SUPPRESION OF MITOCHONDRIAL RESPIRATION IS A FEATURE OF CELLULAR GLUCOSE TOXICITY

ABSTRACT

Glucose toxicity is central to the myriad complications of diabetes and is now believed to encompass neurodegenerative diseases and cancer as well as microvascular and macrovascular disease. Due to the widespread benefits of SGLT2 inhibitors, which affect glucose uptake in the kidney proximal tubular cell, a focus on cell metabolism in response to glucose has important implications for overall health. We previously found that a -Warburg-type effect underlies diabetic kidney disease and involves metabolic reprogramming. This is now supported by quantitative measurements of superoxide measurement in the diabetic kidney and systems biology analysis of urine metabolites in patients. Further exploration of mechanisms underlying mediators of mitochondrial suppression will be critical in understanding the chronology of glucose-induced toxicity and developing new therapeutics to arrest the systemic glucose toxicity of diabetes.

INTRODUCTION

Given the global pandemic of diabetes, the incidence and prevalence of diabetic complications are expected to continue rising. Classically, the microvascular complications of diabetes such as diabetic retinopathy, neuropathy, and nephropathy were well recognized and known to be mitigated with tight glycemic control, especially in the early stages of the disease (1). Subsequently, macrovascular diseases, including myocardial infarctions, heart failure, stroke, and peripheral vascular disease, were recognized to be the major causes of mortality in patients with diabetes. More recently, diabetes has been found to be a risk factor for cognitive dysfunction leading to Alzheimer's disease and various types of cancers (2).

Ultimately, the manifestations of a particular or multiple diabetic complications will involve an interplay of genetic predisposition, age of onset, and cellular response to the hyperglycemic milieu. Of these factors, the genetic predisposition to diabetic complications has not as yet been completely unraveled and may be a minor contributor (3,4). Therefore, substantial research effort has focused on understanding the cellular response to hyperglycemia to direct new therapies. Examination of the cellular response to glucose will involve a close examination of bioenergetics and mitochondrial function. In recent studies, based on new imaging data and metabolomics, mitochondrial dysfunction has been found to play a central role in understanding cellular physiology to glucose, and a host of mitochondrial-based therapies are being developed. Our group examined recent developments in the field and proposed a basis of glucose toxicity involving both the Warburg and Crabtree effects (PMID 29602394 and PMID 37091239). These two effects have been described in cancer cells originally and describe a phenomenon of suppressed mitochondrial oxidative phosphorylation despite availability of oxygen. In kidney cells, both a short-term Crabtree effect and a chronic Warburg effect have been recently found, and the search is on for identifying the mechanisms of the acute Crabtree and chronic Warburg effects.

MATERIALS AND METHODS

Superoxide Measurement in Mice With Dihydroethidium

Superoxide measurements were previously published by our group (5). Briefly, a model of type 1 diabetes (T1D) F1 C57BL/6J-Ins2^Akita male mice and age-matched C57BL/6J wild type (WT) mice (n=5-6 per group) were administered dihydroethidium (DHE) (50 mg/kg) by i.p. injection as previously described (6). Measurements of DHE and its oxidation products ethidium (E⁺) and 2-hydroxyethidium (2-OH-E⁺) in kidney tissues were performed by high performance liquid chromatography (HPLC) using published protocols with minor optimization (7,8). Comparisons of DHE oxidation products in kidney tissues between Akita T1D mice and WT controls were conducted using t test by GraphPad. Significance was defined as p < 0.05.

Bioinformatics Analysis Based on Urine Metabolites

Network analysis of urine metabolites with enzymes and protein-protein interaction (PPIs) was performed using Cytoscape (9,10). Integrated multi-omics analyses of metabolites and proteins (i.e., enzymes) identified mitochondrial-related metabolic pathways and networks. Briefly, the data necessary to create the shown network (“MetBridge” network) were metabolic pathways and PPIs, and they were obtained from global human metabolic pathway map (http://www.genome.jp/kegg/) and BioGRID database (http://thebiogrid.org). These two data sources were integrated by converting enzymes that appear in metabolic pathways into protein information. We also added metabolic pathways involving the 13 urinary metabolites that we discovered to be linked to diabetic kidney disease (DKD) (9). For further details, please refer to our published methods (11).

RESULTS

In Vivo Imaging With DHE Demonstrates Reduction of Superoxide in Diabetic Kidneys

Mice with T1D were administered DHE, and superoxide was evaluated by HPLC analysis of ethidium (E⁺) or 2-hydroxyethidium (2-OH-E⁺). We had previously found via confocal imaging a reduction in kidney levels of DHE fluorescence indicating reduction of superoxide in T1D (5); however, E⁺ and 2-OH-E⁺ have overlapping “red fluorescence” spectra. As it is challenging to quantify these two moieties using fluorescent-based techniques (7) in the current study, E⁺ and 2-OH-E⁺ were quantified using HPLC. We found a reduction of both moieties specific for superoxide responsive to DHE (Figure 1) in the model of Akita T1D mice. Interestingly, E⁺ was the major oxidation product of superoxide in the kidneys of both WT control and Akita T2D mice (Figure 1).

Fig. 1.

Akita mouse kidneys had a reduction of the moieties specific for superoxide responsive to DHE. DHE oxidation products such as ethidium (E⁺) or 2-hydroxyethidium (2-OH-E⁺) were significantly lower in Akita T1D mice compared to WT controls (data were normalized to DHE, n=5-6 per group).

Bioinformatic Analysis of Urine Metabolomics Indicates Reduced Mitochondrial Function

Urine metabolites that were differentially regulated in patients with diabetes in the presence or absence of kidney disease (based on eGFR and ACR) were analyzed via Cytoscape (Figure 2). Based on the network analysis and overall reduction of urine metabolites linked to mitochondria, there was an overall reduction of mitochondrial function as indicated by the bioinformatic analysis.

Fig. 2.

Network analysis by Cytoscape showed an overall reduction of mitochondrial function. Integrated multi-omics analysis of metabolites and proteins/enzymes identified novel links and pathways and 13 mitochondrial-related metabolites, which were reduced in diabetic patients with kidney disease. Each node and edge represent a molecule (metabolite, enzyme, or protein) and interaction among a pair of two molecules, respectively. Metabolites (hexagons) with thick borders represent one of 13 metabolites which were localized in mitochondria. Enzymes (squares) with thick borders represent those which are involved in the reaction with the metabolites -localized in mitochondria. Such reactions are represented by wavy edges. An edge width representing protein-protein interactions is proportional to confidence of the -interaction. The size of proteins are proportional to the calculated significance of the number of the interactions.

DISCUSSION

The present study provides in vivo perspective of superoxide production in the diabetic kidney by demonstrating that DHE measurable superoxide levels are actually reduced in the kidney of diabetic mice using a quantitative approach. Further, the metabolites that were found to be reduced in patients with DKD identify a global reduction of mitochondrial function by a systems biology approach. These results indicate that reduced mitochondrial function may form the basis for chronic long-term decline of mitochondrial function and structure as a characteristic of DKD.

Reactive Oxygen Species, Superoxide, and Diabetes

Superoxide production is typically considered to be a byproduct of respiring mitochondria, and it was believed that excess glucose input would lead to a consequent increase in mitochondrial superoxide levels (12). However, it appears that the majority of cells within the mouse kidney have a reduction in superoxide levels with diabetes. The in vivo quantification of superoxide by DHE in the present study provides further evidence that superoxide levels are reduced in the diabetic kidney; therefore, mitochondrial oxidative phosphorylation is also likely to be reduced. If mitochondrial respiration is reduced, one would also need to consider the concepts of the Warburg and Crabtree effects as central to DKD.

Systems Biology and Implications for Human Diabetic Complications

Mitochondrial function is a dynamic process and difficult to quantify in the context of nutrient stress conditions. Several studies have determined that mitochondrial dysfunction is a characteristic of DKD (13). The term “mitochondrial dysfunction” is broad and represents findings related to disrupted mitochondrial dynamics, uncoupling of mitochondria, oxidative damage to mitochondria, decreased mitochondrial respiratory capacity, decreased (mitochondrial DNA) mtDNA content, and decreased antioxidant capacity (13). Due to the difficulty in quantifying oxidative phosphorylation and mitochondrial function in a heterogeneous organ such as the kidney, a global measure of mitochondrial function in the human condition would provide a useful perspective. Several studies have measured urine metabolomics in patients with DKD (9,14) and bulk metabolomics from experimental animal models (15). These studies have consistently found an altered mitochondrial function in DKD and suppression of glucose-induced mitochondrial oxidative phosphorylation (16). The systems biology approach in the present study supports a global dysfunction of oxidative phosphorylation in human DKD based on key metabolites that are reduced in the urine of patients with DKD.

Warburg and Crabtree to Explain Cellular Physiology to Glucose Toxicity

Both the Warburg (17,18) and Crabtree (19,20) effects were first described in cancer cells, and both describe a process of reduced mitochondrial respiration despite availability of oxygen and a shift to aerobic glycolysis. However, their differences are partly due to timing, with the Crabtree effect describing the acute effect of glucose in suppressing oxidative phosphorylation and the Warburg effect showing a sustained effect of cells in suppressing oxidative phosphorylation and enhancing glycolysis. In the diabetic kidney, the sustained effect of tubular cells in enhancing glycolysis and suppressing oxidative phosphorylation is consistent with a Warburg-type effect. Interestingly, this characteristic Warburg effect is common in several types of kidney disease, such as polycystic kidney disease, and inhibition of aerobic glycolysis appears to be protective against disease progression (21,22), possibly by inhibiting interstitial fibroblast activity (23). A recent study suggested that fructose and uric acid play roles in chronic kidney disease to mediate the Warburg effect (22).

A role for the Crabtree effect has not yet been demonstrated but may underlie some of the initial effects of glucose to initiate a metabolic program leading to inflammation, matrix production, and a sustained Warburg-type effect. It is now well described that acute effects of glucose have initial signaling effects that could lead to changes in key bioenergetic pathways such as downregulation of 5’ adenosine monophosphate-activated protein kinase (AMPK) and upregulation of mammalian target of rapamycin (mTOR). These changes to key energy sensing pathways are fundamental to initiation and progression of DKD and may well be driven by an initial Crabtree-type effect (PMID 37091239). The identification of key intermediates of Crabtree and Warburg effects will likely have important implications in mediating the acute and sustained effects of enhanced glucose exposure.

DISCUSSION

Weir, Baltimore: Very nice, Kumar. I have a quick question for you. At what level of glucose do you start to see these effects, both in vitro and in vivo, and is there a correlation between the glucose levels and the degree of changes that you're describing here?

Sharma, San Antonio: Yes, that's a great question. We actually found it to be caused even with very low levels of glucose in cell culture. The proximal tubular cells actually like zero glucose, at least for the short period of time, during which the cells have the maximum oxygen consumption rate and mitochondrial activity. So even with 2.5 and 5 millimolar levels of glucose, we start seeing a reduction of oxidative phosphorylation (OXPHOS) activity in vitro. Obviously, evaluating zero glucose can't be done in the in vivo situation, but in the in vivo state with humans, we were starting to see lactic acid elevation even with modest levels of increased glucose.

Weir, Baltimore: And do those changes affect cellular autophagy as well? That may be another hypothesis linking these alterations with obvious impairment of cellular function.

Sharma, San Antonio: Yes, that's certainly likely, and I think turnover of proteins and mitochondrial turnover is likely going to be affected as well.

Desir, New Haven: Good morning. Your findings are fascinating. Do you think that what you're finding in kidney cells actually might be relevant to type 1 diabetes where the switch from OXPHOS to glycolysis increases cell toxicity in the pancreas and kills the beta cells?

Sharma, San Antonio: That's a great question. We did our hyperglycemic studies in type 1 diabetic patients, so the glycolytic processes are occurring in patients with type 1 diabetes. However, we have not evaluated the effect on islet cells. I am not sure if other people have looked at the Crabtree effect in islet cells.

Desir, New Haven: The other thing I would say is that there are certain circulating factors that actually drive the Warburg effect or the Crabtree effect and that may be one way to look at it also.

Sharma, San Antonio: Great, thank you.

Chi-yuan Hsu, San Francisco: Hi, Kumar. Thanks for the talk. Can you explain a little bit about whether these would correlate with the observation in humans that glycemic controls seem much more important in human disease early on, to prevent the onset of microvascular disease, whereas once you have established disease, glycemic control seems less useful and potentially even harmful, at least in human diabetic kidney disease?

Sharma, San Antonio: Yes, those are great questions. It's hard to correlate these findings with human observations, but we do see that patients with early diabetes and poor control have a faster rate of progression. It is likely that an early effect of hyperglycemia to reduce mitochondrial function sets the stage for progressive disease once they have loss of mitochondria. We think that may be happening, but it hasn't been proven and may help to explain part of the glycemic memory effect that we see with short-term poor glycemic control in patients.

Carethers, Michigan: You pointed out that diabetes affects many organs and the diabetic findings in the liver. Are the same processes in the Crabtree pathway going on in the liver as well as perhaps in nervous tissue? These two tissues at the structural level of mitochondria are a little bit different.

Sharma, San Antonio: We did look at liver cells, and there seems to be sort of an in-between effect. They act more like proximal tubular cells than many other cell types though not quite as profoundly, but we still have to complete those studies as we're just in the beginning stages.

Reeves, San Antonio: Great talk, Kumar. One of your slides showed the GLUT transporters (membrane proteins that transport glucose and related hexoses) as mediating the glucose entry. I’m curious what your thoughts are about sodium-dependent glucose co-transporters (SGLT) and whether this mechanism might explain some of the beneficial effects we see clinically.

Sharma, San Antonio: Yes, great question, Brian. We find that if we incubate these proximal tubular cells with sodium-glucose transport protein 2 (SGLT2) transporters we completely block lactic acid production in response to glucose, so we do think that blocking the Crabtree effect could explain part of the benefit of SGLT2 inhibition.