Mitochondrial DNA Polymerase Editing Mutation, Polg , Reduces The

Home , POLG

Mitochondrial DNA polymerase editing mutation, PolgD257A, reduces the diabetic phenotype of Akita male mice by suppressing appetite

Raymond Foxa,b, Hyung-Suk Kimb, Robert L. Reddickc, Gregory C. Kujothd,e, Tomas A. Prollad, Shuichi Tsutsumif, Youichiro Wadaf, Oliver Smithiesa,b,1, and Nobuyo Maedaa,b,1

aCurriculum in Genetics and Molecular Biology, bDepartment of Pathology and Laboratory Medicine, University of North Carolina, Chapel Hill, NC 27599; cDepartment of Pathology, University of Texas Health Sciences Center, San Antonio, TX 78229; Departments of dGenetics and Medical Genetics and eNeurological Surgery, University of Wisconsin, Madison, WI 53792; and fInstitute of Systems Biology and Medicine, University of Tokyo, Tokyo 153-8904, Japan

Contributed by Oliver Smithies, April 20, 2011 (sent for review February 17, 2011) Diabetes and the development of its complications have been The enzymes required for mitochondrial biogenesis are enco- associated with mitochondrial DNA (mtDNA) dysfunction, but ded in the nuclear genome, including mtDNA polymerase γ causal relationships remain undetermined. With the objective of (Polg), which replicates and proofreads the mitochondrial ge- testing whether increased mtDNA mutations exacerbate the di- nome. Polg is essential for life, and its complete absence leads to abetic phenotype, we have compared mice heterozygous for the early embryonic death (10). An amino acid substitution, D257A, Akita diabetogenic mutation (Akita) with mice homozygous for the in the exonuclease domain II of Polg ablates proofreading without D257A mutation in mitochondrial DNA polymerase gamma (Polg) signiﬁcantly affecting the capacity of the polymerase to replicate or with mice having both mutations (Polg-Akita). The Polg-D257A mtDNA (11, 12). Mice homozygous for the D257A mutation protein is defective in proofreading and increases mtDNA muta- (Polg) die prematurely by 13 mo of age, displaying a series of tions. At 3 mo of age, the Polg-Akita and Akita male mice were aging-associated phenotypes due to increased random accumu- equally hyperglycemic. Unexpectedly, as the Polg-Akita males aged lations of mtDNA mutations that lead to increased apoptosis, to 9 mo, their diabetic symptoms decreased. Thus, their hypergly-

particularly in tissues containing cells that are metabolically active MEDICAL SCIENCES cemia, hyperphagia and urine output declined significantly. The (11, 12). The D257A mutation, which does not itself induce di- decrease in their food intake was accompanied by increased plasma abetes, consequently provides a tool that can be used to determine leptin and decreased plasma ghrelin, while hypothalamic expres- whether an increased frequency of mtDNA mutations can in- sion of the orexic gene, neuropeptide Y, was lower and expression fluence the pathophysiology of diabetes and its complications. of the anorexic gene, proopiomelanocortin, was higher. Testis func- In the present work, we take advantage of this tool to examine tion progressively worsened with age in the double mutants, and the effects of the Polg-D257A mutation in mice that are diabetic plasma testosterone levels in 9-mo-old Polg-Akita males were because of the dominant Akita mutation (C96Y) in the Ins2 gene. significantly reduced compared with Akita males. The hyperglyce- The C96Y mutation prevents the formation of an important mia and hyperphagia returned in aged Polg-Akita males after disulfide bond in insulin and leads to misfolding of the protein. testosterone administration. Hyperglycemia-associated distal tu- The misfolded protein has impaired transport through the en- bular damage in the kidney also returned, and Polg-D257A-associated doplasmic reticulum (ER), which triggers an ER stress reaction proximal tubular damage was enhanced. The mild diabetes of leading to pancreatic β-cell apoptosis (13). Males heterozygous female Akita mice was not affected by the Polg-D257A mutation. for the Akita mutation, but not females, develop hyperglycemia We conclude that reduced diabetic symptoms of aging Polg-Akita begining at ≈3 wk of age and becoming progressively more severe males results from appetite suppression triggered by decreased with age. Surprisingly, our experiments show that the diabetic testosterone associated with damage to the Leydig cells of the testis. symptoms of Akita male mice, which also have the PolgD257A mutation (Polg-Akita), become progressively less severe after iabetes mellitus is becoming increasingly common world- 3 mo of age. We demonstrate that the overall improvement in the Dwide. A large body of data demonstrates links between mi- diabetic phenotype of the Polg-Akita mice is at least in part due tochondrial dysfunction, insulin resistance, and diabetes (1, 2). to a reduction in the Akita-induced hyperphagia triggered by Deleterious mitochondrial DNA (mtDNA) mutations have their a Polg-induced decrease in testis production of testosterone. greatest effect in cells that require high energy production, or Glomerular changes were mild in all of the diabetic mice. Prox- already have high oxidative stress, including neurons, hair cells of imal tubular damage occurred only in aged double mutant males. the inner ears, heart and skeletal myocytes, pancreatic beta cells, Distal tubular damage occurred in all hyperglycemic males. – and gut and kidney epithelial cells (3 7). Complications of di- Results abetes also involve damage to these metabolically active cell Mitochondrial Mutations in Polg-Akita Mice. The reported median types, leading to a hypothesis that mtDNA mutations contributes lifespan of Polg mice homozygous for Polg-D275A mutation is to the complications. Consistent with these findings, we have demonstrated in mice that a genetic absence of the bradykinin B2

receptor (B2R) combined with diabetes caused by the Akita mu- Author contributions: R.F., O.S., and N.M. designed research; R.F., H.-S.K., R.L.R., S.T., and C96Y tation, Ins2 , progressively increases modiﬁcations to kidney Y.W. performed research; G.C.K. and T.A.P. contributed new reagents/analytic tools; R.F., mtDNA, including 8-hydroxy-2′-deoxyguanosine content, point O.S., and N.M. analyzed data; R.F., O.S., and N.M. wrote the paper; and G.C.K. and T.A.P. provided Polg mice. mutations, and deletions (8, 9). The mtDNA mutations in these The authors declare no conﬂict of interest. mice are also associated with enhanced nephropathy and senes- Freely available online through the PNAS open access option. cence-related phenotypes in multiple tissues (8, 9). However, it 1To whom correspondence may be addressed. E-mail: [email protected] remains unclear whether the mtDNA mutations act causally with or [email protected]. respect to diabetes and its complications, or are consequences of This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. these conditions. 1073/pnas.1106344108/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1106344108 PNAS Early Edition | 1of6 Downloaded by guest on September 24, 2021 13 mo, and the amount of mtDNA mutations accumulating in variables were markedly higher than in nondiabetic mice. Urine − Polg mice has been estimated to be as high as 1 × 10 3 mutations glucose levels at 3 mo of age (Fig. 2C) were comparable in Polg- − per bp in most tissues compared with <5 × 10 4 mutations per bp Akita and Akita mice. However, as the hyperglycemia of the in wild-type mice (11, 12, 14, 15). Because untreated Akita di- male Polg-Akita mice improved with age, their water consump- abetic mice deteriorate quickly after 9 mo (16), we determined tion, urine output, and urinary glucose also declined and, at 9 mo mtDNA mutation load in the liver of 9-mo-old mice, which are of age, they were no longer different from those of nondiabetic histologically intact at this age. As expected, the average number mice. Urinary potassium, sodium, and chloride levels did not of base substitution per site per hundred mtDNA molecules, differ in the diabetic mice regardless of the presence of the without correcting for errors associated with deep sequencing PolgD257A mutation. Despite the improvement of hyperglycemia, and heteroplasmic polymorphisms, was significantly higher in the daily urinary albumin excretion was similar in the Polg-Akita and Polg mice than in their wild-type (WT) littermates (0.425 ± 0.005 Akita males although it was only modestly greater than in non- mutations per site per hundred molecules sequenced vs. 0.375 ± diabetic males throughout the 9 mo of study (Fig. 2D). 0.004; P ≤ 0.001). Diabetes, again as expected, further increased The glomeruli showed no notable differences between the base substitution rate (P < 0.001), with the Polg-Akita mice diabetic Akita mice and the Polg-Akita mice under both light having significantly more mutations than their Akita littermates microscopy (Fig. 2 E and F) or Electron microscopy (Fig. S3). No (0.557 ± 0.007 vs. 0.494 ± 0.008; P ≤ 0.001). We conclude that evidence of fibrosis or areas of necrosis were found. In the distal the mtDNA of 9-mo-old Polg-Akita mice has accumulated sig- tubular epithelial cells, glycogen deposits in the central portion nificantly more mutations than the mtDNA of their Akita siblings. of nuclei were present in both the Akita and Polg-Akita kidneys, The expressions of mitochondrial transcription factor A but were more prominent in the Akita mice than in Polg-Akita (Tfam), mtDNA polymerase γ (Polg), and cytochrome b (CytB) mice (Fig. 2 E and F, long arrows). In the proximal tubular ep- have been correlated, respectively, with mitochondrial biogenesis ithelial cells, in contrast, cytoplasmic clear foamy inclusions were (17), replication (18), and function (19). However, renal and present only in the Polg-Akita kidneys (Fig. 2F, arrowheads), hepatic gene expression of Tfam, Polg, and CytB did not differ whereas no such changes were seen in either the Akita kidneys or fi signi cantly among the groups (Table S1). Similarly, mtDNA in nondiabetic Polg kidneys of the same age. Ultrastructurally, fi content in these tissues was not signi cantly different between the foamy cytoplasmic inclusions seen by light microscopy in the the two genotypes (Fig. S1). We conclude that the Polg mutation proximal tubular cells of the Polg-Akita kidneys proved to be did not change mitochondrial biogenesis and replication in our markedly enlarged lysosomes containing electron-dense material diabetic mice. and calcifications (Fig. 2H). Lamellated inclusions of glycolipids mtDNA damage associated with diabetes also includes an in lysosomes are correlated with increased damage and turnover increased incidence of deletions (8, 9). However, we found no fi of mitochondria within cells, and some inclusions can be seen in signi cant differences in the incidence of D-17 deletions in the Polg-Akita kidneys, although most of the mitochondria appeared liver, pancreas, kidneys, and small intestine of the Polg-Akita normal except for some variations in sizes and density. No mice compared with the Akita mice (Fig. S1). inclusions were seen in the Akita littermates (Fig. 2G). Taken together, these data show that the distal renal tubular damage, Improved Diabetic Symptoms in Aging Male Polg-Akita Mice. At which occurs in the Akita mice in association with their severe 3 mo of age, both Akita and Polg-Akita male mice exhibited all hyperglycemia, disappears in the Polg-Akita mice, in association of the symptoms normally associated with type 1 diabetes. Thus, with their markedly decreased hyperglycemia, but proximal tu- both had markedly elevated plasma glucose levels (491 ± 33 mg/ bular damage is now apparent because of mitochondrial damage dL; Fig. 1A) and food intake (8.5 g/d; Fig. 1B) compared with their nondiabetic counterparts. In contrast, at 6 mo of age, caused by the Polg mutation. plasma glucose and food consumption in the Polg-Akita mice began to show significant decreases relative to the Akita mice. By 9 mo of age, the fasting plasma glucose levels in the Polg-Akita mice (280 ± 50 mg/dL) had decreased still further compared with their littermate Akita mice (475 ± 54 mg/dL), although they were still significantly higher than in nondiabetic WT (167 ± 10 mg/ dL) and Polg mutant (151 ± 10 mg/dL) mice. The food intake of the Polg-Akita mice was dramatically lower than that of littermate Akita mice, and was no longer different from that of nondiabetic mice. In contrast, the body weight of the Polg-Akita mice increased steadily with age, and at 9 mo, they were significantly heavier than their Akita littermates (26.3 ± 1.2 g vs. 21.6 ± 1.2 g; P ≤ 0.01, Fig. 1C). The weight gain in the Polg-Akita mice was associated with significantly larger amounts of s.c. and visceral fat (Fig. 1D), although still below the amount observed in WT mice. Diabetes in our female Akita mice was mild, as has been reported (16); it was not affected by the PolgD257A mutation (Fig. S2). Thus, plasma glucose levels are similar in Akita (276 ± 15 mg/dL) and Polg-Akita females (282 ± 20 mg/dL) through 9 mo of age. Likewise, food consumption is similar in Akita (3.78 ± 0.16 g/d) and Polg-Akita females (3.71 ± 0.27 g/d) through 9 mo of age. Together, these data show that, starting Fig. 1. Amelioration of diabetic profiles in Polg-Akita mice. Plasma glucose ≈3 mo of age, the PolgD257A mutation causes an age-dependent fi (A), food consumption (B), and body weight (C) in male wild-type (W), Polg and male-speci c amelioration of the Akita diabetic phenotype. (P), Akita (WA), and Polg-Akita (PA) mice. Gross weight of s.c. (SF) and visceral (VF) fat in 9-mo-old Akita (WA) and Polg-Akita (PA) male mice (D). All Kidney Function. At 3 mo of age, Polg-Akita males had small but data are represented as mean ± SEM. *P < 0.05, **P < 0.01, and ***P < 0.001 significant decreases (≈20%) in water consumption (Fig. 2A) and between Polg-Akita and Akita mice at each time point by the Student t test. urine output (Fig. 2B) compared with Akita mice, although both n ≥ 20, n ≥ 10, and n ≥ 7 at 3, 6, and 9 mo, respectively.

2of6 | www.pnas.org/cgi/doi/10.1073/pnas.1106344108 Fox et al. Downloaded by guest on September 24, 2021 Fig. 2. Renal function and histology in 9-mo-old Polg-Akita mice. Water consumption (A), urine output (B), urine glucose (C), and urinary albumin (D)inwild- type (W), Polg (P), Akita (WA), and Polg-Akita (PA) mice. Renal histology of diabetic Akita (E) and Polg-Akita (F) mice with proximal tubules (blue outline) and distal tubules (yellow outline). (G) Electron micrograph of Akita distal tubular cells with glycogen deposits (yellow arrowheads) within the nuclei. (H) Electron micrograph Polg-Akita proximal tubular cells with mitochondria and lamellated inclusions (blue arrowheads) in the cytoplasm. All data are represented as mean ± SEM. *P < 0.05 and ***P < 0.001 between Polg-Akita and Akita mice at each time point by the Student t test. n ≥ 6 at 3, 6, and 9 mo.

Plasma Insulin and Insulin Sensitivity. Insulin mRNA amounts in signiﬁcantly change the already impaired β-cell function of the the pancreas of both Akita and Polg-Akita mice were similar, at Akita mice, and that the impaired glucose handling exhibited by 20–30% of the levels in nondiabetic wild-type mice (Fig. 3A). the Akita mice was likewise not changed by the Polg mutation.

Plasma insulin levels in the Akita and Polg-Akita mice were also MEDICAL SCIENCES ≈20% of those in nondiabetic mice. Plasma insulin levels in the Small Intestine Function. In agreement with a previous report (11), D257A Polg-Akita mice were slightly higher than those in the Akita we find that the Polg mutation causes an increase in apo- littermates, but the difference was not significant (Fig. 3B). No ptosis within the villi of the small intestine of the Polg-Akita mice differences in the morphology and numbers of pancreatic islet relative to Akita mice. Additionally, we find an increase in ap- cells were found between the Akita (Fig. 3C) and Polg-Akita optosis and a decrease in cell proliferation in the small intestine (Fig. 3D) mice. At 9 mo, the Polg-Akita males exhibited impaired crypts of the Polg-Akita mice relative to Akita mice (Fig. S4). However, these profiles were found at similar levels in non- clearance of oral glucose loads (Fig. 3E). However, plasma insulin D257A levels after the oral glucose challenge did not differ significantly diabetic Polg mice, indicating that these Polg effects were in the Akita and Polg-Akita mice (Fig. 3F). Taken together, these not altered by diabetes. In agreement with this result, the gross data demonstrate that adding the PolgD257A mutation did not morphology of the small intestine remained unchanged in 9-mo- old Polg-Akita mice relative to Akita mice. The mRNA levels of intestinal transporters that affect glucose uptake were also similar in the Polg-Akita and Akita mice (Table S2). Pancreatic expression of mRNAs for amylase, trypsinogen, chymotrypsino- gen, and lipase in 9-mo-old Polg-Akita mice was not significantly affected by the PolgD257A mutation (Fig. S4). Taken together, these observations show that the Polg-Akita mice at 9 mo of age display no significant indications of digestive problems or nutrient malabsorption that might account for the amelioration of their diabetic hyperphagia and hyperglycemia.

Appetite Control. Prompted by the decrease in food consumption that developed in the Polg-Akita mice as they aged, we measured two known regulators of appetite: leptin (a suppressor of appetite) and ghrelin (a stimulator of appetite). At 9 mo of age, the plasma leptin levels in Polg-Akita mice were significantly higher (Fig. 4A), whereas circulating ghrelin levels were significantly lower, relative to Akita littermates (Fig. 4B). Expression in the hypothalamus of the anorexic gene, Pomc (coding for proopiomelanocortin), and the orexic gene, Npy (coding for Neuropep- tide Y), are critical for controlling appetite (20), and we found that hypothalamic expression of the orexic Npy gene was significantly increased in Akita diabetic mice relative to nondiabetic Fig. 3. Insulin levels in Polg-Akita and Akita mice. Insulin mRNA levels (A) wild-type mice (Fig. 4C), whereas expression of the anorexic and plasma insulin (B) in wild-type (W), Akita (WA), and Polg-Akita (PA) mice Pomc gene was significantly decreased (Fig. 4D). However, in the at 6 and 9 mo of age. All data are represented as mean ± SEM Pancreas Polg-Akita mice, these diabetes-associated shifts in Npy and (H&E) from 9-mo-old Akita (C) and Polg-Akita (D) mice. Glucose tolerance fi test (GTT) (E) and plasma insulin (F) during GTT at 9 mo of age in W, WA, and Pomc expression were reversed, returning to levels not signi - PA. **P < 0.01 between Polg-Akita and Akita, ***P < 0.001 compared with cantly different from those in nondiabetic wild-type mice. Thus, D257A wild type by the Student t test. adding the Polg mutation to the Akita mutation increases

Fox et al. PNAS Early Edition | 3of6 Downloaded by guest on September 24, 2021 Plasma testosterone in wild-type mice at 9 mo of age was 220 ± 12 ng/dL. Plasma testosterone in the Akita mice was not de- monstrably different from this value at 6 mo age, but had decreased to ≈50 ng/dL at 9 mo of age (Fig. 5F). The Polg-Akita mice had signiﬁcantly lower plasma testosterone levels than the Akita mice at both ages (≈30%; P < 0.01). Together, these data demonstrate that the PolgD257A mutation has a strong and direct negative impact on testosterone levels and testicular function in the Polg-Akita mice.

Testosterone Replacement. To test whether decreased testosterone levels are responsible for the reduced appetite and hyperglycemia, we implanted pellets into 6-mo-old Polg-Akita mice (n =3) that deliver the physiological dose of testosterone (0.14 mg/d) commonly used in replacement therapy in mice (21). Within 15 d Fig. 4. Appetite. Plasma leptin (A), plasma ghrelin (B), and hypothalamic after implantation of the testosterone pellet, plasma glucose in gene expression of Neuropeptide Y (C) and Proopiomelanocortin (D) in wild- the Polg-Akita mice increased from 300 mg/dL to 500 mg/dL, type (W), Akita (WA), and Polg-Akita (PA) mice at 9 mo of age. All data are indistinguishable from the level in Akita mice without testos- represented as mean ± SEM. *P < 0.05 and ***P < 0.001 indicate significant terone supplementation (Fig. 6A). Their food consumption also difference by the Student t test. increased compared with untreated Polg-Akita mice and were essentially the same level as in untreated Akita mice (Fig. 6B). Taken together, these data show that testosterone is a key com- plasma leptin, decreases plasma ghrelin, decreases hypothalamic ponent that directly regulates food consumption and, thereby, expression of Npy, and increases hypothalamic expression of influences the hyperglycemia of Akita male mice. Surprisingly, Pomc. It is likely that these changes in modulators of food intake, however, despite the return of hyperglycemia in the Polg-Akita all known to suppress appetite, result in the decreased appetite mice treated with testosterone, their daily urine output, glucose we observed in Polg-Akita mice relative to Akita mice. excretion, and urine albumin excretion were not increased (Fig. 6 C–E). This result indicates that the testosterone-treated Polg- Testicular Function. The testis is severely affected by the PolgD257A Akita mice are reabsorbing substantial amounts of glucose and mutation (11), but not by diabetes. Thus, light microscopy suggested to us that the kidneys of these animals were likely to be showed that the testes of our 9-mo-old Akita mice (Fig. 5C) were metabolically stressed. In support of this idea, we found that the not notably different from those of nondiabetic wild-type mice ultrastructure of the proximal tubules of the Polg-Akita mice (Fig. 5A). In contrast, the seminiferous tubules of the Polg (Fig. treated with testosterone for 3 mo showed considerably more 5B) and Polg-Akita (Fig. 5D) mice were severely atrophic and lamellated inclusions (Fig. S3) than were present in the untreated contained vacuoles. Marked loss of germ cells and spermato- Polg-Akita mice of the same age (Fig. 2H). genesis was evident within the seminiferous tubules of both Polg and Polg-Akita mice. Light microscopy showed a significant in- Discussion crease in the number of Leydig cells per cluster in the Polg mice Because of the importance of mitochondria in metabolism, we relative to WT and in the Polg-Akita mice relative to Akita, al- expected that the PolgD257A mutation, which leads to an accu- though the increase was less in the diabetic mice (Fig. 5E). Ul- trastructurally, Leydig cells of the testis in Polg-Akita mice have a markedly degenerative morphology compared with Akita mice, including the presence of lipofuscin indicative of mitochondrial damage (Fig. S5).

Fig. 6. Testosterone replacement in Polg-Akita mice. (A) Plasma glucose Fig. 5. Diminished testis function in Polg-Akita mice. Comparable histology levels in testosterone-implanted Akita (WA+T) and Polg-Akita (PA+T) mice sections of the testis of 9-mo-old wild-type (W; A) Polg (P; B), Akita (WA; C), and Control Akita (WA or WA-T) and Polg-Akita (PA or PA-T). Food con- and Polg-Akita (PA; D) mice. Total number of cells per Leydig cluster (E) sumption (B), urine volume (C), urine albumin (D), and urine glucose (E)in and plasma testosterone (F). All data are represented as mean ± SEM. *P < WA, PA, and PA-T at 22 d after implant. All data are represented as mean ± 0.05, **P < 0.01, and ***P < 0.001 indicate signiﬁcant difference by the SEM. (Scale bar: 100 μm.) *P < 0.05, **P < 0.01, and ***P < 0.001 indicate Student t test. signiﬁcant difference by the Student t test.

4of6 | www.pnas.org/cgi/doi/10.1073/pnas.1106344108 Fox et al. Downloaded by guest on September 24, 2021 mulation of random mtDNA mutations, would exacerbate the ducing the food intake of Akita mice also reduces their hyper- diabetes exhibited by male Akita mice and would increase the glycemia (20). severity of their diabetic complications. Surprisingly, we found Testosterone increases appetite in castrated rats and appears to the opposite result: a dramatic improvement of diabetic symp- do so independently of its conversion to estrogen (32). Whether toms (hyperglycemia, food consumption, and water consump- testosterone acts on adipocytes to regulate production of leptin, tion) in the Polg-Akita male mice as they aged. Accompanying circulating leptin, or directly on the genes in the hypothalamus these changes, we found alterations in the expression of anorexic remains to be elucidated. In addition, because plasma testoster- (Pomc) and orexic (Npy) genes in the hypothalamus, indicating one levels also declined in aged diabetic Akita mice without ap- that the Polg mutation had caused appetite suppression. We did parent loss of appetite, other factors must be also involved. For not find any improvement in the overall function of either the example, a significant reduction in circulating ghrelin works in pancreas or small intestine that would be sufficient to alter the unison with the testosterone-leptin mechanism to reduce appe- diabetes so markedly. However, there was a dramatic aging-as- tite. We also note that this aging-related amelioration of diabetes sociated increase in testicular damage in the Polg-Akita mice caused by the PolgD257A mutation could be exaggerated in the relative to the Akita mice, including damage to the Leydig cells, Akita model of diabetes in which β-cell destruction is directly tied and a reduction in plasma testosterone. Testosterone replace- to insulin production. Whether it is applicable to other animal ment increased food consumption in the double mutants, and models of diabetes remains to be determined. their hyperglycemia returned, confirming the influence of tes- The nephropathy in our experimental diabetic Akita mice was tosterone and appetite in the development of the diabetic symp- mild, in part because of their C57BL/6J genetic background, toms of Akita male mice. which confers resistance to diabetic nephropathy (33, 34). Con- In nondiabetic animals, an increase in plasma glucose after cordant with the dramatic improvement of hyperglycemia and a meal stimulates the release of insulin. Insulin binds to insulin reduced urine output, the Polg-Akita mice no longer had gly- receptors in peripheral tissues and triggers glucose uptake via cogen deposits in the nuclei of their distal tubular cells. Instead, insulin-sensitive glucose transporters. Insulin, together with the despite the lessening of their hyperglycemia, the Polg-Akita mice β amylin released by the -cells simultaneously with insulin (22), now had abnormalities in their proximal tubular cells. Formation plays an important role in the regulation of appetite by sup- of enlarged lysosomes filled with lamellated inclusions and cal- pressing orexic genes within the hypothalamus (23, 24). In Akita fi β ci cations indicate that the increase in mtDNA mutations in- mice, misfolded insulin protein causes ER stress and -cell duced by the PolgD257A mutation leads to increased mitochondrial death, resulting in reduced insulin production (13). Insufficient

destruction in the proximal tubular cells that have high-energy MEDICAL SCIENCES insulin production initiates two pathways that accelerate hyper- requirements and are highly packed with mitochondria. We note A glycemia in Akita mice (Fig. S6 ). First, it reduces the ability of that diabetes is necessary to cause the observed abnormalities in peripheral tissues to take up glucose. Second, the reduced insulin the proximal tubular epithelial cells, because they were not impairs appetite suppression in the hypothalamus causing hy- present either in nondiabetic Polg males or in mildly diabetic perphagia. The increased food intake induces an increased de- Polg-Akita females of the same age. Our observation that the mand for insulin, which increases the problems of the already urine volumes and glucose excretion of testosterone-treated Polg- stressed insulin-producing β-cells leading to further β-cell death. Akita mice remained low, despite a return of hyperglycemia, This damaging feedback can be partly interrupted by reducing fi further emphasizes this interaction. Thus, the treated mice are food intake, as illustrated by the well-documented bene ts of fi fi reduced food intake in the management of type 1 diabetes before reabsorbing suf cient amounts of ltered glucose to avoid os- the discovery of insulin (25). motic diuresis and glycosuria even with plasma glucose levels exceeding 500 mg/dL, which requires a large amount of energy. In the present study, we have demonstrated that the mtDNA fi mutator, Polg-D257A, can break the self-induced cycle of hy- Accompanying this high energy demand, we nd an accumulation perglycemia and hyperphagia of Akita male mice as they age of laminated bodies in the proximal tubules of the testosterone- (Fig. S6B). The most striking age-dependent damage in the Polg- treated hyperglycemic Polg-Akita mice exceeding that in the Akita males is in the testis, accompanied with a decrease in euglycemic-untreated Polg-Akita mice. In conclusion, we have demonstrated that the mtDNA editing circulating testosterone. Recent studies have suggested that ag- D257A ing-related reduction of food consumption may in part result mutation, Polg , causes an age-dependent reduction of the from reduced testosterone and a consequent increase in leptin diabetic phenotype of Akita male mice. We have shown that this production (26). Epidemiological studies have also demon- reduction is, at least in part, due to appetite suppression caused strated an association between these factors (27). In addition, by a premature decline in testosterone production associated testosterone administration has been shown to suppress the el- with increased damage to Leydig cells. Because female Akita evated leptin levels in older hypogonadal men (28, 29), and and Polg-Akita mice are both only moderately hyperglycemic dihydrotestosterone suppresses both leptin mRNA levels and and their food consumption remains similar throughout 9 mo of leptin secretions in adipocytes in vitro (30). Leptin, produced in age, the Polg-D257A does not affect the severity of their di- adipose tissues, is crucial in the regulation of appetite by sig- abetes. The accumulation of mitochondrial debris in the proxi- naling through the hypothalamus (31). Our observations in aged mal tubules of the Polg-Akita males, even when euglycemic, Polg-Akita males show the involvement of these factors in the demonstrates clearly that increased mutations in mtDNA exac- changes in food consumption and appetite induced by the Polg erbate the kidney pathology in untreated type 1 diabetes. mutation. Thus, as they age, Polg-Akita mice develop significant increases in plasma leptin, increased Pomc, and decreased Npy Methods gene expression compared with Akita males. Furthermore, both Mice. All mice were maintained on normal chow containing 5.3% fat and hyperglycemia and hyperphagia returned in Polg-Akita males 0.019% cholesterol (Prolab Isopro RMH 3000, ref 5P76; Agway). The exper- after testosterone administration. We conclude that the age- imental animals were Akita heterozygous males homozygous for the PolgD257A mutation. Controls were male Akita littermates that were either dependent decline of testis function and testosterone production +/+ D257A wild type at the Polg locus (Polg Akita) or heterozygous for the mutation caused by Polg underlies the reduced diabetic phenotypes (PolgD257A/+Akita). Heterozygous PolgD257A mutants are phenotypically in- of the Polg-Akita mice. This interpretation is supported by a re- distinguishable from wild-type mice (11). Both male and female animals cent report that gonadectomy causes a marked reduction of were monitored for diabetes and characterized at 3, 6, and 9 mo of age. The hyperglycemia in diabetic Akita males and normalizes Pomc breeding process to generate doubly mutant and control mice is described and Npy mRNA levels within their hypothalamus, and that re- in Fig. S7. All animal experiments were performed in accordance with the

Fox et al. PNAS Early Edition | 5of6 Downloaded by guest on September 24, 2021 Institutional Animal Care and Use Committee at the University of North for glucose measurement at time points of 0, 15, 30, 60, and 120 min Carolina at Chapel Hill. after gavage.

Plasma Analyses. Animals were fasted in the morning 4 h before the collection Urine Analyses. Mice were housed in metabolic cages, allowed to acclimate to of blood. Plasma glucose was measured by using a colorimetric kit (439– the new environment for 24 h, and urine samples were collected over 90901; Wako Chemical). Plasma levels of leptin, insulin, and ghrelin were a second period of 24 h. Body weight, food, and water were measured before measured by using ELISA kits (90030 and 90080; Crystal Chem and EZRGRT- and at the end of the second period of 24 h. Urinary Albumin and creatinine 91K; LINCO Research). Testosterone levels were measured in plasma pooled were measured by using kits (Albuwell, 1011 and Creatinine Companion, from 3 mice per sample with RIA kits (TKTT2; Siemens Medical Solutions 1012; Exocell). Urinary levels of Na, K, and Cl were measured by using an Diagnostics) by the Reproductive Biology Core at the University of Virginia. Automatic Chemical Analyzer (VT250; Johnson & Johnson). Mitochondrial DNA Mutations. Mitochondria was isolated from 200 mg of liver Histology. Tissues were fixed in 4% paraformaldehyde (PFA). Paraffin sections samples pooled from 3 to 5 mice of each genotype by differential centri- μ fugation in a sucrose gradient as described (35). DNA was purified by using at 5 m were stained with Hematoxylin and Eosin, Periodic acid-Schiff, or ’ standard isolation protocols and fragmented by Covaris S220 adaptive fo- Masson s Trichrome. For transmission electron microscopy (JOEL USA), small cused acoustic apparatus into 300 bp in length, and subjected to high tissue fragments were postfixed in 2% osmium tetroxide and 1-μm-thick throughput sequencing of 5,000–27,000 reads per base by using Illumina sections were stained with uranyl acetate followed by lead citrate. GAII. DNA sequence was processed through the bandled Solexa image extraction pipeline and aligned with ELAND software against murine mito- Testosterone Replacement. Six-month-old Polg-Akita and Akita mice were chondrial genome sequence (NCBL Build 36) as a reference. Default quality implanted with 12.5 mg/90-day testosterone time-release pellet (Innovative score was adjusted to allow 2 mismatches per 36 bp reads. The uniquely Research America). This dose was shown to normalize plasma testosterone mapped reads were used for further analyses. levels in orchidectomized mice (21).

fi Gene Expression. Total RNA was puri ed from hypothalamus and pancreas Data Analysis. Values are reported as mean ± SEM. Statistical analyses were – samples by using TRIzol (15596 026; Invitrogen) as described (36). Total conducted with JMP 6.0.2 software (SAS Institute). P values <0.05 were RNA from other tissues was purified by using an Automated Nucleic Acid considered significant. Workstation ABI 6700. Real-time PCR was performed in an ABI PRISM 7700 Sequence Detector (Applied Biosystems). β-Actin mRNA was used for ACKNOWLEDGMENTS. We thank Drs. E. Wilson, K. Pandya, N. Takahashi, normalization. L. Johnson, R. Thresher, and S. Magness for advice and discussions, and Marcus McNair for skillful assistance. This work was supported by National Glucose Tolerance Tests. Male mice of each genotype were fasted for 4 h Institutes of Health Grants DK07613, HL087946, and DK34987. R.F. was sup- before oral gavage of glucose (2 g/kg of body weight). Plasma was collected ported by National Institutes of Health Grant T32-HL69768.

1. de Andrade PB, et al. (2006) Diabetes-associated mitochondrial DNA mutation 19. Blakely EL, et al. (2005) A mitochondrial cytochrome b mutation causing severe A3243G impairs cellular metabolic pathways necessary for beta cell function. respiratory chain enzyme deﬁciency in humans and yeast. FEBS J 272:3583–3592. Diabetologia 49:1816–1826. 20. Toyoshima M, et al. (2007) Dimorphic gene expression patterns of anorexigenic and 2. Patti ME, Corvera S (2010) The role of mitochondria in the pathogenesis of type 2 orexigenic peptides in hypothalamus account male and female hyperphagia in Akita diabetes. Endocr Rev 31:364–395. type 1 diabetic mice. Biochem Biophys Res Commun 352:703–708. 3. Simmons RA, Suponitsky-Kroyter I, Selak MA (2005) Progressive accumulation of 21. Nathan L, et al. (2001) Testosterone inhibits early atherogenesis by conversion to mitochondrial DNA mutations and decline in mitochondrial function lead to beta-cell estradiol: Critical role of aromatase. Proc Natl Acad Sci USA 98:3589–3593. failure. J Biol Chem 280:28785–28791. 22. Moore CX, Cooper GJ (1991) Co-secretion of amylin and insulin from cultured islet 4. Someya S, et al. (2008) The role of mtDNA mutations in the pathogenesis of age- beta-cells: Modulation by nutrient secretagogues, islet hormones and hypoglycemic related hearing loss in mice carrying a mutator DNA polymerase gamma. Neurobiol agents. Biochem Biophys Res Commun 179:1–9. – Aging 29:1080 1092. 23. Morris MJ, Nguyen T (2001) Does neuropeptide Y contribute to the anorectic action 5. Trifunovic A, Larsson NG (2008) Mitochondrial dysfunction as a cause of ageing. of amylin? Peptides 22:541–546. – J Intern Med 263:167 178. 24. Mayer CM, Belsham DD (2009) Insulin directly regulates NPY and AgRP gene 6. Wang J, Markesbery WR, Lovell MA (2006) Increased oxidative damage in nuclear and expression via the MAPK MEK/ERK signal transduction pathway in mHypoE-46 mitochondrial DNA in mild cognitive impairment. J Neurochem 96:825–832. hypothalamic neurons. Mol Cell Endocrinol 307:99–108. 7. Zhang D, et al. (2003) Mitochondrial DNA mutations activate the mitochondrial 25. Westman EC, Yancy WS, Jr., Humphreys M (2006) Dietary treatment of diabetes apoptotic pathway and cause dilated cardiomyopathy. Cardiovasc Res 57:147–157. mellitus in the pre-insulin era (1914-1922). Perspect Biol Med 49:77–83. 8. Kakoki M, et al. (2006) Senescence-associated phenotypes in Akita diabetic mice are 26. Morley JE (2001) Decreased food intake with aging. J Gerontol A Biol Sci Med Sci 56: enhanced by absence of bradykinin B2 receptors. J Clin Invest 116:1302–1309. 81–88. 9. Kakoki M, Takahashi N, Jennette JC, Smithies O (2004) Diabetic nephropathy is 27. Baumgartner RN, et al. (1999) Serum leptin in elderly people: Associations with sex markedly enhanced in mice lacking the bradykinin B2 receptor. Proc Natl Acad Sci hormones, insulin, and adipose tissue volumes. Obes Res 7:141–149. USA 101:13302–13305. 28. Sih R, et al. (1997) Testosterone replacement in older hypogonadal men: A 12-month 10. Hance N, Ekstrand MI, Trifunovic A (2005) Mitochondrial DNA polymerase gamma is randomized controlled trial. J Clin Endocrinol Metab 82:1661–1667. essential for mammalian embryogenesis. Hum Mol Genet 14:1775–1783. 29. Kapoor D, Clarke S, Stanworth R, Channer KS, Jones TH (2007) The effect of 11. Kujoth GC, et al. (2005) Mitochondrial DNA mutations, oxidative stress, and apoptosis testosterone replacement therapy on adipocytokines and C-reactive protein in in mammalian aging. Science 309:481–484. – 12. Trifunovic A, et al. (2004) Premature ageing in mice expressing defective mitochon- hypogonadal men with type 2 diabetes. Eur J Endocrinol 156:595 602. drial DNA polymerase. Nature 429:417–423. 30. Gupta V, et al. (2008) Effects of dihydrotestosterone on differentiation and pro- 13. Wang J, et al. (1999) A mutation in the insulin 2 gene induces diabetes with severe liferation of human mesenchymal stem cells and preadipocytes. Mol Cell Endocrinol – pancreatic beta-cell dysfunction in the Mody mouse. J Clin Invest 103:27–37. 296:32 40. 14. Vermulst M, et al. (2008) DNA deletions and clonal mutations drive premature aging 31. Friedman JM, Halaas JL (1998) Leptin and the regulation of body weight in mammals. – in mitochondrial mutator mice. Nat Genet 40:392–394. Nature 395:763 770. 15. Vermulst M, et al. (2007) Mitochondrial point mutations do not limit the natural 32. Gentry RT, Wade GN (1976) Androgenic control of food intake and body weight in lifespan of mice. Nat Genet 39:540–543. male rats. J Comp Physiol Psychol 90:18–25. 16. Yoshioka M, Kayo T, Ikeda T, Koizumi A (1997) A novel locus, Mody4, distal to 33. Gurley SB, et al. (2006) Impact of genetic background on nephropathy in diabetic D7Mit189 on chromosome 7 determines early-onset NIDDM in nonobese C57BL/6 mice. Am J Physiol Renal Physiol 290:F214–F222. (Akita) mutant mice. Diabetes 46:887–894. 34. Qi Z, et al. (2005) Characterization of susceptibility of inbred mouse strains to diabetic 17. Ekstrand MI, et al. (2004) Mitochondrial transcription factor A regulates mtDNA copy nephropathy. Diabetes 54:2628–2637. number in mammals. Hum Mol Genet 13:935–944. 35. Graham JM (2001) Isolation of mitochondria from tissues and cells by differential 18. Foury F (1989) Cloning and sequencing of the nuclear gene MIP1 encoding the centrifugation. Curr Protoc Cell Biol, Chapter 3:Unit 3.3. catalytic subunit of the yeast mitochondrial DNA polymerase. J Biol Chem 264: 36. Chomczynski P, Sacchi N (1987) Single-step method of RNA isolation by acid 20552–20560. guanidinium thiocyanate-phenol-chloroform extraction. Anal Biochem 162:156–159.

6of6 | www.pnas.org/cgi/doi/10.1073/pnas.1106344108 Fox et al. Downloaded by guest on September 24, 2021