the incidence of chronic kidney disease (estimated to be 15% of the U.S. adult population in 2019) (7), the need for better therapeutic strategies to deal with infections in these patients is urgent and likely to increase with time.

ESKD is often accompanied by uremia, which is characterized by the accumulation of more than 900 metabolites in the blood. A subset of these metabolites (~100) negatively affects physiological systems and is termed uremic toxins (8, 9). Uremia is also linked to an increased risk for infectious disease, including disseminated candidiasis (1012). Seventy-seven percent of all infections in uremic patients occur without direct vascular access from invasive medical procedures (13). Thus, uremia itself is a predisposing factor for infections in ESKD independent of hemodialysis. Data from observational clinical studies suggest that uremic patients are more likely to die from these infections (14, 15). However, the mechanisms by which uremia impairs antifungal immunity are poorly defined.

During systemic candidiasis, an innate response dominated by neutrophils is the major driver of fungal clearance (1618). Accordingly, neutropenia is a substantial risk factor for infection in humans, and mice depleted of neutrophils are highly susceptible to disseminated candidiasis (16, 17, 19). Upon fungal recognition through Dectin-family receptors, neutrophils phagocytose C. albicans. The phagosomes fuse with lysosomes and neutrophil granules containing proteolytic enzymes and antimicrobial peptides (AMPs). Activated nicotinamide adenine dinucleotide phosphate oxidase (NADPH oxidase) generates reactive oxygen species (ROS), which, along with other oxidants, kill fungi (20). Our understanding of the infectious sequelae of neutrophil dysfunction in part derives from individuals with chronic granulomatous disease (CGD), a rare inherited defect in neutrophil development where infections are common (21). Although these studies yielded valuable insights, they do not reveal how neutrophil function is affected by uremia in individuals with an otherwise functional immune system.

Metabolic pathways are critical determinants of immune function, a topic that has received intense interest in recent years in the context of T cells (22, 23), but there has been far less emphasis on understanding metabolic regulation of neutrophil function. Neutrophils are metabolically distinct from other immune cells, because they are terminally differentiated, short-lived, and have comparatively few mitochondria (23, 24). Thus, unlike mitochondria-sufficient T cells, neutrophils rely solely on glycolysis for their function (25). In terms of fungal control mechanisms, metabolism of glucose by glycolysis and the pentose phosphate pathway (PPP) yields NADPH (26). Subsequently, NADPH oxidase catalyzes ROS production by the formation of superoxide through the transfer of an electron from NADPH to oxygen on the cell membrane, where fungal killing occurs (20, 27). Mice with CGD or with myeloperoxidase deficiency (where neutrophils cannot produce ROS) succumb to C. albicans infections that are typically sublethal (28). A combination of ROS and cations cause yeast cell death by inhibiting the fungal transcription factor Cap1 (29). Neutrophils also extrude neutrophil extracellular traps (NETs) by a ROS-dependent mechanism, and NETs target the hyphal form of C. albicans (30). The antimicrobial function of neutrophils is a metabolically demanding process, necessitating increased glucose uptake (31). In this regard, neutrophils express the Glut1 and Glut3 glucose transporters, belonging to a superfamily of 14 Gluts (32). Although Glut1 has been extensively studied in insulin-regulated cells such as muscle and adipose tissue (33), its roles in antifungal activity in ESKD are less clear.

Here, we took advantage of two mouse models of ESKD to assess susceptibility against disseminated candidiasis and reveal mechanisms by which uremia affects antifungal immunity. Knowledge gained from these studies was applied to develop preclinical therapeutic strategies to boost immune response against disseminated candidiasis, with implications for other fatal infections in ESKD.


Mice with aristolochic acid I–induced uremia showed increased susceptibility to disseminated candidiasis

To create a system in which we can interrogate the role for kidney disease or uremia on antifungal immunity, we combined a mouse model of chemically induced kidney impairment/uremia with an intravenous model of disseminated candidiasis. Aristolochic acid (AA) is found in Aristolochia and Asarum species of herbs and causes AA nephropathy (AAN) in humans (34). There are 11 known derivatives of AA, with AAI and AAII the most prevalent. Both are equally mutagenic, but only AAI causes nephrotoxicity. AAII was used as a non-nephrotoxic control for AAI throughout the study. A single intraperitoneal injection of AAI (10 mg/kg) drives kidney damage and fibrosis in C57Bl/6 mice (fig. S1A), as shown previously (35). Mice injected with AAI develop uremia as evidenced by elevated concentrations of serum blood urea nitrogen (BUN), creatinine, and uremic toxins starting at day 3 (fig. S1, B and C). Kidney pathology and BUN concentration peaked at 6 to 10 days (fig. S1, D and E), and ~100% mortality was observed by 14 days after AAI injection (fig. S1F), as shown before (34, 35). AAI-injected mice exhibited normal blood glucose and liver enzyme concentrations (fig. S1G and table S1).

To determine whether uremia in mice affects disseminated candidiasis, mice were given AAI, AAII, or phosphate-buffered saline (PBS) (control) followed by intravenous infection of C. albicans 4 days after AAI/AAII administration, a time point at which AAI-injected mice exhibited uremia without much weight loss (Fig. 1A and fig. S1H). At 48 hours post-infection (p.i.), AAI-injected animals (henceforth referred to as uremic mice) exhibited ~1.5 to 2 log higher fungal burdens in the organs than AAII-injected or control mice (Fig. 1B). Periodic acid–Schiff (PAS) staining of kidneys and brain revealed increased hyphal invasion in uremic animals (Fig. 1C). Uremic mice succumbed to infection (~95% mortality) by day 4, a time point at which AAII-injected and control mice were typically alive (Fig. 1D). Infected uremic mice demonstrated a higher BUN concentration than the uninfected uremic group (Fig. 1E). Thus, mice with AAI-induced uremia show heightened susceptibility to systemic fungal infection and therefore serve as a valid model of this pathology in humans.

Fig. 1 Uremic mice show increased susceptibility to disseminated candidiasis.

(A) Schematic representation of the experimental plan. C57Bl/6 male mice were subjected to AAN. On day 4 (d4) after AAI injection, mice were infected with 1 × 105 CFU C. albicans. (B) Fungal burdens were measured at 48 hours p.i. (n = 6 to 7). (C) Kidney and brain sections were stained with PAS to assess hyphal invasion and infiltration of inflammatory cells (black arrows). Representative images from one of three mice per group. Magnification, ×40. (D) Survival was evaluated over a period of 4 days p.i. (n = 8 to 11). (E) Serum BUN concentrations were measured at 48 hours p.i. (n = 6 to 7). (F) Uremic, control, and AAII mice (n = 6) were systemically infected with MRSA at day 4 after AAI injection, and bacterial colony counts were determined at 6 hours p.i. Data pooled from at least two experiments for (B), (D), (E), and (F) and expressed as means ± SD (B, E, and F). Statistical analyses by log-rank test (D), and one-way ANOVA (B, E, and F). ns, statistically not significant.

Uremic patients are also at higher risk for other bloodstream infections including Staphylococcus aureus (36). Uremic mice showed significantly higher methicillin-resistant S. aureus (MRSA) burdens than control animals (Fig. 1F). Thus, enhanced susceptibility of uremic mice is not only limited to disseminated candidiasis.

Uremia causes defective fungal clearance in disseminated candidiasis

C. albicans use urea as a source of nitrogen (37). Hence, we asked whether uremia has any direct impact on fungal growth in AAI-injected mice. Uremic mice were infected with a strain of C. albicans lacking urea amidolyase (Dur1,2−/−), an enzyme that enables the fungus to use urea as a nitrogen source (Fig. 2A). Uremic mice infected with this strain still exhibited increased fungal burdens similar to the control strain (Fig. 2B), suggesting that uncontrolled infection in uremia is not due to the ability of C. albicans to metabolize excess urea for growth. An alternative explanation for the above observations is that AAI directly affects immunity. To rule out this possibility, uremic mice were subjected to infection and treated with probenecid, an organic anion transporter inhibitor that limits damage to renal tubular epithelial cells and prevents uremia without neutralizing AAI (38, 39). Probenecid-treated uremic mice showed reduced BUN and lower fungal loads than untreated uremic animals, arguing against a direct effect of AAI on the immune system (Fig. 2C and fig. S2, A to C).

Fig. 2 Uremia is critical for increased susceptibility to fungal infection.

(A) Schematic diagram of the experimental design. Uremic and control C57Bl/6 mice (n = 5 to 6) were infected with Dur1,2−/− and Dur1,2rev C. albicans strain. (B) Mice were evaluated for fungal loads at 48 hours p.i. (C) Groups of AAI-injected mice were either treated with probenecid (AAI + PRB) or left untreated. Control mice received probenecid only (PRB). Four days later, mice were subjected to infection and fungal burdens were measured at 48 hours p.i. (n = 4 to 6). C57Bl/6 mice (n = 4 to 5) were either subjected to 5/6 nephrectomy (5/6 Nx) or sham-operated (non–5/6 Nx). Three weeks later, mice were infected, and (D) fungal loads, (E) percentage weight loss, and (F) BUN were quantified at 48 hours p.i. (G) WT mice (n = 4) were either subjected to UUO or sham-operated (control). One week after surgery, mice were infected and fungal colonies were enumerated at 48 hours p.i. Pooled results from at least two experiments for (B) to (G) and expressed as means ± SD. Statistical analyses by Student’s t test (B, D, E, and G) and one-way ANOVA (C and F).

We next assessed fungal burdens in a surgically induced mouse model of uremia. The model of 5/6 nephrectomy involves excision of the upper and lower poles of one kidney followed by nephrectomy of the converse kidney (fig. S2, D and E) (40). The 5/6 nephrectomized mice exhibited increased fungal loads, weight loss, and BUN compared to sham-nephrectomized animals after C. albicans infection (Fig. 2, D to F).

To distinguish between the effects of kidney damage on fungal susceptibility independent of uremia, we used the unilateral ureteral obstruction (UUO) model, in which one of the ureters is ligated and the converse ureter is left intact, allowing for normal renal filtration (41). Consequently, UUO causes unilateral kidney inflammation and tissue damage but not uremia (fig. S2, F and G). UUO mice infected with C. albicans showed comparable fungal burdens to the non-UUO group, supporting the hypothesis that increased susceptibility to infection is due to uremia and not kidney damage per se (Fig. 2G).

Uremia impairs the antifungal activity of neutrophils by inhibiting ROS production

During infection, a robust innate response dominated by neutrophils is essential to the clearance of C. albicans (17, 18). To test whether neutrophils play a similar antifungal role in uremia, we used MRP8-cre mice crossed with diphtheria toxin receptor (DTR) transgenic mice to generate mice with DTR expression restricted mostly to neutrophils (PMNDTR), as described before (42). Diphtheria toxin injection depleted neutrophils in the bone marrow (BM) and spleen of PMNDTR mice (fig. S3A) (42). Neutrophil-depleted uremic mice showed increased fungal burdens compared to nondepleted uremic animals (Fig. 3A). Uremic mice showed similar infiltration of neutrophils and monocytes/macrophages to organs after infection (Fig. 3B and fig. S3B). This observation is unexpected because kidney-infiltrating neutrophils expressed marginally less CD11b, an integrin responsible for transendothelial migration (43) (fig. S3C). Moreover, BM neutrophils exhibited reduced in vitro migration in the presence of uremic serum (fig. S3D), as reported before (44). These results indicate that uncontrolled fungal loads and resulting increased inflammatory cytokines production (fig. S3E) may have compensated for the defect in neutrophil migration in mice. Granulopoiesis was unaffected under uremic conditions (fig. S3F).

Fig. 3 Uremia impairs ROS production by neutrophils.

(A) PMNWT and PMNDTR mice were given PBS (control) or AAI (n = 4 to 5) followed by two injections with diphtheria toxin (DT) at days 3 and 4 after AAI injection. Mice were infected with 2.5 × 104 CFU of C. albicans at day 4 after AAI injection and evaluated for fungal loads at 30 hours p.i. (B) Percentage of neutrophils (live CD45+Ly6G+CD11b+) in the organs of uremic, control, and AAII mice (n = 4) was quantified by FACS at 24 hours p.i. (C) C57Bl/6 BM neutrophils were assessed for in vitro fungal killing activity in the presence of 50 and 10% uremic, control, and AAII serum at 180 min. (D) Uremic, control, and AAII-injected mice (n = 6 to 7) were infected with 5 × 106 CFU of streptavidin AF633-dTomato+ C. albicans and assessed for in vivo fungal killing in the kidney-infiltrating neutrophils at 2 hours p.i. Representative plots show live (dTomato+AF633+) and killed C. albicans (dTomatoAF633+) within the Ly6G+CD11b+ population. (E) ROS production by C57Bl/6 BM neutrophils ± uremic or control or AAII serum ± C. albicans was measured by FACS at 180 min. (F) Isolated neutrophils from the spleens of uremic, control, and AAII mice were stimulated with C. albicans, and ROS production was assessed at 180 min. (G) In vivo ROS production by neutrophils was quantified in the organs of mice (n = 4) at 24 hours p.i. In vitro ROS (norm) = MFI of DHR in C. albicans stimulated by unstimulated cells; in vivo ROS (norm) = MFI of DHR in C. albicans infected by uninfected mice. The data are pooled from at least two experiments for (A), (B), (D), and (G) and three to four experiments for (C), (E), and (F) and expressed as means ± SD. Statistical analyses by one-way ANOVA (B to G) and by Student’s t test (A).

We assessed the in vitro fungal killing capacity of neutrophils in the presence of 50% serum from AAI-injected, AAII-injected, or control mice (henceforth denoted to as AAI or uremic, AAII, and control serum, respectively). BM neutrophils exhibited diminished fungicidal activity in the presence of AAI serum (Fig. 3C), which was not due to any defect in in vitro phagocytosis, apoptosis, or AMP gene expression (fig. S3, G to I). There was also no direct impact of AAI on fungicidal activity of neutrophils or fungal growth, when used at a concentration seen in the serum of AAI-injected mice (~10 μg/ml) (fig. S4, A and B) (38).

Next, we infected uremic, control, or AAII-injected mice with a streptavidin AF633-dTomato+ C. albicans reporter strain to visualize in vivo fungal killing. In uremic mice, kidney-infiltrating neutrophils showed diminished intracellular fungal killing compared to controls or AAII-injected animals (Fig. 3D). In contrast to results from the in vitro phagocytosis assay, there was a modest decrease in phagocytosis by neutrophils in uremic mice (Fig. 3D). These data suggest that uremia inhibits the candidacidal function of neutrophils and may have some impact on phagocytosis as well.

Upon fungal recognition and phagocytosis, NADPH oxidase–mediated release of ROS is essential for the elimination of C. albicans (30). BM neutrophils incubated in the presence of AAI serum demonstrated reduced ROS production compared to those incubated with control or AAII serum (Fig. 3E). Neutrophils isolated from AAI-injected mice also showed diminished ROS (Fig. 3F). Consistently, neutrophils from infected uremic mice demonstrated compromised ROS generation (Fig. 3G). AAI at a concentration seen in uremic mice showed no effect on ROS generation by neutrophils (fig. S4C). The deficiency is restricted to the cytoplasmic ROS, because mitochondrial ROS in neutrophils was comparable between the groups (fig. S4D).

We also investigated NET formation, which is partially dependent on the ROS production, during uremia. As reported before (45), we observed an increase in spontaneous NETs in the presence of uremic serum without C. albicans (fig. S4E). In the presence of fungus, a large number of neutrophils incubated with either control or AAII serum formed NETs while swarming around hyphae under in vitro condition. During uremia, NETs not only were reduced but also were formed far away from hyphae. These results suggest that spontaneous NET formation in uremia occurs remotely and may not be sufficient for hyphal clearance.

Uremia inhibits Glut1-mediated glucose uptake by neutrophils

Activated neutrophils consume oxygen and produce ROS in a metabolically demanding process, known as respiratory burst (27). The NADPH required for ROS generation is produced by the breakdown of glucose via glycolysis and PPP (26). Neutrophils incubated with AAI serum showed impaired glycolysis, as evidenced by reduced adenosine 5′-triphosphate and lactate (Fig. 4, A and B). Selective inhibition of glycolysis but not fatty acid oxidation or oxidative phosphorylation diminished fungicidal activity and ROS production by neutrophils (Fig. 4, C and D, and fig. S5, A and B). The rapid antimicrobial function of neutrophils is supported by increased glucose uptake (31), and uremic serum inhibited basal glucose uptake by neutrophils (Fig. 4E). Accordingly, kidney-infiltrating neutrophils of infected uremic mice showed compromised glucose incorporation than nonuremic animals (Fig. 4F). These results imply that uremia impairs glycolysis and PPP by limiting glucose uptake, an event upstream of ROS production (46).

Fig. 4 Reduced glucose uptake in neutrophils during uremia.

C57Bl/6 BM neutrophils + AAI or control or AAII serum + C. albicans were evaluated for (A) intracellular adenosine 5′-triphosphate (ATP) and (B) lactate production at 60 and 180 min, respectively. Neutrophils ± 2-DG (at indicated concentrations) ± C. albicans were assessed for (C) in vitro fungicidal activity and (D) ROS production at 180 min. (E) Basal glucose uptake by neutrophils was quantified in the presence of AAI, control, or AAII serum at 90 min. (F) In vivo glucose uptake by neutrophils was measured in the kidneys of mice at 24 hours p.i. Neutrophils + AAI or control or AAII serum were evaluated for (G) Glut1 and Glut3 protein by immunoblot at 180 min and quantified by ImageJ software. Representative image of one of four and three experiments for Glut1 and Glut3, respectively. 2-NBDG, 2-[N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino]-2-deoxyglucose. (H) C57Bl/6 BM neutrophils were evaluated for fungal killing activity in the presence of increased concentrations of glucose in uremic or control serum. (I) In vitro fungicidal activity and (J) ROS production were assessed in neutrophils ± C. albicans ± WZB117 at 180 min. WT mice (n = 7) were treated with WZB117 and evaluated for (K) fungal burdens and (L) ROS production by kidney-infiltrating neutrophils at 48 and 24 hours p.i., respectively. In vitro ROS (norm) = MFI of DHR in C. albicans stimulated by unstimulated cells; in vivo ROS (norm) = MFI of DHR in C. albicans infected by uninfected mice. Results pooled from three to four experiments for (A) to (E) and (G) to (J) and at least two independent experiments for (F), (K), and (L) and expressed as means ± SD. Statistical analyses by Student’s t test (D, J, K, and L) and one-way ANOVA (A to C and E to I).

The primary glucose transporters expressed in neutrophils are Glut1 and Glut3 (encoded by Slc2a1 and Slc2a3, respectively). Uremic serum suppressed Glut1 but not Glut3 protein expression (Fig. 4G). However, neutrophils incubated with uremic or control serum showed comparable expression of Slc2a1 and Slc2a3 mRNA (fig. S5C), suggesting that uremia negatively regulates Glut1 at the protein level. Increasing glucose concentration in the media did not rescue the defect in antifungal activity of neutrophils in the presence of uremic serum (Fig. 4H). To determine the contribution of Glut1 during infection, neutrophils were treated with a Glut1 inhibitor (WZB117) (47). WZB117-treated BM neutrophils showed reduced fungicidal activity and ROS production (Fig. 4, I and J). Accordingly, WZB117-treated mice demonstrated an increase in fungal loads and reduction in ROS production by neutrophils compared to controls (Fig. 4, K and L, and fig. S5D). These results indicate that uremia inhibits Glut1-mediated glucose uptake by neutrophils to impair antifungal immunity.

Uremia causes aberrant activation of GSK3β

On the basis of a report showing a diminished expression of pAKTSer473 in BM stromal cells in uremic mice (48), we measured AKT phosphorylation (pAKT) in neutrophils incubated in the presence of uremic or control serum. AAI serum inhibited pAKTSer473 in neutrophils (Fig. 5A and fig. S5E). Inhibition of phosphatidylinositol 3-kinase (PI3K) diminished fungicidal activity and ROS production by neutrophils (Fig. 5, B and C). Wild-type (WT) mice treated with a pan-PI3K inhibitor showed increased fungal burdens compared to untreated animals (Fig. 5D), thus signifying a vital role for PI3K/AKT signaling in fungal clearance. One of the ways that PI3K/AKT regulates metabolism is through inhibitory phosphorylation of glycogen synthase kinase 3 beta (GSK3β), a serine protease that regulates glycogen metabolism, cell cycle progression, and mammalian target of rapamycin (mTOR) signaling (49). We observed that AAI serum suppressed the inhibitory phosphorylation of GSK3 isoform GSK3β (pGSK3βSer9), but not GSK3α (pGSK3αSer21) (Fig. 5, E to G, and fig. S5F). Thus, GSK3β is aberrantly activated in neutrophils in the presence of uremic serum. Heat inactivation of AAI serum did not alter PI3K/AKT signaling and downstream GSK3β activation in neutrophils (fig. S5G), indicating that protein-based inflammatory mediators in uremia is not responsible for the defect in PI3K/AKT/GSK3β pathway in neutrophils. GSK3β is a negative regulator of mTOR function (49). Uremic serum down-regulated the phosphorylation of ribosomal protein S6 kinase beta-1 (pS6K1Thr389) and eIF4E-binding protein 1 (p4E-BP1Thr37/46), but not pPKCαSer657, indicating that mTORC1 but not mTORC2 function is compromised in neutrophils during uremia (Fig. 5, H and I, and fig. S5, H and I) (50).

Fig. 5 Aberrant activation of GSK3β in neutrophils in uremia.

C57Bl/6 BM neutrophils + AAI or control or AAII serum were analyzed for pAKTSer473 expression by (A) FACS and immunoblot at 45 min. Histogram and images represent of one of four and three experiments, respectively. Neutrophils ± pan PI3K inhibitor (at the indicated concentrations) ± C. albicans were assessed for (B) fungicidal activity and (C) ROS generation. (D) C57Bl/6 WT mice (n = 6) were treated with LY294002 and evaluated for fungal burden at 48 hours p.i. Neutrophils + AAI or control or AAII serum were analyzed for (E and G) pGSK3βSer9 expression by immunoblot and FACS, (F) pGSK3αSer21, (H) pp70S6KThr389, and (I) p4E-BP1Thr37/46 by immunoblot at 45 min. Immunoblot images represent one of three experiments. WT mice (n = 6) were either treated with Torin1 or left untreated. Mice were evaluated for (J) fungal burdens and (K) in vivo glucose uptake by the neutrophils in the kidneys. In vitro ROS (norm) = MFI of DHR in C. albicans–stimulated/DHR MFI-unstimulated cells. Pooled data from three to four experiments for (A) to (C) and (E) to (I) and two experiments for (D), (J), and (K) expressed as means ± SD. Statistical analyses by Student’s t test (C, D, and J) and one-way ANOVA (A, B, G, and K).

To define the in vivo significance of mTOR activation in antifungal immunity, WT mice were treated with the mTOR inhibitor Torin 1 and subjected to systemic infection. Torin 1–treated mice exhibited ~0.5 to 1 log higher fungal loads than untreated groups (Fig. 5J), and kidney-infiltrating neutrophils in treated mice showed diminished glucose uptake (Fig. 5K). These observations highlight an essential role for PI3K/AKT/GSK3β/mTORC1-axis in restraining Glut1-mediated glucose uptake in neutrophils during uremia.

Inhibition of GSK3β activation restores the antifungal activity of neutrophils in uremia

Granulocyte-macrophage colony-stimulating factor (GM-CSF), which induces ROS by directly activating p47phox (51), failed to rescue the defect in ROS production in the presence of uremic serum (Fig. 6A). We reasoned that blocking GSK3β in neutrophils during uremia might be an alternate strategy to rescue neutrophil function by exploiting the activity of a clinically approved GSK3β inhibitor, lithium chloride (LiCl) (52). In vitro application of LiCl restored pGSK3βSer9 and Glut1 expression, glucose incorporation, and candidacidal function of neutrophils in the presence of uremic serum, with no impact on fungal morphology or replication (Fig. 6, B to D and F, and fig. S6, A and B). Treatment of neutrophils with a more specific GSK3β inhibitor, SB415286 (53), similarly restored glucose uptake and antifungal activity of neutrophils (Fig. 6, E and F). LiCl-treated WT mice demonstrated increased glucose uptake by neutrophils and reduced fungal loads compared to untreated mice (Fig. 6, G and H). These results imply that treatment with GSK3β inhibitors can restore neutrophil function during uremia.

Fig. 6 GSK3β inhibitors rescue neutrophil dysfunction in uremia.

(A) C57Bl/6 BM neutrophils + AAI or control serum ± GM-CSF (10 ng/ml) ± C. albicans were evaluated for ROS production at 180 min. In vitro ROS (norm) = MFI of DHR in C. albicans stimulated divided by unstimulated cells. Neutrophils + AAI or control serum ± LiCl were assessed for (B) pGSK3βSer9 levels, (C) Glut1 expression, and (D) glucose uptake by the neutrophils. (E) Basal glucose uptake by neutrophils was measured after treatment with the GSK3β inhibitor SB415286. (F) Antifungal activity of neutrophils + AAI or control serum ± LiCl or SB415286 + C. albicans was measured at 180 min. WT mice (n = 4 to 5) were either treated with LiCl or left untreated. (G) Fungal burdens and (H) glucose uptake by neutrophils in the kidneys were assessed at 48 hours p.i. GSK3β expression was measured in (I) purified neutrophils from BM and (J) neutrophils in the spleen of PMNΔGSK3β mice by immunoblot and FACS analyses, respectively. (K) PMNΔGSK3β and control mice (n = 5 to 8) were subjected to AAN. A cohort of PMNΔGSK3β or control mice (n = 5 to 6) received PBS. Mice were infected with C. albicans and fungal loads were quantified at 48 hours p.i. The histograms (B and J) and images (C and I) represent one of two to four experiments. For (C), lanes were run on the same gel but were nonconsecutive (splicing indicated by black dashed line). The data are pooled from three experiments for (A), (B), and (D) to (F) and two experiments for (G), (H), and (K), and expressed as means ± SD. Statistical analyses by Student’s t test (G and H) and one-way ANOVA (A, B, D, E, F, and K).

To determine whether the effects of GSK3β are specific to neutrophils, we crossed MRP8-Cre-ires/GFP and Gsk3βfl/fl mice to delete Gsk3β in neutrophils (PMNΔGSK3β) (42). PMNΔGSK3β mice successfully lacked GSK3β in neutrophils from BM and spleen (Fig. 6, I and J, and fig. S7A). The absence of GSK3β had no effect on granulopoiesis or the number of blood neutrophils in PMNΔGSK3β mice (fig. S7, B and C). Uremic PMNΔGSK3β mice showed ~1.5 to 2 log reduction in the fungal burdens compared to uremic PMNGSK3β controls (Fig. 6K). Neutrophil tissue infiltration was similar between uremic PMNΔGSK3β and control groups (fig. S7D).

GSK3β blockade showed preclinical efficacy in improving the function of mice and human neutrophils in uremia

We assessed the preclinical efficacy of LiCl in uremic mice after disseminated candidiasis. To that end, C. albicans–infected uremic mice were either treated with LiCl or left untreated (Fig. 7A). LiCl-treated mice showed reduced fungal burdens and had prolonged survival compared to untreated uremic animals (Fig. 7, B and C). LiCl treatment increased both ROS production and glucose uptake by neutrophils in uremic mice (Fig. 7, D and E).

Fig. 7 LiCl treatment shows preclinical efficacy in correcting neutrophil defects.

(A) Schematic representation of the experimental design. Uremic mice (n = 5 to 7) were either treated with LiCl starting day 0 (relative to infection) and then daily for the next 2 to 4 days or left untreated. Mice were assessed for (B) fungal burdens at 48 hours p.i., (C) survival (n = 7 to 9), (D) ROS production, and (E) glucose uptake by neutrophils. (F) In vitro fungal killing activity by healthy donor neutrophils (n = 2) incubated with paired pre- or postdialysis sera (n = 9) at 180 min. (G) ROS generation by healthy donor (n = 2) neutrophils incubated with pre- or postdialysis sera (n = 10) or serum of healthy individuals (n = 10) ± C. albicans at 180 min. (H) Predialysis neutrophils ± LiCl and postdialysis neutrophils (n = 7) were assessed for fungicidal activity in the presence of healthy serum + C. albicans at 180 min. In vitro ROS (norm) = MFI of DHR in C. albicans stimulated by unstimulated cells; in vivo ROS (norm) = MFI of DHR in C. albicans infected by uninfected mice. Pooled data from two to three experiments for (B) to (E) expressed as means ± SD. Statistical analyses by one-way ANOVA (B, D, E, and H), log-rank test (C), and Student’s t test (F and G).

Several in vitro studies have identified a role for uremia in neutrophil dysfunction (5458). However, these studies used polyclonal activators rather than assessing anti–C. albicans activity in the setting of uremia (59, 60). Hence, we evaluated the candidacidal function of healthy human donor neutrophils in the presence of serum from patients with ESKD receiving hemodialysis. Serum samples were obtained immediately predialysis and 2 hours after dialysis and were used in a paired fashion to avoid confounders. Healthy donor neutrophils incubated with postdialysis serum showed markedly improved fungicidal activity compared to incubation with predialysis serum (Fig. 7F). Healthy donor neutrophils in the presence of predialysis serum showed diminished ROS compared to neutrophils incubated with serum from healthy volunteers (Fig. 7G). Moreover, there was an increase in ROS production in healthy donor neutrophils when incubated with postdialysis serum (Fig. 7G).

We also isolated neutrophils from patients pre- or postdialysis and assessed their antifungal activity with or without LiCl. As shown in Fig. 7H, postdialysis neutrophils showed improved candidacidal activity compared to predialysis neutrophils. In line with our mouse model studies, LiCl treatment improved the antifungal activity of predialysis neutrophils. Thus, our data demonstrate the preclinical therapeutic effectiveness of GSK3β inhibitors in reversing mice and human neutrophil dysfunction and limiting fungal infection in uremia.


Features of a chronically activated immune system are linked to several inflammatory complications in patients with ESKD and commonly observed comorbidities including cardiovascular disease (61). On the other hand, patients with ESKD mount poor immune response to infections (62). Whereas much emphasis has been placed on understanding the mechanisms of inflammatory diseases in these patients, unexpectedly little is known about immune deficiencies that can induce susceptibility to infections. The higher disseminated candidiasis mortality rate in patients with ESKD has long been known but never understood at a fundamental mechanistic level. Using two clinically relevant mouse models of kidney disease, namely, AAI nephropathy and 5/6 nephrectomy, we show that uremia causes a neutrophil-specific defect in ROS production by inhibiting cellular metabolic pathways, which are, in turn, critical for antifungal immunity (fig. S8). We also show that similar defect in antifungal function of neutrophils exists in a mouse model of MRSA infection. Thus, the results from this study have broader implications for other infectious diseases including MRSA, which is a major cause of mortality in patients with ESKD and undergoing hemodialysis.

The identity of uremic toxin(s) that inhibits the antifungal activity of neutrophils is unknown. In this context, a previous study showed that fibroblast growth factor 23 (FGF23) limits neutrophil recruitment and impairs host defense against pulmonary bacterial infection in chronic kidney disease (CKD) (63). In contrast, we did not see any defect in neutrophil recruitment in target organs. The reasons for this discrepancy are currently unclear. Increased fungal load and resulting inflammatory milieu may have negated the inhibitory effect of FGF23 on neutrophil recruitment. Despite this discordance, FGF23 treatment resulted in diminished ROS production from neutrophils. However, heat inactivation of uremic serum was unable to rescue the defect in PI3K/AKT/GSK3β pathway, indicating that FGF23 or other protein mediators of the uremic serum are not responsible for the defect in neutrophil function. Thus, these data point toward organic solutes as potential culprits for causing neutrophil dysfunction by suppressing the PI3K/AKT pathway. Organic solutes such as guanidines, indoxyl sulfate, para-cresyl sulfate, para-cresyl glucuronide, and indole-3-acetic acid have been linked to causing immune dysfunction in human settings but never studied at a fundamental mechanistic level (6466). Hence, comparing the concentrations of uremic toxins between pre- and postdialysis serum will provide a unique opportunity to study potential uremic toxin(s) in predialysis serum with inhibitory effect on PI3K/AKT/GSK3β signaling in neutrophils. Understanding the specific inhibitory molecules in the serum of patients with chronic kidney disease may lead to the improvement of dialysis techniques to better remove these compounds and improve immunity.

The reagents to detect ROS and glucose are commercially available as fluorescein isothiocyanate conjugated only, thus precluding in vivo detection in green fluorescent protein (GFP)–expressing PMNΔGSK3β mice. Therefore, we examined ROS and glucose uptake in WT mice treated with LiCl. We show that uremia causes a neutrophil-specific defect in ROS production by inhibiting the glycolytic pathway, which is required for NADPH production. In vascular smooth muscle and endothelial cells, PI3K activity triggers assembly and activation of NADPH oxidase (67). One report suggests a role for PI3Kγ in chemoattractant-induced superoxide production in mouse neutrophils (68). Thus, it is plausible that uremia inhibits the assembly of functional NADPH oxidase. However, GM-CSF, a cytokine that can directly activate NADPH oxidase without the need for glucose uptake for glycolysis or PPP, failed to rescue the defect in ROS generation in uremia (51). Hence, these data indicate that the defect in ROS production cannot be attributed to the impairment of NADPH oxidase assembly and function in uremia. Whereas ROS generation is inhibited by impaired glucose uptake, phagocytosis is only modestly affected. A previous study has shown that energy requirements for neutrophils during phagocytosis are exclusively met by the breakdown of stored glycogen to glucose, and chemotaxis and antimicrobial activities of neutrophils rely on the uptake of glucose (69). Thus, it is expected to see that neutrophils exhibit a modest defect in phagocytosis, albeit ROS production was severely affected during uremia.

Unlike Glut4, which is insulin sensitive, expression of Glut1 at the cell membrane is not regulated by blood glucose concentrations (70). Rather, Glut1 expression is controlled by transcription, translation, and/or translocation from an intracellular pool. Here, we show that uremia negatively regulates Glut1 expression and consequently glucose uptake in neutrophils by modulating GSK3β/mTORC1/Glut1 axis. In the insulin pathway, serum- and glucocorticoid-inducible kinase 1, a kinase downstream of PI3K, enhances the translocation of Glut1 from intracellular vesicle to the plasma membrane without altering the total protein levels (71). Thus, uremia could also prevent the cell surface localization of Glut1 in neutrophils. We did not investigate the status of PKCβ1- and PKCδ-driven Glut1 phosphorylation in neutrophils, known to play an important role in cell surface localization of Glut1 and glucose uptake (72). We show that WT mice treated with a Glut1 inhibitor demonstrate increased fungal loads, but because Glut1 is ubiquitously expressed, a neutrophil-specific role for Glut1 in antifungal immunity is unknown. We have used only male mice to define the mechanisms of neutrophil dysfunction in uremia. Thus, the effect of uremic toxins on antifungal immune response between male and female mice remains to be elucidated.

The short- and long-term impact of uremia on human neutrophil function is poorly understood. The limited availability of human neutrophil samples from dialysis patients precludes elaborate cellular and molecular characterizations of these innate cells in uremic patients in this study. The aberrant activation of GSK3β and neutrophil dysfunction in uremia suggests a new approach for treating or preventing systemic candidiasis in kidney disease. Because complete removal of uremic toxins cannot be achieved by hemodialysis, boosting immunity is an attractive alternative to prevent infections in these patients. Our data suggest that targeting the GSK3β/Glut1 pathway may lead to safe, inexpensive, and rapidly implementable treatments for disseminated candidiasis in patients with ESKD because drugs targeting this pathway such as LiCl are already in use. Long-term treatment with LiCl causes adverse effects including nephrotoxicity (73), but the use of more specific GSK3β inhibitors may help circumvent the problem of toxicity. Thus, these studies may provide justification for future clinical studies of GSK3β inhibitors to treat or prevent bloodstream infections in patients with ESKD.


Study design

The objectives of this study were (i) to develop a mouse model suitable to study susceptibility to disseminated candidiasis in ESKD, (ii) to define mechanisms of immune dysfunction in the context of kidney disease, and (iii) to exploit this knowledge for therapeutic benefit in a preclinical setting. To test this objective, we combined two clinically relevant mouse models of kidney disease with a model of disseminated candidiasis. Groups of C57BL/6 mice with or without kidney injury were either systemically infected with C. albicans or left uninfected. Mice were evaluated for survival, fungal loads, neutrophil function at various days p.i. We used male mice for all the experiments described. The study size for mouse experiments was selected on the basis of power calculations of historical data from our laboratory. The results from mouse studies were validated with samples from a cohort of patients with ESKD undergoing hemodialysis. For studies involving human biospecimens, sample size is determined on the basis of our pilot study. Each in vitro and in vivo experiment was repeated at least two to three times (as specified in the figure legend) to ensure reproducibility, and no outliers were excluded. Investigators were not blinded. Primary data are reported in data file S1.


Eight-week-old male WT C57Bl/6NTac mice were directly purchased from Taconic Biosciences Inc. ROSA26iDTR mice and MRP8-Cre+ mice were purchased from the Jackson laboratory. MRP8-Cre+ mice were crossed with ROSA26iDTR mice to generate PMNDTR and PMNWT littermate controls. To generate mice with conditional deletion of GSK3β in neutrophils, we crossed the Gsk3βfl/fl mouse (obtained from F. Cambi, University of Pittsburgh) with MRP8-Cre+ mice. All mice were housed under specific pathogen–free conditions under the supervision of Division of Laboratory Animal Resources. All animal experiments were conducted following the National Institutes of Health guidelines under protocols approved by the University of Pittsburgh Institutional Animal Care and Use Committee.

Mouse models of AAN

C57Bl/6 mice were injected intraperitoneally with AAI (10 mg/kg) (Sigma) in PBS once. Mice in non-nephrotoxic control group were intraperitoneally injected with AAII (10 mg/kg) (Sigma), and control animals received similar volume of PBS.

Human biospecimens

Pre- and postdialysis serum samples (n = 10) were obtained as part of a study previously approved by the Maine Medical Center Institutional Review Board. Additional blood samples were obtained immediately pre- and postdialysis from a cohort of patients with ESKD (n = 7) undergoing chronic hemodialysis at University of Pittsburgh Medical Center (UPMC) In-patient Dialysis Unit. All studies were approved by the University of Pittsburgh Institutional Review Board. The samples were obtained after providing written informed consent. Healthy donor serum samples (n = 10) were obtained as part of a study entitled “Banking of Biological Samples and Collection of Clinical Data for Connective Tissue Disease” at the Division of Rheumatology and Clinical Immunology, University of Pittsburgh. All demographic information of the study participants and cause of ESKD are provided in tables S2 and S3. Patients treated with immunosuppressive therapy were excluded from the study.

Disseminated candidiasis

C. albicans (strain SC5314) was grown in yeast extract-peptone-dextrose (YPD) broth at 30°C for 18 to 24 hours in shaker incubator. Dur1,2−/− and Dur1,2rev C. albicans were kindly provided by K. W. Nickerson, University of Nebraska (37). Mice were infected with 0.25 × 105 to 1 × 105 C. albicans yeast cells via the lateral tail vein and daily monitored for signs of illness, weight loss, and survival. Formalin-fixed tissue sections were stained with PAS for visualization of hyphae.

MRSA infection

Mice were infected with 1 × 107 colony-forming units (CFU) MRSA (strain USA300) (provided by J. F. Alcorn, University of Pittsburgh) via the lateral tail vein, and bacterial loads in the liver, kidneys, and spleen were assessed at 6 hours p.i. (74).

In vitro assays

C57Bl/6 BM neutrophils (0.5 × 105 to 1 × 105 cells per well) were incubated with C. albicans (1 × 104 to 2 × 104 per well) in the presence of AAI, control, or AAII (50%) heat-inactivated serum.

Fungal killing assay and ROS measurement. BM neutrophils were incubated with C. albicans for 180 min. The number of surviving yeasts was assessed by plating serial dilutions on YPD agar at 180 min after incubation. The percent killing of C. albicans was expressed as [1 − (CFU of C. albicans in the presence of neutrophils/CFU of C. albicans cells in the absence of neutrophils)] × 100. The production of ROS was measured using the Dihydrorhodamine 123 (DHR) (Invitrogen). Results are expressed as normalized ROS [ROS (norm)], which is calculated by dividing the mean fluorescence intensity (MFI) of DHR in C. albicans stimulated by unstimulated sample.

Glucose uptake. C57Bl/6 BM neutrophils were assayed for in vitro glucose uptake in the presence of AAI or control serum at 90 min. Briefly, 2-deoxyglucose (2-DG) was added to the cells and, 10 min later, 2-DG uptake was measured using Glucose Uptake-Glo Assay (Promega Corp.).

In vivo experiments

ROS measurement. For ROS measurement in tissue-infiltrating neutrophils, DHR reagent was added to tissue cell suspension and incubated at 37°C for 30 min and evaluated for ROS by fluorescence-activated cell sorting (FACS).

Glucose uptake. Mice were injected (intravenously) with 2-[N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino]-2-deoxyglucose (0.5 mM), and 30 min later, organs were harvested and analyzed for glucose uptake by FACS.

Inhibitors for in vivo studies

Organic anion transporter 1 and 3 inhibitor. Mice were intraperitoneally injected with Probenecid (Invitrogen) [150 mg/kg i.p. (intraperitoneally)] or vehicle control (PBS) once at the day of AAI injection (39).

Glut1 inhibitor. Mice were intraperitoneally injected with WZB117 (Sigma) (10 mg/kg per day i.p.) or vehicle control [5% dimethyl sulfoxide (DMSO) + PBS] starting at day 0 (relative to infection) and daily for next 2 days (47).

GSK3β inhibitor. Mice were treated (intraperitoneal injection) with LiCl (40 mg/kg per day) or PBS control starting at day 0 (relative to infection) and daily for next 2 days (for fungal load) or 4 days (for survival) p.i. (75).

mTOR inhibitor. Mice were intraperitoneally injected with Torin1 (Selleckchem) (20 mg/kg per day) or vehicle control [20% N-methyl-2-pyrrolidone + 50% polyethylene glycol 400 (PEG 400)] starting at day 0 (relative to infection) and daily for next 2 days p.i. (76).

Pan PI3K inhibitor. Mice were intraperitoneally injected with LY294002 (Selleckchem) (50 mg/kg per day) or vehicle control (4% DMSO + 30% PEG 300 + 5% Tween 80) starting at day 0 (relative to infection) and daily for next 2 days p.i. (77).

Statistical analysis

All data are expressed as means ± SD. Statistical analyses were performed using the analysis of variance (ANOVA), Mann-Whitney, and unpaired or paired Student’s t test through GraphPad Prism 7 program. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.


Materials and Methods

Fig. S1. Mouse model of AAN.

Fig. S2. Kidney dysfunction or uremia is required for neutrophil dysfunction.

Fig. S3. Neutrophil function in uremia.

Fig. S4. Effect of uremia on antifungal activity of neutrophils.

Fig. S5. Cell signaling events in neutrophil during uremia.

Fig. S6. In vitro treatment of neutrophils and C. albicans with LiCl.

Fig. S7. PMNΔGSK3β mice.

Fig. S8. Working model.

Table S1. Blood basic chemistry of mouse serum.

Table S2. Demographic information for human participants.

Table S3. Cause of kidney disease for human participants.

Data file S1. Primary data.


Acknowledgments: We thank M. McGeachy and P. Pagano for helpful suggestions and discussions. We thank Unified Flow Core, Department of Immunology and UPMC In-patient Dialysis Unit, University of Pittsburgh for FACS and recruitment of dialysis patients, respectively. Funding: This work is supported in part by NIH grants TL1R001858 to A.J.P.; R01-GM107122 to T.D.N.; Division of Intramural Research (DIR) of the NIAID, NIH to M.S.L.; DE022550 and AI107825 to S.L.G.; and DK104680 and AI145242 to P.S.B. Author contributions: C.V.J., T.D.N., G.M.D., M.S.L., S.L.G., and P.S.B. conceptualized and designed the study. C.V.J., K.R., D.-d.L., B.M.C., R.S.O., S.K., J.V.D., L.L., and A.J.P. performed the experiments. C.V.J., K.R., and P.S.B. analyzed the data. F.H.B. recruited and consented the dialysis patients for this study. C.V.J. and P.S.B. wrote the manuscript. 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 or the Supplementary Materials.

Stay Connected to Science Translational Medicine

Navigate This Article