Determinants of Whole-Body Protein Metabolism in Subjects With and Without Type 2 Diabetes

RESEARCH DESIGN AND METHODS

Subjects and diet

Caucasian adult subjects (88 [31 men and 57 women] without and 40 [15 men and 25 women] with type 2 diabetes) were recruited and screened by medical history, physical examination, and laboratory investigation (3,12). Although recruitment was done over a period of years, type 2 diabetic and nondiabetic subjects were studied concurrently, minimizing potential temporal confounding. They were admitted to the clinical investigation unit of the Royal Victoria Hospital for 7 days, after written informed consent. The study protocols were approved by the institutional human ethics review board. Obesity was defined as BMI >30 kg/m². They had no hypertension, hepatic, thyroid, cardiovascular, hematologic, renal, or pulmonary dysfunction. Smokers, individuals requiring insulin, or those with unusual protein intakes were excluded. Their only exercise was walking in or around the hospital. Three days before admission, oral antihyperglycemic agents were stopped. They received a weight-maintaining liquid formula diet (Ensure; Ross Laboratories, Montreal, Canada) based on their resting energy expenditure (REE), determined by the Harris-Benedict equation or indirect calorimetry and multiplied by a factor of 1.5–1.7 for physical activity. Protein intake was 15% of energy, at 0.9–1.3 g/kg body wt [1.73 ± 0.02 g/kg FFM/day]. Six equal meals per day were taken during the ¹⁵N-glycine protocol (3,4). Capillary glucose was measured six times daily in type 2 diabetic and three in other subjects (Accusoft Advantage; Roche Diagnostics, Laval, Canada). Daily weight and water intake were recorded. Daily 24-h urine collections were made, nitrogen balance was calculated on the last 3 days (3,4), and urine glucose was measured in type 2 diabetic subjects. Energy lost as glycosuria was added to the diet as 50% glucose polymer and 50% soy oil. Body composition was assessed by bioelectrical impedance analysis (RJL-101A Systems, Detroit, MI) (n = 99) using equations validated for nonobese (19) and obese (20) subjects. In others (n = 13 type 2 diabetic and 16 obese subjects without type 2 diabetes), lean body mass was estimated from daily urinary creatinine excretion (21). The two methods correlated with a slope of 0.83 and r² of 0.933 (n = 9). Waist circumferences were measured at the minimal value between the xiphoid process and the iliac crest and hips at the maximal protuberance in the trochanteric region.

¹⁵N-glycine kinetic studies were done on days 5–7, as detailed (4,5). Briefly, 99 atom percent excess ¹⁵N-glycine (Isotec; Sigma-Aldrich, St. Louis, MO) was given at a rate of 0.5 mg ¹⁵N · kg body wt⁻¹ · 24h⁻¹ in 5 ml water every 3 h for 60 h to achieve steady-state isotopic enrichment in urine urea. Urine was collected every 3 h. Urea nitrogen was isolated, after removal of ammonia and other nitrogenous compounds with an ion-exchange resin, using urease (U 1500; Sigma-Aldrich) and the Conway procedure (22). Urea ¹⁵N enrichment was measured with a dual-inlet, double-collector isotope ratio mass spectrometer (Vacuum Generators, Micromass 602D; Winsford, U.K.) or by mass isotopomer distribution analysis of hexamethylenetetramine (formed by reacting ammonium released from urea with formaldehyde) measured by gas chromatography/mass spectrometry. This method, adapted from Yang et al. (23), permits the measurement of low isotopic enrichment of ¹⁵N in urine byproducts, through amplification on a secondary molecule. Correlation of ¹⁵N enrichment between the two methods was excellent with a slope of 0.93 and r² of 0.97. Enrichment curves were drawn for each subject, with pretest background subtracted. The method requires that urea ¹⁵N enrichment reaches an isotopic steady state within 60 h, defined as the first plateau with at least four consecutive points (12 h) with a slope not different from zero and before ¹⁵N recycling (24). The rate of entry or exit of nitrogen (flux) into the metabolic nitrogen pool was calculated from the mean steady-state value, assuming that the fraction of the administered isotope excreted as urinary ¹⁵N urea is the same as the fraction of total amino nitrogen entering the metabolic pool excreted as urinary urea nitrogen (24). We used urinary ¹⁵N urea enrichment because, unlike ammonia as end product, urea does not underestimate the synthesis rate (22). Nitrogen intake was known from the controlled diet, and total urinary nitrogen was measured. Thus, protein synthesis and breakdown were calculated from the Picou Taylor-Roberts equation as follows (25): Q = I + B = S + E, where Q is nitrogen flux; I, nitrogen intake, B, protein breakdown; S, protein synthesis; and E, total urinary nitrogen excretion.

Analytical methods

Postabsorptive venous blood samples were drawn on days 1 and 3 of the ¹⁵N-glycine study and processed as detailed (3–6). Glucose, urea, lipid, electrolyte, and A1C determinations were made in the hospital's clinical laboratory. Free fatty acid concentrations were measured using the NEFA C test kit (Wako Chemicals, Richmond, VA), immunoreactive insulin by a single antibody, and charcoal precipitation radioimmunoassay using human standards and labeled hormone from Linco Research (St. Charles, MO), with no cross-reactivity with proinsulin. The homeostasis model assessment of insulin resistance (HOMA-IR) was calculated from fasting glucose and insulin concentrations (26). REE was measured by indirect calorimetry (Deltratrac; Sensor Medics, Yorba Linda, CA).

Statistical methods

Results are presented as means ± SE. Subject characteristics were analyzed by ANOVA, with the Bonferroni adjustment for multiple comparisons. The effects of diabetes and obesity were assessed by ANCOVA, with FFM and protein intake as covariates when found to have a significant predictive value from prior regression analysis (27). HOMA-IR values were log transformed to yield a normal distribution for analyses. Pearson's coefficient was used for all correlations, and when they required controlling for other variables, partial correlation was performed. Stepwise linear regression analysis was conducted with individual predictors of flux, synthesis, breakdown, and net balance found by simple correlations. The analyses were performed with SPSS 11.0 for Windows (SPSS, Chicago, IL). Significance was at P < 0.05.

RESULTS

Subject data are presented in Table 1. Diabetes duration was 4.7 ± 0.7 years. There were more women in the two groups of obese subjects. Mean age differed among groups, but each group included younger and older subjects. Weight, BMI, FFM, and percent body fat were greater in obese subjects but did not differ between obese groups. Body circumferences were greater in obese subjects, with no sex difference. Waist-to-hip ratio was greater in men than women and, for both sexes, greatest in type 2 diabetic patients. REE was lowest in nonobese and highest in type 2 diabetic subjects. Protein intake did not differ per kilogram FFM. Daily nitrogen balance was less in type 2 diabetic subjects and negative (P = 0.013 vs. 0). Fasting triacylglycerol, glycemia, A1C, free fatty acids, and daily mean capillary glucoses were elevated in type 2 diabetes. Fasting plasma insulin was lowest in nonobese and highest in type 2 diabetic subjects. HOMA-IR and lnHOMA-IR were highest in type 2 diabetic subjects.

Whole-body protein kinetics are shown in Fig. 1A–C as means adjusted for FFM and protein intake as covariates for flux and FFM as a covariate for synthesis and breakdown. Within each group, there were no sex effects, and an age effect was seen in type 2 diabetic subjects only. There was a diabetes effect on flux (Fig. 1A), synthesis, and breakdown (Fig. 1B), which were significantly greater, and on net balance (Fig. 1C), which was less in type 2 diabetic subjects than in the other groups. The diabetes effect on all variables remained significant when BMI was also used as a covariate to control for adiposity. There was an obesity effect on flux, synthesis, and breakdown, with rates greater in obese subjects without diabetes than in nonobese subjects; however, net balance did not differ. Flux, synthesis, and breakdown correlated positively with indexes of body fatness and circumferences, and REE, fasting triacylglycerol, glucose and insulin, lnHOMA-IR, and A1C (for flux and breakdown) (r = 0.236–0.431, all P < 0.021) controlled for FFM and protein intake. Net balance correlated negatively with triacylglycerol (r = −0.282, P = 0.002), glucose (r = −0.412, P < 0.001), A1C (r = −0.413, P < 0.001), and lnHOMA-IR (r = −0.255, P = 0.004).

In type 2 diabetic subjects, lnHOMA-IR and insulin, but not glucose, correlated with waist circumference and BMI. Flux, synthesis, and breakdown correlated negatively with age (r = −0.433 to −0.513, all P ≤ 0.006). In women with type 2 diabetes, but not in men, net balance correlated positively with hip circumference (Fig. 2A). In men with type 2 diabetes, but not in women, net balance correlated negatively with fasting glycemia (Fig. 2B).

Stepwise linear regression models were as follows (all P < 0.001). In type 2 diabetic subjects, waist-to-hip ratio and lnHOMA-IR explained 40% of the variance in flux and 43% of that in synthesis; in all the subjects without diabetes, FFM and lnHOMA-IR explained 62% of the variance in flux. In all subjects, fasting glucose was a negative predictor of 16% of the variance in net balance.

CONCLUSIONS

Our approach to whole-body integrated fed-fasted protein metabolism was deployed in a large sample of adults varying in body composition and especially in insulin sensitivity of glucose metabolism. The strategy was to safely seek, as marked, a diabetes effect as feasible in a short-term, closely supervised inpatient protocol. The subjects were nonetheless representative of community-dwelling, suboptimally controlled type 2 diabetes. They demonstrated significantly higher values of all components of whole-body kinetics of protein metabolism, with a lesser net balance. The obese subjects without diabetes had flux, synthesis, and breakdown rates intermediate between nonobese and type 2 diabetic subjects. The type 2 diabetic subjects had greater insulin resistance of glucose than the obese subjects, by HOMA-IR. These diabetes effects on whole-body protein metabolism were independent of age, sex, FFM, and body fatness.

Our diet maintained weight and provided sufficient protein to ensure nitrogen equilibrium and thereby preserve lean tissue. However, in the hyperglycemic type 2 diabetic subjects, nitrogen balance was worse and net endogenous balance significantly less than in nonobese subjects by an average of 0.89 g N (5.6 g protein, 28 g lean tissue) per day. These data suggest that protein needs in type 2 diabetes may be increased with poor glycemic control. Type 2 diabetes thus mirrors chronic disease states associated with elevated protein turnover and rapid loss of body protein (28). These type 2 diabetes alterations can be corrected by restoration of normoglycemia with intensive insulin therapy (5,6) or with near-normal glycemia using oral antihyperglycemic agents and sufficient protein intake even with moderate energy restriction (3).

All protein kinetic parameters correlated positively and net balance negatively with fasting glucose, lnHOMA-IR, and A1C. The correlations with net balance remained significant even with nonobese data excluded from the analysis. Also, lnHOMA-IR was a significant predictor of the variance in flux and fasting glucose of net balance. This suggests a worsening of protein metabolism that is concurrent with that of glucose and, since A1C reflects metabolic state before the study, that it is possibly affected by the duration of hyperglycemia. We have reported such relationships between glucose and protein metabolism in obese women (14) and type 2 diabetic men (13), using the hyperinsulinemic, euglycemic, isoaminoacidemic clamp to simultaneously measure insulin resistance of both. Rates of glucose disposal correlated with the protein anabolic response to hyperinsulinemia (14). We had predicted a dose-response relationship between glycemic control and the derangement of protein metabolism (3), with a glucose threshold above which abnormal kinetics appear. In type 2 diabetic women as a group, no significant correlation between protein kinetics and glycemia was found, but in men net balance decreased significantly as hyperglycemia worsened (Fig. 2B). In obese men without diabetes, the lower boundary of the 95% CI for net balance (mean 2.2 g N/day), was 1.4 g N/day, a value that crosses the regression line using data from type 2 diabetic men (Fig. 2B) at a fasting glucose level of 8.2 mmol/l and crosses the regression line using data from all men (r = −0.555, P < 0.001) at 7.6 mmol/l. The same calculations in women, in whom the lower boundary for net balance was 1.9 g N/day, crosses the regression line (r = −0.354, P < 0.001) at a fasting glucose of 6.7 mmol/l. These concentrations may reflect glycemic thresholds above which protein anabolism is blunted.

HOMA-IR scores reflect the postabsorptive state, whereas measurements of protein kinetics and net balance are integrated over the day; still, they were correlated. These findings indicate that the magnitude of derangements in the metabolism of protein relates to that of insulin resistance of glucose. We consider lnHOMA-IR to be an appropriate surrogate for insulin resistance of glucose (26), since three subjects only had impaired fasting glucose, two subjects with type 2 diabetes had low fasting insulin levels, and none were on insulin therapy, factors claimed to limit its validity (29). Indeed, lnHOMA-IR correlated with lnM (glucose uptake/[insulin]) (r = −0.850, P < 0.001, n = 65) in our subjects in whom clamps were performed. The correlation was significant in the nonobese (n = 19), obese (n = 29), and type 2 diabetic (n = 17) groups (r = −0.508, −0.740, and −0.594, respectively; P ≤ 0.012).

We tested whether protein metabolism would worsen as waist circumference increased in type 2 diabetes. Although it correlated with flux, synthesis, and breakdown with all subjects included, waist circumference did not do so within the type 2 diabetic group. However, our type 2 diabetic women with larger hips had greater net balance. This may indicate that lower-body fat has a “protective” protein-sparing effect. There is evidence that abdominal obesity may be a marker of “dysfunctional adipose tissue” characterized by a relative inability for subcutaneous adipose tissue to store excess energy, which then accumulates at undesired sites (30). In our viscerally obese type 2 diabetic women, having larger hips may explain why we did not discern an effect of the magnitude of hyperglycemia. Women with upper-body obesity have been found to have blunted insulin-mediated decrease in protein breakdown compared with women with lower body obesity (31).

The ¹⁵N-glycine method used cannot differentiate between the effects of insulin and amino acids on protein metabolism or elucidate the mechanisms underlying the results. Amino acids have protein anabolic effects in whole-body studies in healthy subjects (32) and share with insulin the intracellular mammalian target of rapamycin (mTOR) pathway to stimulate skeletal muscle protein synthesis (33). Assessment of whole-body insulin sensitivity in the fed part of the feeding-fasting cycle requires a separate technique for quantifying it.

All parameters of protein kinetics were highly correlated with REE, and REE was a predictive factor of flux and synthesis in type 2 diabetes (data not shown). This is not surprising, as the energy requirement of protein turnover may account for ∼20% of REE (34). We have reported greater REE in hyperglycemic type 2 diabetic subjects, with glycemia as an independent predictor of REE (35). Others have found the same with poor diabetes control and attributed it to greater protein turnover (8).

Mean age differed among groups, but each group was composed of younger and older subjects, which may explain the absence of an age effect on protein kinetics. Age did not correlate with kinetic parameters, controlling for FFM. Others have found a significant linear decline in postabsorptive whole-body protein turnover by advancing decades but no significant difference when comparing groups of middle-aged to younger or older subjects (27). However, within the type 2 diabetic group, age was negatively correlated with flux, synthesis, and breakdown and was a significant predictor of flux in men. We have reported a decrease in flux with age (18). This effect was attributed mostly to changes in body composition with aging. In the present study, there were trends of FFM to decrease with age in the type 2 diabetic group and for the older men with type 2 diabetes to be leaner.

In conclusion, protein flux, synthesis, and breakdown are elevated, while net balance is diminished in type 2 diabetes. While in men with type 2 diabetes fasting glycemia relates to net balance, this is not the case in women, in whom lower body fat appears to be protective. We have also shown that abnormalities in protein metabolism and insulin resistance of glucose coexist in obesity (14) and that insulin sensitivity of protein metabolism is blunted in type 2 diabetes (13). For these reasons, there is need to define protein requirements for type 2 diabetes in relation to concurrent obesity, weight loss, and level of glycemic (metabolic) control.