Ketosis ameliorates renal cyst growth in polycystic kidney disease

Summary

Mild reduction in food intake was recently shown to slow polycystic kidney disease (PKD) progression in mouse models, but whether the effect was due solely to reduced calories or some other aspect of the diet has been unclear. We now show that the benefit is due to induction of ketosis. Time-restricted feeding, without caloric reduction, strongly inhibits mTOR signaling, proliferation, and fibrosis in the affected kidneys in a PKD rat model. A ketogenic diet had a similar effect and led to regression of renal cystic burden. Acute fasting in rat, mouse and feline models of PKD results in rapid reduction of cyst volume, while oral administration of the ketone β-hydroxybutyrate (BHB) in rats strongly inhibits PKD progression. These results suggest that cystic cells in PKD are metabolically inflexible, which could be exploited by dietary interventions or supplementation with BHB, representing a new therapeutic avenue to treat PKD.

Graphical Abstract

eTOC Blurb

Dietary interventions that induce a state of ketosis were found to inhibit disease progression in animal models of polycystic kidney disease (PKD), a commonly inherited disease that leads to kidney failure. The beneficial effects of a ketogenic diet on PKD could be replicated in pre-clinical models by oral treatment with the natural ketone beta-hydroxybutyrate.

Introduction

Autosomal Dominant Polycystic Kidney Disease (ADPKD) is a common genetic disease characterized by slowly progressive cyst growth in both kidneys, which leads to deterioration of renal function over the course of several decades necessitating dialysis or kidney transplantation (Chebib and Torres, 2018; Cornec-Le Gall et al., 2019). Cysts derived from tubule epithelial cells exhibit changes in proliferation (Shillingford et al., 2006, 2010) and metabolism (Chiaravalli et al., 2016; Magistroni and Boletta, 2017; Padovano et al., 2017), and activation of signaling pathways that are normally dormant in healthy adult kidneys such as mTOR and STAT signaling (Weimbs et al., 2013). ADPKD results from mutations in either the PKD1 or PKD2 genes, coding for polycystin-1 (PC1) and polycystin-2, respectively, but how these two proteins link to the many cellular changes observed within cyst-lining cells is still poorly understood (Ong and Harris, 2015).

A key signaling molecule activated in ADPKD is the kinase mTOR that regulates many cellular behaviors including proliferation, cell growth and energy metabolism, and is normally regulated by numerous inputs including growth factor signaling, cellular energy status and nutrient availability (Cornu et al., 2013; Saxton and Sabatini, 2017). mTOR inhibitors, such as rapamycin, have proved highly effective in PKD rodent models (Ravichandran et al., 2014; Shillingford et al., 2006, 2010; Wahl et al., 2006). but, unfortunately, failed in clinical trials (Serra et al., 2010; Walz et al., 2010), most likely because high enough doses cannot be administered over the long-term in humans due to significant extra-renal side effects and toxicities (Weimbs et al., 2018).

The high mTOR activity found in ADPKD may be related to recently observed changes in energy metabolism of renal cysts in animal models, and in cultured cells that lack PC1. These mutant cells appear to exhibit an altered metabolism characterized by increased glycolysis, mitochondrial abnormalities, decreased oxidative phosphorylation, and defective fatty acid oxidation similar to the Warburg effect in cancer (Magistroni and Boletta, 2017; Menezes et al., 2016; Rowe et al., 2013), making PC1-deficient cells apparently more dependent on glucose and gluconeogenic substrates like glutamine (Flowers et al., 2018; Hwang et al., 2015) in order to meet their energetic requirements. Similarly, abnormalities in mitochondria and energy metabolism can be triggered by mutations in the PKD2 gene as well as the PKHD1 gene, a principal gene mutated in autosomal recessive PKD (Chumley et al., 2018). Treatment of PKD cells and Pkd1-null mice with an inhibitor of glycolysis, 2-deoxyglucose (2DG), leads to a reduction in cyst growth (Chiaravalli et al., 2016; Riwanto et al., 2016). Similar effects on cyst growth have been observed utilizing other pharmacological interventions targeting metabolism including the AMPK activator metformin, the PPAR-α activator fenofibrate, the PPAR-γ activator rosiglitazone and the GLS1 inhibitor BPTES (Weimbs et al., 2018).

As an alternative to pharmacological intervention, we considered influencing mTOR activity in the kidney through dietary intervention. Since mTOR is under control of nutrient availability, we recently investigated whether a reduction in food intake may affect renal cyst growth in an orthologous Pkd1 mouse model (Kipp et al., 2016). Surprisingly, a very mild reduction in food intake, by only 23%, profoundly inhibited renal cyst growth and mTOR activity while not affecting body weight gain in these animals. An independent study, using two different orthologous mouse models, had the same conclusion and showed that food restriction by 40%, and even by only 10%, significantly inhibited renal cyst growth (Warner et al., 2016). These striking results raise questions about the mechanism underlying the beneficial effect of dietary restriction. Is the inhibition of renal cyst growth due to overall caloric restriction, or due to the restriction of a particular macro- or micro-nutrient, or could it be caused by a different mechanism entirely?

We report here that dietary restriction strongly inhibits renal cyst growth due to metabolic changes caused by intermittent fasting. We show that time-restricted feeding, in comparison to isocaloric ad libitum feeding, causes intermittent decrease of blood glucose, an increase in ketogenesis, and leads to strong inhibition of renal cyst growth, proliferation and fibrosis. These effects are not merely due to circadian feeding rhythm because ad libitum administration of a ketogenic diet similarly inhibits renal cyst growth. We show that acute fasting in mouse, rat and feline models of PKD induces significant apoptosis in cyst-lining epithelial cells and a striking reversal of renal cystic burden. Finally, we show that supplementation of a normal chow diet with just the natural ketone β-hydroxybutyrate (BHB) replicates the beneficial effects of dietary restriction. Our results indicate that dietary restriction has profound inhibitory effects on PKD progression and this depends on induction of ketosis as renal cyst cells in PKD appear to be metabolically inflexible and thus unable to adapt to alternative fuel sources. These results are consistent with the observation that hyperglycemia and obesity correlate with faster disease progression in individuals with ADPKD. Our results strongly suggest that disease progression in ADPKD could be controlled by dietary interventions such as time-restriction or ketogenic diets, and/or by BHB supplementation. These dietary interventions are safe for long-term use and well-established for the control of conditions such as type 2 diabetes, obesity and childhood epilepsy, so they could be adapted for the treatment of ADPKD.

Results and Discussion

Glucose levels predict the rate of kidney volume change in individuals with ADPKD

Individuals with ADPKD and type 2 diabetes have significantly larger total kidney volume (TKV) than those with ADPKD alone (Reed et al., 2012), and overweight/obesity associates with faster progression in early-stage ADPKD (Nowak et al., 2018). To investigate whether serum glucose levels even in normoglycemic, non-diabetic subjects with ADPKD correlate with disease progression, we explored this relationship in a cohort of individuals with ADPKD with preserved renal function that were recruited to the CRISP study (Yu et al., 2019). We used linear regression models to analyze average yearly change in TKV (average 87.8 mL/year, SD 99.2) and change in height-adjusted TKV (htTKV; average change 50.1 mL/m/year, SD 55.6) between baseline and the last follow up with TKV and glomerular filtration rate (GFR) measurements (on average 9.7 yrs, SD 3.6 yrs) as a function of baseline serum glucose levels (after overnight fasting, average 89.6 mg/dL, SD 10.6), also adjusting for baseline TKV or htTKV and PKD gene mutation (PKD1 vs PKD2 vs no mutation detected). The baseline serum glucose in this cohort uniquely predicted greater TKV (Fig. 1A) and htTKV (Fig. 1B) increase over time (for TKV change: n = 232, b = 0.95, P = 0.041; for htTKV change: b = 0.54, P = 0.042; See Suppl. Table 1). Similar analyses found no effect of the baseline serum glucose on average yearly change in GFR.

Figure 1: Serum glucose predicts TKV increase in individuals with ADPKD, and time restricted feeding ameliorates PKD disease progression in a PKD rat model.

A. and B.) Linear regression analysis of the relationship between serum glucose levels (after overnight fasting) and average annual kidney volume change (TKV (A) or height-adjusted TKV (htTKV, B) in normoglycemic, non-diabetic individuals with ADPKD. reveals positive correlation. Line represents best fit and dotted lines represent 95% confidence intervals. C.) Serum β-hydroxybutyrate (BHB) levels from PKD1^cond/cond:Nes^cre mice (PKD; n =6) and wild-type controls (WT; n =6) from our previous research (Kipp et al., 2016) on reduced food intake (RFI; n =8) vs. ad libitum (AL; n =8) controls. D.) Schematic outline of the feeding regimens used in this study. Time-restricted-fed (TRF) rats were allowed access to food from T14 to T22. Food access is indicated by the orange boxes. E.) Average food intake per animal, per gram bodyweight, per week for rats on AL or TRF diet. F.) Average change in body weight for male and female PKD and WT rats on AL or TRF diet. G. and H.) H&E stained kidney sections. Scale bars are 3 mm or 100μm, respectively. I.) Cystic index (percent cystic area) for rats on TRF or AL diet. J.) 2-kidney weight to body weight ratios of TRF and AL treated rats. K.) Serum creatinine levels of AL and TRF animals. L.) Total number of cysts per kidney section in AL and TRF animals. M.) Cyst sizes of cysts counted in L. N.) BHB levels at the start of the experiment (week 3) and the final time point of the experiment (week 8). O.) Blood glucose levels at the start of the experiment (week 3) and the final time point of the experiment (week 8). (See also Fig. S2) Error bars represent SD. Statistical significance was determined using Mann-Whitney analysis. n =8 male and 12 female Cy/+ rats; n =13 male and 8 female wild-type rats for TRF experiments.

Together these results suggest that glucose availability may directly affect renal cystic disease progression in ADPKD. This hypothesis is also supported by the previous findings that streptozotocin-induced hyperglycemia accelerates cystogenesis in a cilia-defective mouse model of PKD (Sas et al., 2015), and that glucose in the growth media promotes cyst growth in in vitro models (Kraus et al., 2016). Conversely, this hypothesis suggests that limiting glucose availability may ameliorate cystic progression in PKD.

Time restricted feeding ameliorates disease progression in the Han:SPRD rat model of PKD

We previously reported that a mild reduction in food intake (RFI) by 23% compared to ad libitum fed controls (AL) strongly inhibits PKD progression in a mouse model of the disease (Kipp et al., 2016). In our study we noticed a behavioral difference between animals in the RFI and AL cohorts. Whereas AL-fed controls had continuous access to food that they consumed frequently, mice in the RFI cohort consumed their entire daily pre-weighed food allotment shortly after feeding, usually within one hour. This behavioral difference makes it difficult to separate effects of caloric restriction and those of time-restricted feeding in which animals are restricted from food for a set time period each day. We measured blood β-hydroxybutyrate (BHB) levels from the animals in our previous RFI study and found significantly elevated levels in the RFI cohort compared to the AL cohort (Fig. 1C) indicating that mice in the RFI group had indeed experienced intermittent fasting leading to ketosis.

To test whether the beneficial effect on renal cyst growth may be due to the effects of intermittent fasting, rather than reduced food intake per se, we placed Han:SPRD rats, a non-orthologous model of PKD with a mutation in Anks6 (Brown et al., 2005), on a time-restricted feeding (TRF) regime wherein TRF animals had access to food for an 8-hour period within their 12-hour dark cycle (Fig. 1D). Animals were treated for 5 weeks from postnatal weeks 3–8. This experimental time point was chosen because Han rat heterozygotes (Cy/+) exhibit rapid cystogenesis beginning at 3 weeks of age until around 8 weeks when cyst progression begins to slow. Male animals are more strongly affected than females in this model and exhibit very mild renal function impairment at 8 weeks (Cowley et al., 1993). Therefore, all outcomes were sex-stratified. TRF animals consumed a comparable number of calories (Fig. 1E) and gained weight similar to the AL fed controls (Fig. 1F) indicating that TRF does not lead to food or caloric restriction. Similarly, there were no negative effects measured on any other organ (Fig. S1. Related to Figure 1). After 5 weeks of this dietary regimen, animals in the TRF cohort exhibited strikingly reduced renal cystic disease progression compared to animals in the AL cohort. Kidneys of TRF-treated animals were significantly less cystic (Fig. 1G–I) with a marked reduction in the 2-kidney/body weight ratio (Fig. 1J), and improved kidney function as indicated by normalization of serum creatinine (Fig. 1K). We next analyzed whole kidney sections for total cyst number (Fig. 1L) and cyst size (Fig. 1M), each of which were decreased in TRF animals compared to controls, indicating that TRF inhibited both cystogenesis and cyst expansion. Additionally, TRF-treated animals had increased levels of BHB (Fig. 1N) indicating that the TRF regimen induced a state of ketosis. This was accompanied by reduced blood glucose (Fig. 1O) which was consistently decreased at the end of the fasting period over the experimental timecourse (Fig. S2. Related to Figure 1).

Time restricted feeding reduces mTORC1 and STAT3 signaling, interstitial fibrosis, and proliferation in cystic kidneys

We assessed changes in known PKD associated pathways and pathologies. The kinase mTOR is important in nutrient sensation and its activity is known to be elevated in PKD and is a driver of disease progression (Shillingford et al., 2010). We found a significant reduction in phosphorylated S6^S235/236 (Fig. 2A,D), a downstream target in the mTOR pathway, in cyst-lining epithelial cells of TRF-treated male, but not female, Cy/+ animals indicating inhibition of mTORC1 activity. PKD also leads to interstitial fibrosis, a hallmark of progressive renal disease. We found that collagen deposition was markedly reduced in TRF compared to AL animals (Fig. 2B). To determine whether the reduction of fibrosis may be due to a change in the myofibroblast population we probed for the myofibroblast marker smooth muscle actin-1 (SMA). Whereas cysts in AL animals are associated with numerous SMA-positive myofibroblasts consistent with previous results (Norman, 2011; Song et al., 2017), myofibroblasts were strongly reduced, and frequently absent, in kidneys of TRF animals (Fig. 2C, E, F). Staining with the cell cycle marker Ki-67 revealed a significant reduction in cyst-lining cells in TRF compared to AL animals indicating inhibition of proliferation (Fig. 2G, H). We investigated STAT3 which has previously been shown to be activated in PKD and drives cystic progression (Leonhard et al., 2011; Takakura et al., 2011; Talbot et al., 2011, 2014; Torres et al., 2019; Weimbs et al., 2013). Active, phosphorylated-STAT3 was reduced or absent in cysts in TRF-treated animals (Fig. 2I).

Figure 2: Time restricted feeding reduces mTORC1 signaling, fibrosis, and proliferation in cystic kidneys.

A.) Immunostaining of the mTORC1 marker pS6^S235/236(cyan), in kidneys from WT and PKD rats on AL diet (top and middle) or PKD rats on TRF diet (bottom). B.) Sirius Red stained kidney sections from AL and TRF rats showing collagen deposits (red). C.) Immunostaining of the myofibroblast marker SMA (cyan), in kidneys from WT and PKD rats on AL diet (top and middle) or PKD rats on TRF diet (bottom) D.) Quantification of pS6-positive cells in cyst lining epithelia as a percent of the total number of DAPI stained nuclei. E.) Higher magnification images SMA-stained kidney sections showing pericystic myofibroblasts (cyan). F.) Quantification of SMA-positive cells as the percent of total cells. G.) Ki-67 immunostaining in kidneys from WT and PKD rats on AL diet (top and middle) or PKD rats on TRF diet (bottom). H.) Quantification of Ki-67 positive cells separated by location from the cyst lining, tubular, or interstitial regions and expressed as a percent of the total number of DAPI stained nuclei. I.) Immunostaining of P-STAT3 (cyan), in kidneys from WT and PKD rats on AL diet (top and middle) or PKD rats on TRF diet (bottom). Scale bars=100μm. Error bars represent SD. Statistical significance determined using Mann-Whitney analysis. n =8 male and 12 female Cy/+ rats; n =13 male and 8 female wild-type rats for TRF experiments.

A ketogenic diet ameliorates disease progression in juvenile Han:SPRD rats

To confirm that the observed beneficial effect of TRF on PKD kidneys is due to ketosis as opposed to other possible consequences of time-restricted feeding, such as alterations in circadian rhythms, we explored the use of a ketogenic diet. Han:SPRD rats were given ad libitum access to a high fat, very low carbohydrate ketogenic diet (KD; caloric ratio of ~91% fat, ~2% carbohydrates and ~5% protein) and compared to animals with ad libitum access to normal chow (NC; caloric ratio of ~62% carbohydrates, ~25% protein and ~13% fat). Animals were treated again for 5 weeks from postnatal weeks 3–8. Renal cystic disease progression was strongly inhibited in animals on the KD regimen compared to those on NC (Fig. 3A,B) along with a remarkable reduction in the 2-kidney to body weight (Fig. 3C) and cystic area (Fig. 3D). KD-treated rats exhibited decreased blood glucose (Fig. 3G) and increased BHB levels (Fig. 3H) compared to NC controls, and the level of ketosis was more profound than that achieved with the TRF regimen (Fig. 1N). Animals on the KD appeared phenotypically normal and did not exhibit any signs of distress during the study. However, KD-treated animals exhibited reduced bodyweight gain (Fig. 3E), and only gained approximately 2–10% of their body weight each week during the study (Fig. 3I). This lack of growth was not dependent on caloric intake as the KD group consumed comparable calories to NC controls relative to their body weight (Fig. 3J). Similar stunting induced by the commonly used ketogenic rodent chow has previously been described in juvenile rats (Liśkiewicz et al., 2017), which would normally be rapidly growing during the treatment period, and we attributed the stunting effect as most likely resulting from protein restriction. However, comparing effects of the KD on multiple organs (Fig. S3. Related to Figure 3) clearly showed that it has a disproportionate inhibitory effect on the growth of polycystic kidneys as compared to normal kidneys, for example, as illustrated after normalizing to heart weights (Fig. 3F). Importantly, KD-feeding led to improved kidney function (Fig. 3K) as well as inhibiting both cystogenesis (Fig. 3L) and cyst expansion (Fig. 3M).

Figure 3: A ketogenic diet ameliorates disease progression in juvenile Han:SPRD rats.

A. and B.) H&E stained kidney sections from WT or PKD rats fed ad libitum a Normal Chow (NC) or Ketogenic diet (KD). Scale bars=3mm or 100μm, respectively. C.) 2-kidney weight to body weight ratios of NC and KD treated rats. D.) Cystic index (percent cystic area) from NC and KD cystic rats. E.) Body weight of wild-type and cystic animals on NC and KD. F.) 2-kidney weight to heart weight ratios of NC and KD treated rats. G.) Blood glucose at initial (week 3) and final time point (week 8) of the experiment. H.) Blood β-hydroxybutyrate levels at initial (week 3) and final time point (week 8). I.) Percent change in mass of. J.) Calories consumed per gram of bodyweight per week. K.) Serum creatinine. L.) Total number of cysts per kidney section. M.) Cyst sizes of cysts counted in L. Error bars represent SD. Statistical significance determined from Mann-Whitney analysis. n =8 male and 7 female Cy+ rats; n =13 male and 6 female wild-type rats for ketogenic diet experiments (* P<0.05, **** P<0.0001)

Effects of KD-treatment on PKD progression were generally similar to those caused by TRF-treatment. We found that mTOR (Fig. 4A, E) and STAT3 (Fig. 4B) activity in cystic epithelia were both blunted in KD-fed rats. SMA1-positive myofibroblasts were less in KD-fed animals compared to NC controls (Fig. 4C, F) and this was accompanied by a decrease in collagen deposition (Fig. 4D, G). A decrease in the percent of Ki-67 positive cells again suggests inhibition of cell cycle entry and proliferation (Fig. 5H, I).

Figure 4: The ketogenic diet reduces markers of PKD in juvenile Han:SPRD rats.

A.) Immunostaining of the mTORC1 marker pS6^S235/236(cyan), in kidneys from WT and PKD rats on Normal Chow diet (NC, top and middle) or PKD rats on Ketogenic Diet (KD) (bottom). B.) Immunostaining of P-STAT3 (cyan). C.) Immunostaining of the myofibroblast marker SMA (cyan),. Scale bars=100μm D.) Sirius Red/Fast Green stained kidney sections showing collagen deposits (red). Scale bar=300μm E.) Quantification of pS6-positive cells in cyst lining epithelia as a percent of the total number of DAPI stained nuclei in cysts. F.) Quantification of SMA positive cells as a percent of the total number of cells. G.) Quantification of collagen positive area. H.) Ki-67 immunostaining. Scale bar=100μm I.) Quantification of Ki-67 positive cells separated by location from the cyst lining, tubular, or interstitial regions and expressed as a percent of the total number of DAPI stained nuclei. J.) Immunoblot for the indicated markers. K.) Quantification of immunoblots. Error bars represent SD. Statistical significance determined using Mann-Whitney analysis. n =8 male and 7 female Cy+ rats; n =13 male and 6 female wild-type rats for ketogenic diet experiments (* P <0.05, **P <0.01).

Figure 5: The ketogenic diet reverses kidney disease progression in adult Han:SPRD rats.

A.) Blood glucose levels at initial (week 8) and final time point (week 12) of the experiment. B.) Blood β-hydroxybutyrate levels. C. and D.) Percentage change in NC and KD treated animal’s body, kidney and heart mass over the 4-week treatment period. Bracketed masses reflect the difference between 8 week and 12 week values or differences between NC and KD rats. E. and F.) H&E stained kidney sections. Scale bars=3mm or 100μm, respectively. G.) 2-kidney to body weight ratios. H.) Cystic index (percent cystic area). I.) Serum creatinine. J.) Total number of cysts per kidney section. K.) Cyst sizes of cysts counted in L.) Sirius Red/Fast Green stained kidney sections showing collagen (red). M.) SMA immunostaining (cyan). N.) Immunostaining of the mTORC1 marker pS6^S235/236 (cyan). O.) Ki-67 immunostaining. P.) Quantification of collagen positive area from animals in L and compared to 8 week rats. Q.) Quantification of SMA positive cells as a percent of the total number of cells. R.) Quantification of Ki-67 positive cells separated by location into cyst lining, tubular, or interstitial and expressed as a percent of the total number of DAPI stained nuclei. Scale bars=100μm. Error bars represent SD. Statistical significance was determined using Mann-Whitney analysis. n =12 male and 10 female cystic rats; n =10 male and 14 female wild-type rats for adult ketogenic diet experiments. (* P <0.05, **** P <0.0001).

A KD is known to raise AMPK activity (Bae et al., 2016) and to promote fatty acid uptake by increasing expression of CPT1α via induction of PPARα (Grabacka et al., 2016). We found that KD feeding increased the levels of phospho-AMPK in the kidneys of cystic animals and increased CPT1α levels in male cystic, but not female cystic, rats (Fig. 4J, K).

A ketogenic diet ameliorates disease progression in adult Han:SPRD rats

To exclude possible confounding factors related to the suppression of body weight gain, we next fed Han:SPRD rats at an older age (postnatal week 8–12) with a KD vs NC. At this age, there is no longer cystic kidney enlargement (Cowley et al., 1993) and we observed no further change in kidney size (compare Fig. S3, S4. Related to Figure 5). The 4-week KD regimen resulted in effective ketosis as illustrated by decreased blood glucose (Fig. 5A) and increased blood BHB (Fig. 5B).

During the treatment period, these adult rats only exhibit a modest body weight gain on NC (Fig. 5C, D). While weight gain was mildly affected in rats on the KD, there was no difference between wild-type or cystic animals (Fig. 5C, D). In contrast, the total mass of polycystic kidneys remained nearly the same in animals on NC but lowered by 35% in male rats on KD (Fig. 5C) which was accompanied by diminished 2-kidney to bodyweight ratio (Fig. 5G) and percent cystic area (Fig. 5H). Similar to what was found in our juvenile experiments (Fig. S3), the KD again affected the organ size of polycystic kidneys disproportionally because wild-type kidneys or other organs, such as the heart, were largely unaffected (Fig. 5C, D and Fig. S4). Unlike our juvenile experiments, serum creatinine was unaffected by KD treatment in adult rats (Fig. 5I) as well as the total number of cysts per animal (Fig. 5J). There was however a marked reduction in cyst size (Fig. 5K) indicating that KD feeding prevents cyst growth in animals at that age. It should be noted that the number of cysts in untreated 12-week animals are similar to those in untreated 8-week animals (Fig. 1L, 3M) indicating that there are few or no new cysts formed after 8 weeks of age. KD feeding led to a reduction in fibrosis in male animals as indicated by reduced collagen (Figures 5L and 5P) but there was no detectable difference in myofibroblasts (Fig. 5M, Q), mTOR activity (Fig. 5N) or the cell cycle marker Ki-67 (Fig. 5O, R).

Acute fasting leads to cell death and renal cyst regression in multiple PKD animal models

The strong inhibitory effects on PKD progression we observed with ketosis induced by TRF and KD regimens occurred during the course of several weeks. To investigate whether ketosis has any fast-acting effects on renal cysts, we subjected Han:SPRD rats to acute ketosis via fasting. 8-week old animals were fasted for 48 hours with free access to water. As expected, acute fasting led to decreased blood glucose (Fig. 6A) and increased blood BHB values (Fig. 6B) indicating the induction of ketosis. The fasting-induced modest decrease in body weight (~12%) is expected due to depletion of fat and glycogen reserves, primarily in adipose tissue, liver and skeletal muscle, respectively, and was consistent between wild-type and cystic animals (Fig. 6C). Consequently, liver weights lowered among all groups (Fig. S5. Related to Figure 6). Strikingly, acute fasting strongly affected cystic kidneys leading to ~20% reduction in cystic area (Fig. 6D) and a concurrent reduction in kidney mass (Fig. 6E) and the 2-kidney/heart weight ratio (Fig. 6F). Fasting had minimal to no effect on the weight of hearts (Fig. 6G) and normal kidneys (Fig. 6E) as expected, indicating that the effect of fasting is specific to PKD kidneys. The loss of mass was not due to any loss of glycogen because neither normal kidneys nor polycystic kidneys contain a significant amount of glycogen (Gatica et al., 2015). Instead, the extent of the reduction of cystic volume almost precisely accounted for the observed loss in kidney mass, suggesting that acute fasting leads to a loss of cyst fluid.

Figure 6: Acute fasting leads to rapid cystic cell death, and reduced cyst and kidney size in polycystic kidneys.

A.) Blood glucose and B.) β-hydroxybutyrate levels of untreated and 48-hour fasted Han:SPRD and WT rats. C.) Animal mass. D.) Cystic index (percent cystic area). E.) Single kidney masses. F.) 2-kidney to heart mass ratios. G.) Heart masses. H.) TUNEL stain. Kidneys from WT rats did not demonstrate significant levels of apoptosis. Scale bar=100μm I.) Percentage of TUNEL positive cells as the percent of the total number of DAPI stained nuclei in cystic epithelia, tubules, interstitial or cells within lumen. J.) Sirius red/fast green stain of ad libitum and fasted Han:SPRD rats indicating collagen and basement membranes in red. Arrowheads denote fragmented epithelial cells. Arrows denote denuded epithelium. Asterisks denote red blood cells in cysts. K.) Oil O red stain. Scale Bar=50μm. L.) Blood glucose values and M.) BHB levels from 24-hour fasted Pkd1^cond/cond:Nes^cre mice. N.) Single kidney mass of AL and fasted Pkd1^cond/cond:Nes^cre mice. O.) 2-kidney to bodyweight ratio of AL and fasted Pkd1^cond/cond:Nes^cre mice. P.) TUNEL stain of kidney sections from Pkd1^cond/cond:Nes^cre mice after a 24-hour fast. Scale bars=100μm. Q.) Percentage of TUNEL positive epithelial cells or cells within cyst lumens as a percent of the total number of DAPI stained nuclei in AL vs. fasted Pkd1^cond/cond:Nes^cre mice in cystic. R.) CT scans of PKD cats before and after 72 hours fast. * Denotes the same cyst in each image. S.) Quantification of total kidney volumes of PKD cats determined ~1.5 years prior to the study, immediately before fasting or immediately after fasting. Error bars represent SD. Statistical significance was determined using Mann-Whitney analysis. n =8 male and 5 female cystic rats; n =7 male and 5 female wild-type rats, n =4 polycystic cats, and n =4 female cystic Pkd1^cond/cond:Nes^cre mice; n =3 male and 5 female wild-type mice were used for fasting experiments. (* P <0.05, ** P <0.01, **** P <0.0001)

TUNEL staining revealed that acute fasting led to greater cell death of cystic cells (Fig. 6H, I) but did not cause greater cell death in normal, non-cystic kidneys (Fig. 6H). In particular, there was a conspicuous number of TUNEL-positive, detached cells in cyst lumens indicating that these dead cells originated from sloughed-off cyst-lining cells (Fig. 6H, I). Renal cysts in fasted animals frequently exhibited fragmented cells and denuded epithelium which was not observed in renal cysts of control-fed cystic animals (Fig. 6J). These results suggest that the loss of cyst fluid caused by acute fasting is due to cell death and disruption of the epithelial barrier of cysts leading to draining of cyst fluid presumably via the insterstitium and lymphatics, leading to overall shrinking of polycystic kidneys.

To explore a possible reason for the induction of cell death of cystic cells by acute fasting, we considered the possibility that the increased supply of fatty acids during ketosis may affect cystic cells. As expected, intracellular oil droplets are rarely observed in normal or cystic kidneys in control-fed animals (Fig. 6K). In contrast, acute fasting leads to accumulation of oil droplets in tubule cells in normal kidneys consistent with previous reports (Krzystanek et al., 2010; Scerbo et al., 2017). Cyst-lining cells in polycystic kidneys, however, exhibited a much-exaggerated degree of cytoplasmic oil droplet accumulation (Fig. 6K). Frequently, most of the cytoplasm of the affected cells appeared to be occupied by oil droplets suggesting a state of severe steatosis. These findings suggest that cyst-lining cells have the ability to take up circulating fatty acids during acute fasting but are unable to sufficiently metabolize them which may lead to lipotoxicity and greater observed cell death.

To test whether acute ketosis due to fasting has similar effects in an orthologous mouse model of PKD, we used the previously described orthologous Pkd1:Nestin-Cre model (Kipp et al., 2016, 2018; Shillingford et al., 2010). Due to the smaller body size and faster metabolism of mice, we were only able to fast mice for 24 hours as opposed to the 48 hours in the rat model above. Fasting for 24 hours, with free access to water, led to pronounced decrease in blood glucose (Fig. 6L) and increase in blood BHB (Fig. 6M) but no significant change in kidney mass (Fig. 6N) or 2-kidney to bodyweight (Fig. 6O). However, TUNEL staining revealed that acute fasting again induced apoptotic cell death in cyst-lining cells and greater TUNEL positive luminal cells, in polycystic kidneys but not in normal kidneys (Fig. 6P, Q). We speculate that the short duration of fasting in mice is insufficient to lead to significant cyst draining and a measurable effect on cystic burden, as compared to rats.

To investigate whether these dramatic effects of acute fasting may extend to a non-rodent model of ADPKD, we employed a feline model in a longitudinal study. A naturally occurring mutation in the orthologous PKD1 gene leads to a high prevalence of ADPKD in cats of Persian cat descent (Lyons et al., 2004). This feline model of ADPKD is arguably the model that most closely approximates human ADPKD because the underlying genetics appear to be identical. In contrast to most genetically engineered animal models, only one allele of the PKD1 gene is affected in the germline in cats which presumably leads to a similar second-hit mechanism as in human ADPKD. Due to heterogeneity in the strain background, the rate of disease progression in feline ADPKD is naturally as variable as in human ADPKD which necessitates a longitudinal study design. The trial included four adult cats with confirmed polycystic kidneys and PKD1 DNA mutation, two males and two females, one of each sex was desexed. The four cats had varying levels of renal cystic disease as shown by computed tomography (Fig. S6, CT of PKD cats. Related to Figure 6R. Suppl. Table 3. Related to Figure 6S) but no evidence of renal functional impairment. Animals were subjected to fasting for 72 hours with free access to water. Total kidney volumes (TKV) were determined from CT scans ~1.5 years prior to the study, immediately prior to and immediately after fasting. Cats lost between 0–0.3 kg of body weight as would be expected due to loss of glycogen and fat deposits. Remarkably, the TKV values of all four cats decreased during the 72 hour fasting periods by an average of 15% (range 3.5–29.5%, Fig. 6S). In contrast, the TKV values had increased or remained stable during the ~1.5 years prior to the study with an average increase of 1.2 ml/months (range −0.1 to + 2.8 ml/mo, Fig. 6S). Fasting led to substantially reduced size of some of the identifiable cysts (Fig. 6R asterisk) suggesting that fluid draining occurred similar to what we observed in the rat model.

Oral ß-hydroxybutyrate treatment prevents polycystic disease progression in juvenile rats

We hypothesized that BHB itself may contribute to the observed inhibitory effect of ketosis on PKD progression. Recent results indicate that BHB is not only an energy source for cells but has complex effects on cellular signaling and metabolism that are only partially resolved to date (Newman and Verdin, 2017; Rojas-Morales et al., 2016). Importantly, BHB affects numerous signaling pathways that are implicated in PKD including mTOR, AMPK and HDACs (Bartmann et al., 2018; Shimazu et al., 2013). To investigate whether BHB may have a dominant beneficial effect on PKD progression we treated Han:SPRD rats with BHB from postnatal weeks 3–8 in the context of ad libitum feeding with normal, high-carbohydrate chow (NC). BHB was administered as a sodium/potassium salt in the drinking water ad libitum at a dose equivalent to that recommended as a dietary supplement in BHB supplement drinks. Control cohorts received either normal water or water supplemented with the molar equivalents of salt provided in the BHB supplement (a mixture of NaCl/KCl). All treated animals consumed comparable amounts of water and calories (Fig. S7. Related to Figure 7.), and BHB treatment had no effect on body weight (Fig. 7D) or blood glucose levels (Fig. S7).

Figure 7: BHB administration ameliorates disease progression in the Han:SPRD rat model of PKD.

A.) Gross kidney images of untreated cystic (left), BHB-treated cystic (middle) and BHB-treated wild-type (right) rats. Scale bar=1cm B.) Sirius Red/Fast Green stained kidney sections. Scale bar=2mm and 50μm respectively. C.) 2-kidney to bodyweight ratios, D.) animal masses, and E.) cystic areas of water, salt and BHB treated rats. F.) Serum creatinine. G.) Percent collagen area from Sirius Red/Fast Green stained sections. H.) Ki-67 immunofluorescence stain. I.) Ki-67 quantification. J.) SMA immunofluorescence stain for myofibroblasts (cyan). Scale bar =50μm. Error bars represent SD. Statistical significance was determined using Mann-Whitney analysis. n =7 male and 5 female cystic rats; n =3 male and n =5 female wild-type rats treated with BHB and n =4 male and 4 female cystic rats; n =2 male and 5 female wild-type rats treated with Na/K salt. (* P <0.05, ** P <0.01, *** P <0.001).

After 5 weeks of treatment with BHB the kidneys of PKD animals were nearly indistinguishable from kidneys of wild-type animals, both at the gross and histological level (Fig. 7A, B). BHB-treated PKD rats showed striking reductions in the 2-kidney to bodyweight ratio (Fig. 7C) and the cystic area (Fig. 7E) compared to water- and salt-treated controls. Neither wild-type kidneys nor other organs were affected by BHB treatment indicating that the effect is specific to polycystic kidneys (Fig. S7). BHB treatment led to a strong reduction of fibrosis (Fig. 7B, G), almost complete elimination of myofibroblasts (Fig. 7J), improved kidney function (Fig. 7F) and inhibition of proliferation (Fig. 7H, I).

These results demonstrate that BHB acts in a dominant fashion as a suppressor of PKD progression, even in animals that are fed a high-carbohydrate diet and have unaltered blood glucose levels. Altogether, our results suggest that inducing a state of ketosis, either by dietary intervention or by mimicking its effects with BHB, may be an effective treatment for ADPKD in humans.

Conclusions and Clinical Implications

Previous studies have suggested potential beneficial effects of lowering protein intake or changing the dietary protein composition in individuals with ADPKD (Klahr et al., 1995) and PKD rodent models (Devassy et al., 2017; Fair et al., 2004; Ogborn and Sareen, 1995; Tomobe et al., 1994). In most of these studies the level of protein alteration was relatively major but the beneficial effect was relatively minor. Protein alterations cannot account for the effects observed in our study because neither the amount nor composition of dietary protein was altered in our TRF regimen or in BHB treatment experiments. A potential mechanism that has been proposed to explain effects of TRF regimens is alteration in circadian rhythms. Again, such a mechanism cannot account for the effects observed here because the KD and BHB were administered ad libitum. Furthermore, TRF regimens were recently shown to be still effective in animals lacking a functional circadian clock (Chaix et al., 2018).

The sum of our results suggests that the metabolic changes induced during ketosis are responsible for suppressing PKD progression, specifically by inhibiting cyst expansion and through early intervention, cystogenesis. The main metabolic changes during ketosis are reflected in a shift of fuel source utilization (1) decreased glucose availability, (2) increased fatty acid supply and (3) increased ketone body supply. It is possible that a combination of all three factors contributes to the beneficial effect. PKD cells have been shown to have an altered cellular metabolism in which the preferred fuel source is glucose that is primarily metabolized by glycolysis resembling a Warburg-like effect (Magistroni and Boletta, 2017). This phenotype might be caused by, or amplified by, recently observed mitochondrial defects that may impair the efficiency of the mitochondrial TCA cycle and fatty acid beta oxidation (Ishimoto et al., 2017). This could lead to overall metabolic inflexibility that may lock PKD cells in a dependency on glucose and glycolysis. This view is consistent with the recently reported efficacy of the glycolysis inhibitor 2-DG in PKD mouse models (Magistroni and Boletta, 2017). Similarly, in our experiments the TRF and KD regimens led to decreased blood glucose levels which may have contributed to the observed inhibition of PKD progression. However, the effect of TRF on blood glucose was relatively minor. Furthermore, blood glucose was unaffected in our BHB treatment experiment. These results suggest that limiting the supply of glucose may be a contributing factor to the efficacy of ketosis but is not a requirement.

A common factor in all of the interventions carried out here is the elevation of BHB. TRF, KD and acute fasting all raise BHB levels to varying degrees. Remarkably, treatment with BHB alone, even in the context of a normal high-carbohydrate rodent diet proved to be highly effective. This suggests that BHB itself is a major factor in suppressing PKD progression. There are two plausible mechanisms to explain this effect: BHB may affect PKD cells either as a metabolite or as a signaling molecule. As a metabolite, BHB is known for its favorable use in the TCA cycle by replacing succinyl-CoA and increasing the free pool of succinate, improving mitochondrial efficiency and TCA cycle turnover (Henderson, 2008; Veech et al., 2001). This may alter the overall metabolism of PKD cells and affect their ability to proliferate. On the other hand, BHB has numerous effects on signaling pathways, some of which have been implicated in PKD including mTOR, AMPK and HDACs (Bartmann et al., 2018; Newman and Verdin, 2017; Rojas-Morales et al., 2016; Shimazu et al., 2013). Therefore, BHB may affect PKD cells independent of its own metabolism. BHB has been identified as the endogenous ligand of the G-protein coupled receptor GPR109A, which is also activated by unphysiologically high concentrations of the vitamin niacin (Ristic et al., 2017). Interestingly, administration of megadoses of niacinamide has previously been shown to ameliorate PKD in a Pkd1 mouse model although the effect was believed to be due to direct inhibition of SIRT1 activity (Zhou et al., 2013). We speculate that both mega-concentrations of niacin and physiologically relevant concentrations of BHB, as they occur during ketosis, may inhibit PKD progression due to activation of GPR109A. GPR109A signals via inhibition of cAMP (Ristic et al., 2017), and aberrant cAMP signaling is a known driver of PKD progression (Chebib and Torres, 2018; Cornec-Le Gall et al., 2019). Future work is needed to elucidate the mechanism of BHB action in PKD.

Another possible mechanism by which ketosis may contribute to the inhibition of PKD progression involves the increased supply of fatty acids released by the adipose tissue. We show that acute fasting leads to an extreme level of lipid loading in cystlining epithelial cells. This suggests that these cells are fully capable of taking up circulating fatty acids but appear to be unable to catabolize them which necessitates their deposition into lipid droplets as a stress response similar to what has been observed in cancer cells (Petan et al., 2018). During acute fasting, which leads to a rapid onset of ketosis, the ability of PKD cells to sequester fatty acids in lipid droplets may be overwhelmed leading to lipotoxicity and cell death. This mechanism is consistent with the greater cell death, cyst rupture and draining of cyst fluid that we observed as a rapid response to acute fasting. This view is also consistent with the previously described defect of fatty acid oxidation in PKD (Lakhia et al., 2018; Menezes et al., 2016) that can lead to increased lipid loading in PKD cells in vitro (Lin et al., 2018).

Our results show that ketosis not only strongly inhibits renal cyst growth but also fibrosis. During the progression of PKD, cyst proliferation and fibrosis typically go hand in hand, and it is likely that both processes reinforce each other. Myofibroblasts have been shown to be abundant in PKD kidneys and are thought to be a major contributor to fibrosis (Norman, 2011). We find that ketosis greatly reduces the abundance of pericystic myofibroblasts, especially the TRF regimen and BHB treatment, to almost complete absence. Consequently, overall fibrosis, as measured by collagen deposition, is strongly reduced. Future studies are required to determine whether the primary effect of ketosis is on cyst-lining epithelial cells or on myofibroblasts or on both simultaneously.

The KD regimen that we employed to force animals into sustained ketosis utilized a standard high fat, very low carbohydrate diet. This regimen was highly effective in preventing PKD progression which may be surprising in light of previous reports that increases in dietary fat intake worsened PKD progression in the Han:SPRD rat model (Jayapalan et al., 2000) and in Pkd1 mouse models (Menezes et al., 2016). However, in those studies, the fat contents were altered in the context of a high carbohydrate food background which would prevent animals from experiencing ketosis.

We have investigated different interventions to induce ketosis – time-restricted feeding, ketogenic diet, acute fasting, direct oral BHB administration – using multiple PKD models: the non-orthologous Han:SPRD rat model, the orthologous Pkd1 mouse model, and the orthologous feline ADPKD model. These multiple approaches all led to consistent results which suggests that the observed effects are predictive for human ADPKD.

The main significance of our findings lies in the almost immediate clinical applicability and the suggestion that effective ADPKD therapy may be possible by dietary intervention and without the need for pharmacological intervention. Although head-to-head comparisons would have to be made, in our experience the magnitude of the observed beneficial effect of ketosis on PKD progression appears to be greater than the effect of any pharmacological compound that has been tested in rodent models of PKD to date. This includes effects of vasopressin V2 receptor antagonists (Gattone et al., 2003) that eventually led to the first – and only - approved drug therapy for ADPKD. This drug, tolvaptan, was shown to be modestly effective in clinical trials (Torres et al., 2012) but its use is complicated due to side effects including polyuria and potential liver toxicity (Chebib and Torres, 2018), and its high cost has called the cost-effectiveness into question (Erickson et al., 2013). In contrast, dietary interventions to induce ketosis, such as fasting, are not costly and appear relatively safe. Fasting is a long-held practice for both health and religious purposes in many cultures dating back to prehistoric times and may perhaps be considered the oldest form of medical therapy. It has recently been reported that fasting during the month of Ramadan, a form of time-restricted diet, has no negative effects on renal function in individuals with ADPKD (Ekinci et al., 2018). The ketogenic diet has been clinically used as an effective and safe therapy for pediatric epilepsy for nearly 100 years (Huffman and Kossoff, 2006). While this diet is also widely used to control obesity and diabetes, even among lay people, it is important to note that it is not necessarily free of side effects and should be supervised by a physician. Ketogenic diets are contraindicated in the presence of certain rare genetic metabolic disorders, and can also lead to an increased risk of kidney stones (Kossoff et al., 2018). The latter may be a particular concern in the case of ADPKD.

Unfortunately, dietary interventions most frequently fail as therapies due to inability of individuals to adhere to the diets. Our finding that oral treatment with BHB alone, on top of the normal ad libitum-fed high-carbohydrate diet, was highly effective in halting PKD progression suggests that a similar treatment would be feasible in individuals with ADPKD. BHB is readily available in the US as a dietary supplement. We administered BHB with the drinking water mimicking the dose that manufacturers of sports/energy supplements recommend. Such supplementation would avoid the need for lifestyle changes required for time-restricted or ketogenic diets, greatly increasing feasibility.

Most individuals with ADPKD in industrialized societies consume a high carbohydrate diet throughout their waking hours and only rarely, if ever, experience periods of ketosis. A recent nutritional feasibility study recommends high carbohydrate diets for individuals with ADPKD (Taylor et al., 2017) as do many available cookbooks, websites, online forums etc. where individuals with ADPKD may seek dietary advice. Our results suggest that such dietary habits may worsen PKD progression. This concept is also supported by a recent report of faster disease progression in overweight/obese early-stage ADPKD (Nowak et al., 2018). Moderation in caloric intake has sensibly been recommended for the management of ADPKD (Chebib and Torres, 2018). Our results presented here suggest, however, that reducing caloric intake per se would not necessarily lead to inhibition of cystic progression but that the induction of the state of ketosis is instead important, irrespective of caloric intake. Rigorous validation of dietary recommendations in randomized clinical trials is urgently needed and may lead to scientifically validated dietary interventions with the potential to become the standard of care for ADPKD.

Limitations of Study

Several of the experiments in this study make use of the non-orthologous Han:SPRD rat model of PKD. The mutation in this model affects a gene, samcystin, that is different from the PKD genes affected in human ADPKD, and it is unclear if and how samcystin may interact with the polycystins, the products of the PKD genes. Therefore, there is the caveat that results obtained in this model may not necessarily be predictive for the human disease. This caveat is balanced by the fact that the results obtained with this non-orthologous model are in agreement with results from other experiments in this study in which the orthologous Pkd1 mouse model and the orthologous Pkd1 feline model were used. Furthermore, the beneficial effect of dietary restriction – which we now suggest was due to ketosis - was originally observed by two independent laboratories – including ours - using altogether three different orthlogous Pkd1 or Pkd2 mouse models (Kipp et al., 2016; Warner et al., 2016).

STAR Methods

Contact for Reagent and Resource Sharing

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Thomas Weimbs (weimbs@ucsb.edu).

Experimental Model and Subject Details

Animals

All animal studies were conducted in accord with the National Research Council Guide with approval of the Institutional Animal Care and Use Committees at the respective institutions. Whenever possible and applicable, historical controls generated previously were used for comparison. Feline studies were conducted under University of Missouri (MU) IACUC approved protocol 9395. All animals were euthanized at the same time of day beginning at 10AM to control for effects that may be regulated by circadian rhythm, housed in a temperature controlled vivarium at 74°F +/−4°F on a 12-hour light/dark cycle. Cages are bedded with autoclaved Sani-chips and changed every 10–12 days for mice and every 7 days for rats. All animals were healthy at the time of euthanasia.

Pkd1^cond/cond:Nes^cre mice:

Pkd1 mice were group housed and separated by sex at weaning beginning at week 3 of age. These mice were described previously (Shillingford et al., 2010). Pkd1^cond/wt:Nes^Cre+/− mice were bred with Pkd1^cond/cond:Nes^Cre−/− mice to produce experimental cystic mice. Only female PKD mice were used during fasting experiments. Mice were genotyped using Pkd1 primers as follows: sense primer: ATT GGA CAA TGC CTG TGG TGA GTG; anti-sense primer: TCG TGT TCC CTT ACC AAC CCT CTT. Cre primers as follows: sense primer: ATG CTG TTT CAC TGG TTA TGC GGC; anti-sense primer: ACC AGC TTG CAT GAT CTC GGG TAT.

Han:SPRD rats:

Han:SPRD rats were obtained from the University of Oklahoma Health Sciences Center, courtesy of Dr. Benjamin Cowley, and a colony established at UCSB. Rats were group housed and given ad libitum access to food and water unless otherwise specified. Rats were weaned at 3 weeks of age and separated by sex. Rats were genotyped using a known polymorphism in the Ansk6 (Pkdr1) gene coding for SamCystin. Heterozygous females were bred with wild-type males to reduce distressful phenotypes. PCR coupled to restriction digestion was used detect a C to T transition as described previously by Brown et al. (Brown et al., 2005). DNA was extracted from a small piece of ear from each rat and PCR products were obtained from each DNA sample. Primers for rat SamCystin were as follows: sense primer: CTA GAA GCC TCA GTG ACC CC; anti-sense primer: CAG CGT GTG AAC AAG GTA GG. Amplification products were digested by Msp1 Fast Digest (Thermo-Fisher) at 37°C for 15 minutes. Digested PCR products were resolved on an agarose gel, producing bands at 157 and 85 bp; 242, 157, and 85 bp; and 242 bp for +/+, Cy/+, and Cy/Cy rats, respectively.

Time restricted feeding and ketogenic diet in juvenile PKD rats:

Treatment of Han:SPRD (Cy/+) rats on a time-restricted diet began at postnatal week 3 until week 8 of age (n =8 male and 12 female Cy/+ rats; n =13 male and 8 female wild-type rats). Animals were caged in cohorts between 2–6 rats and allowed water ad libitum. Animals were monitored daily for changes in behavior and for adverse effects of the diet. Food measurements and animal weights were performed daily. At weaning litters were separated by sex and then distributed randomly into group housing without consideration of genotype. Animals on a time-restricted diet received standard chow with a caloric ratio of ~62% carbohydrates, ~25% protein and ~13% fat, (Pico Lab Rodent Diet 20; LabDiet) for 8 hours daily starting 3 hours after beginning of the dark cycle. Animals (n =8 male and 7 female Cy+ rats; n =13 male and 6 female wild-type rats) on a ketogenic diet received ketogenic diet (#F3666 diet; Bio-Serv) comprised of a ~91% fat, ~2% carbohydrates and ~5% caloric ratio (See Suppl. Table 2 for detailed information).

Ketogenic diet in adult PKD rats:

Adult treatment began at week 8 until week 12 of age. Han:SPRD rats (n =12 male and 10 female cystic rats; n =10 male and 14 female wild-type rats) were fed ad libitum the pre-formulated ketogenic diet chow Envigo T.D. 110408 diet (Envigo) similar to ketogenic diet #F3666 but supplemented with 1.5g/kg methionine and 5g/kg choline (Suppl. Table 2). Animals were monitored as described above.

Blood Measurements:

Blood was collected via tail prick in all animals. Blood glucose was measured using a blood glucometer (Contour Next; Bayer), and β-hydroxybutyrate was measured using a blood ketone meter (Abbott; Precision Xtra). For TRF experiments, blood was collected prior to feeding after the 18 hour fasting period and is shown in Figure 1N, O and Figure S2. Acutely fasted and ad libitum access animal blood measurements were taken prior to sacrifice. Serum creatinine was measured using a creatinine assay kit (LifeSpan BioSciences, Inc.; LS-K207).

Acute rodent fasting:

Han:SPRD rats (n =8 male and 5 female cystic rats; n =7 male and 5 female wild-type rats), and Pkd1^cond/cond:Nes^cre mice (n =4 female cystic mice; n =3 male and 5 female wild-type mice) were fasted at 8 weeks of age for 48 or 24 hours, respectively, with ad libitum access to water. Animals were monitored for signs of distress during the treatment period. Water consumption was monitored daily.

ß-Hydroxybutyrate supplementation in rats:

Treatment of Han:SPRD rats began at postnatal week 3 until week 8 of age (n =7 male and 5 female cystic rats; n =3 male and 5 female wild-type rats). Animals were caged in cohorts between 2–6 rats and allowed water ad libitum. Animals were monitored daily for changes in behavior and for adverse effects of the diet. Food and water measurements were recorded daily and animal weights recorded weekly. During the experiment, animals had ad libitum access to normal chow (Pico Lab 20). A BHB salt formulation (sodium/potassium, KetoForce; KetoSports) (n =7 male and 5 female cystic rats; n =3 male and n =5 female wild-type rats) delivered in the drinking water at 157.5mM BHB final concentration as previously used in rats (Chang et al., 2017), offered ad libitum and changed every 3 days or sooner if needed. The KetoForce solution is a BHB salt with 1.6g sodium and potassium per 11.7g BHB. Salt controls (n =4 male and 4 female cystic rats; n =2 male and 5 female wild-type rats) were treated with 2.24g:2.24g (1:1 mass ratio) sodium/potassium (from NaCl and KCl) supplemented water equivalent to the BHB solution.

Feline PKD model:

The cats were members of an ADPKD colony housed at the University of Missouri, Columbia. Cats were genotyped for the PKD1 (10063C>A) mutation that causes feline ADPKD (Lyons et al., 2004). All cats were adult, ages ranged from 34.2–64.9 months, and were known to have cystic kidneys based on previous imaging diagnostics, including ultrasound, CT, and MRI (Suppl. Table 3). Cats were fed Persian Adult dry cat food (Royal Canin; protein, min 28.0%; fat, min 20.0%; fiber, max 6.0%) ad libitum. Each cat had preliminary CT imaging ~1.5 years (15–21 months) prior to the study. Cats were fasted for at least 12 hours prior to the preliminary diagnostic imaging and placed under general anesthesia using clinical ansthetic protocols as deemed appropriate by the anesthesia service of the Veterinary Health Center ast MU. Alfaxalone (1–4 mg/kg IV or IM), dexmedetomidine (0.005–0.020 mg/kg IM) buprenorphine (0.005–0.01 mg/kg IM), Propofol (1–8 mg/kg) and isoflurane (inhaled to effect) were used in protocols for sedation/general anesthesia protocols.

All cats were intubated, and anesthesia was maintained using isoflurane gas, and a catheter was placed in the right brachial vein. For CT imaging during the fasting trial, the cats were sedated with 20 mg/kg dexmedetomidine (Zoetis, Parsippany, New Jersey), i.m. Cats were recovered from sedation with 350 – 400 mg/kg atipamezole (Zoetis), i.m. After the final post-fasting CT imaging, two cats were not recovered from sedation and euthanized by barbiturate overdose to obtain kidneys for pathological and histological examinations.

Two cats had ideal weight (body condition score (BCS = 5), one female was slightly overweight, and one cat was overweight (BCS = 7–8). Each cat was bright, alert and active during the entire fast and did not appear dehydrated, lethargic or depressed. Three cats lost 0.3 kg of weight but one cat had the same weight post-fasting.

All cats were healthy and had unremarkable CBC, serum chemistries and urinalyses. No cats had evidence of renal failure previously or at the time of the trial as based on creatinine, urine specific gravity and symmetrical dimethylarginine (SDMA) levels. All PKD1 positive cats were clinically normal but due to the presence of cystic kidneys were classified as feline CKD Stage 1 (International Renal Interest Society; IRIS staging of CKD modified 2016; http://www.iris-kidney.com/guidelines/staging.html, accessed May 13, 2018).

Computed Tomography:

CT imaging of cats was conducted using a third generation 64 slice instrument (Aquillion 64, Canon Medical Systems, Tustin, CA). Each cat was placed in dorsal recumbency and pre- and post-contrast images were obtained from the diaphragm to the anus (slice thickness 1–2 mm, kvp 80–100, mAs 230–500). Intravenous non-ionic, iodinated contrast media (Omnipaque 350 Iohexol injection, GE Healthcare, Marlboro, MA.) was administered at a dose of 0.5 mg/kg and post-contrast images were obtained after a three-minute delay. For the fasting study, cats were fasted for ~12 – 13 hrs prior to sedation for the pre-trial CT imaging. After the pre-trial CT, the fast was continued up to ~60 hrs. The cats were then re-sedated and the post-trial CT imaging and blood sampling was conducted. Total kidney volumes were determined by semi-automated volumetric segmentation using 3D Slicer software (http://www.slicer.org (Fedorov et al., 2012)) using the Segmentation Wizard module (author: Andrew Beers, Athinoula A. Martinos Center for Biomedical Imaging at Massachusetts General Hospital; available at: https://www.slicer.org/wiki/Documentation/Nightly/Extensions/SegmentationWizard, accessed 9 Apr. 2019).

Method Details

Histology and Immunofluorescence microscopy

Tissue embedding and histology:

Tissues were removed and immediately placed into 4% paraformaldehyde for 24 hours then dehydrated through a decreasing alcohol series (2x 100%, 2x 95%, 2x 70%, 2x toluene) for 2 hours each and finally immersed in paraffin for 6 hours twice. Paraffin embedded tissues were then sectioned (5μm) and subjected to standard hematoxylin and eosin staining. To determine the cystic index, images of H&E sections were overlaid with a grid in Adobe Photoshop, and the intersection points on cysts vs. normal tissue were counted manually. To examine fibrosis, sections were deparaffinized and stained with 0.1% Sirius red and 0.1% Fast Green dissolved in picric acid on an orbital shaker, protected from light for 4 hours. Sections were then washed with tap water then dipped 10 times in increasing alcohol (95%, 100%) and cleared three times with xylene. Images were obtained for quantification by grid-overlay as described above. With the exclusion of blood vessels and absent tissue (e.g. Cysts), intersection points on Sirius red-positive structures were counted vs. total intersections.

Oil O Red Staining:

Frozen sections were obtained by fixation of kidneys for 24 hours in formalin followed by incubation in 15% sucrose-PBS until tissues float, then incubation in 30% sucrose-PBS until tissues float again. Tissues were then placed in OCT imbedding compound (Fisher) and frozen using 100% ethanol in dry ice. 10μm sections were then cut and fixed with formalin. Sections were then rinsed in distilled water and washed with propylene glycol 2x 5 minutes followed by Oil O Red staining (0.7% (w/v) Oil O Red (Sigma-Aldrich) in propylene glycol) for 7 minutes with agitation. Sections were then incubated in 85% propylene glycol for 3 minutes, rinsed in distilled water and mounted using gelatin based aqueous mounting media (10g gelatin, 60ml distilled water, 70ml glycerol and 0.25g phenol; prepared in house).

Immunofluorescence staining:

Antigens were retrieved in 10 mM citrate-Na at pH 6.0 in a pressure cooker for 15 minutes. Slides were then blocked in (1% BSA, 0.1% fish skin gelatin, 0.1% sodium azide in TBST) at 37°C for 30 minutes. Sections were then treated with 0.1% Sudan Black in 70% ethanol for 20 minutes to reduce tissue auto fluorescence. After washing with TBST, slides were incubated in primary antibody overnight at 4°C. The following antibodies were use d for immunostaining: Rabbit anti-pS6^S235/236, (#4858; Cell Signaling); Mouse-α-Smooth Muscle Actin (A5228; Sigma-Aldrich), Rabbit-Ki-67 (AB9260; EMD Millipore); Rabbit anti-pSTAT3^Y705 (#9145; Cell Signaling). The sections were then incubated with species-specific fluorescently labeled secondary antibodies, diluted 1:200 in blocking buffer at 37°C for 1 hour. The following secondary antibodies were used for immunostaining: (Goat anti-Rabbit-DyLight 488 and 594, #35552 and #35560, ThermoFisher; goat anti-mouse-DyLight 488, ThermoFisher #35502). After washing in TBST, slides were post-fixed in neutral-buffered formalin for 10 minutes, washed in DI water and mounted with ProLong Gold antifade mounting medium with DAPI (ThermoFisher). The DeadEnd Fluorometric TUNEL system (Promega) was utilized for TUNEL staining and performed as specified by the manufacturer.

Myofibroblast Quantification:

The prevalence of renal myofibroblasts was estimated by quantification of images stained for smooth muscle actin (SMA-1) with the exclusion of blood vessels. Grid intersection points on SMA-1 positive structures were counted vs. total intersections excluding non-tissue intersections.

Ki-67 Quantification:

Ki-67 immunostained kidney sections were imaged and total number of DAPI positive nuclei were analyzed using FIJI (ImageJ; NIH) image software. At least 10 cortical areas were imaged and the number of Ki-67 positive cells were counted and parsed based on their location within the kidney. Percentages of Ki-67 positive cells are expressed as a fraction of the total counted nuclei.

Phospho-S6 Quantification:

pS6 stained sections were imaged and analyzed for pS6 activity within cysts. Approximately 1000 cells were counted within cysts and the percentage of pS6 positive cells were expressed as a percentage of the total number of cells counted.

Cyst Size and Cyst Number Quantification:

Whole kidney sections were imaged and the images were quantified using FIJI (ImageJ; NIH) image software. Individual cysts were counted and their sizes were recorded.

Immunoblotting

For sample preparation, kidneys were removed from euthanized animals, bisected and placed into a mortar cooled with liquid nitrogen, submersed in liquid nitrogen and pulverized using a pestle. These samples were then placed into microcentrifuge tubes and stored at −80°C for future analysis. Approximately 10mg of tissue were lysed in 200μL of 4% SDS-100mM Tris pH 6.8 lysis buffer containing phosphatase inhibitors (Phosphatase inhibitor cocktail 2 & 3 (Sigma-Aldrich) and protease inhibitor cocktail (Sigma-Aldrich). Samples were vortexed and heated at 100°C for 10 minutes followed by vortexing and centrifugation to pellet insoluble debris. Protein concentrations were estimated by reading the absorbance at 280nm using a NanoDrop spectrometer. Following standard SDS-PAGE and transfer onto nitrocellulose membranes, samples were analyzed using the following antibodies: Mouse anti-Actin (N350; Amersham Bioscience); Rabbit phopsho-AMPK^Thr172 (#2535; Cell Signaling); Mouse total-AMPK, (#2793; Cell Signaling); Mouse anti-CPT1α, (#8F6AE9; Abcam).

Quantification and Statistical Analysis

The linear regression model analyses were performed with SAS 9.4 statistical software package (SAS Institute Inc., Cary NC). Following tests for normal distribution, the statistical analyses were performed using Mann-Whitney unpaired one-tailed Student’s t-test. Analysis was performed using Prism (GraphPad) software. Details for statistics can be found within the figure legends. PKD animals were housed with litter mates in groups between 2–4 animals without regard to genotype, resulting in a random distribution of PKD and wild-type animals in cohorts. Experimenters were not blinded to the treatment of genotypes of animals. The analysis of collagen, cystic index and smooth muscle actin images was conducted blinded. Exclusion criteria were based upon animal well-being. No animals were excluded from this study. No power analysis was done to determine sample sizes. Sample sizes were chosen based on experience with previous studies with PKD animals in our lab.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti-Phospho-S6 Ribosomal Protein (Ser235/236) (D57.2.2E) XP® Rabbit mAb	Cell Signaling	Cat#4858; RRID: AB_2721245
Anti-Smooth Muscle Actin (1A4) Mouse mAb	Sigma-Aldrich	Cat#A5228; RRID: AB_262054
Anti-Ki-67 Rabbit pAb	EMD Millipore	Cat#AB9260 RRID: AB_2142366
Anti-Phospho-STAT3 (Tyr705) (D3A7) XP® Rabbit mAb	Cell Signaling	Cat#9145; RRID: AB_2491009
Anti-Rabbit IgG (H+L) Secondary Goat Ab, DyLight 488	ThermoFisher	Cat#35552 RRID: AB_844398
Anti-Rabbit IgG (H+L) Secondary Goat Ab, DyLight 594	ThermoFisher	Cat#35560; RRID: AB_1185570
Anti-Mouse IgG (H+L) Secondary Goat Ab, DyLight 488	ThermoFisher	Cat#35502; RRID: AB_844397
Anti-Actin mAb (AC-15)	Amersham Bioscience	Cat#N350
Anti-Phospho-AMPKα (Thr172) (40H9) Rabbit mAb	Cell Signaling	Cat#2535; RRID: AB_331250
Anti-AMPKα (F6) Mouse mAb	Cell Signaling	Cat#2793
Anti-CPT1α (8F6AE9) Mouse mAb	Abcam	Cat#ab128568; RRID: AB_11141632
Chemicals, Peptides, and Recombinant Proteins
Oil O Red (1-([4-(Xylylazo)xylyl]azo)-2-naphthol, 1-[2,5-Dimethyl-4-(2,5-dimethylphenylazo)phenylazo]-2-naphthol, Solvent Red 27)	Sigma-Aldrich	Cat#O0625
Protease inhibitor Cocktail (ingredients listed below)	In house
Pepstatin 7.3mM	Chemicon	Cat#E19
Leupeptin 11.72mM	Chemicon	Cat#E18
Antipain 14.76mM	Chemicon	Cat#E13
Benzamidine 450mM	Sigma-Aldrich	Cat#12072
Trasylol 10000U/mL	Sigma-Aldrich	Cat#A6279
Phosphatase Inhibitor Cocktail 2	Sigma-Aldrich	Cat#P5726
Phosphatase Inhibitor Cocktail 3	Sigma-Aldrich	Cat#P0044
KetoForce BHB Salt	Ketosports	Cat#PTN1010/929/1501
Dexmedetomidine	Zoetis	CAS:104075-48-1
Antisedan® Atipamezole Hydrochloride	Zoetis	PubChem CID:71310
Propofol		CAS:2078-54-8
Buprenorphine		CAS:52485-79-7
Alfaxalone		CAS:23930-19-0 PubChem ID: 104845
Omnipaque 350 Iohexol	GE Healthcare	CAS:66108-95-0
Msp1 Fast Digest	Thermo-Fisher	FD0544
Critical Commercial Assays
DeadEnd Fluorometric TUNEL system	Promega	Cat#G3250
Experimental Models: Organisms/Strains
Mouse: Pkd1Cond/Cond Nestin-Cre/(C57BL/6)	Shillingford et al. 2010	N/A
Rat: PKD Han:Sprague-Dawley	Cowley et al. 1993	N/A
Oligonucleotides
Pkd1 Genotype sense primer: ATT GGA CAA TGC CTG TGG TGA GTG	This Paper	N/A
Pkd1 Genotype anti-sense primer: TCG TGT TCC CTT ACC AAC CCT CTT	This Paper	N/A
Cre Genotype sense primer: ATG CTG TTT CAC TGG TTA TGC GGC	This Paper	N/A
Cre Genotype anti-sense primer: ACC AGC TTG CAT GAT CTC GGG TAT	This Paper	N/A
Anks6 (Pkdr1) Genotype sense primer: CTA GAA GCC TCA GTG ACC CC	Brown et al. 2005	N/A
Anks6 (Pkdr1) Genotype anti-sense primer: CAG CGT GTG AAC AAG GTA GG	Brown et al. 2005	N/A
Software and Algorithms
FIJI ImageJ	National Institutes of Health	https://fiji.sc/
Prism 8	GraphPad	https://www.graphpad.com/scientific-software/prism/
3D Slicer software	Fedorov et al., 2012	http://www.slicer.org
Photoshop CS5	Adobe
Other
Aquillion 64 Imager	Canon Medical Systems	https://us.medical.canon/
Pico Lab® Rodent Diet 20 5R53	LabDiet	Cat#3002890-742
Ketogenic Diet AIN-76A-Modified, High Fat, Paste	Bio-Serv	Cat#F3666
Teklad Custom Diet-Ketogenic Diet Very Low Protein (VLP)	Envigo	Cat#TD.110408
Persian Adult Dry Cat Food	Royal Canin	Cat#843507
Contour Next Blood Glucose Monitor	Bayer	Model#9628
Precision Xtra Blood Ketone Montior	Abbott	SKU#9881465

Highlights.

1. Ketosis prevents or reverses polycystic kidney disease (PKD) in animal models.

2. Dietary changes that induce ketosis prevent PKD.

3. Oral beta-hydroxybutyrate supplementation alone inhibits PKD progression.

4. Cystic cells are metabolically inflexible which can be exploited for therapy.

Context and Significance Statement.

Polycystic kidney disease (PKD) is a common genetic disease that leads to kidney failure. Currently the only approved treatment for this condition has only modest effects. However, calorie restriction was previously shown to inhibit disease progression in mouse models of PKD but the mechanism remained unknown. Using mouse, rat and cat models of PKD, Thomas Weimbs and colleagues now show that the metabolic state of ketosis is important for inhibiting PKD progression. They also show that simply adding the natural ketone beta-hydroxybutyrate (BHB) to a normal diet replicates the beneficial effects of a ketogenic regimen. These results suggest that cystic cells in PKD are metabolically inflexible, which can be exploited by a simple dietary intervention to provide a new therapy for PKD.