In the present investigation, we sought to determine if the rate of endogenous glucose production declines in response to an abrupt increase in energy expenditure and to ascertain if dietary manipulation- provision of additional dietary protein with concomitant negative energy balance-would attenuate this effect. The primary observations were that a 1000 kcal·d-1 exercise-induced energy deficit caused a decline in GP attributable to a decline in GB, but provision of additional dietary protein mitigated the effects of a sudden increase in exercise on GP. We observed no difference within and between groups in terms of resting RER and oxidation rates. However, we did observe an increase in β-Hydroxybutyrate, and decrease in glucose and insulin, in response to the unaccustomed increase in energy expenditure regardless of dietary protein or energy intake.
As expected, RERs at baseline for all groups were approximately 0.75, indicating that fat was the main substrate oxidized at rest (i.e., ~84% of energy at rest was derived from fat). This is consistent with other reports in aerobically-trained subjects . Lack of within or between group changes over time in RER and oxidation rates was not surprising, given that subjects were already predominantly oxidizing fat at rest prior to the energy deficit phase. It does contrast, however, with the observation that endogenous carbohydrate oxidation is reduced at rest during periods of energy deficit .
The fact that volunteers receiving a higher protein diet were able to maintain GP during the period of increased energy expenditure and negative energy balance suggests that manipulation of macronutrient intake partially offsets the energy deficit-associated fall in glucose production when an increase in exercise to create an energy deficit is undertaken for weight loss or when negative energy balance is inevitable (for example, in the case of military operations or prolonged exercise). The most likely mechanism to explain this effect relates to our observation that maintenance of GB seems to have attenuated the decline in GP associated with the exercise-induced energy deficit. Since the current study did not have adequate power to detect a difference in GB between DEF-HP and DEF-MP using 2-way ANOVA, more rigorous studies are needed to further investigate this speculation. Future studies, aiming to detect a 0.4 g·kg-1·min-1 difference in GB between groups (standard deviation 0.2 g·kg-1·min-1; α = 0.05; power = 0.80), should test at least eight volunteers per group.
The ability of DEF-HP to maintain GP may reflect the increased availability of gluconeogenic substrates. We detected no between-group differences in fgng and free fatty acids, during the period of insufficient energy intake. However, β-Hydroxybutyrate, which stimulates gluconeogenesis, increased in response to the increased exercise, particularly in the two energy deficit groups. More importantly, the increase over time in β-Hydroxybutyrate for DEF-HP was significantly higher compared to BAL-MP and DEF-MP. Increased reliance on alanine-derived gluconeogenesis is another possibility, since the glucose-alanine cycle is up-regulated during exercise and the rate of glucose production from alanine is even higher during periods of starvation , although we have no data to support this hypothesis. Lastly, an increase in the recycling of glucose via fat stores may occur, since periods of starvation stimulate the mobilization of fat  and total glucose production from glycerol may be ≥ 20% during starvation . Our results, however, do not support this mechanism, since we observed no differences in fatty acid availability in response to the exercise-induced energy deficit. Further, the lypolytic response appears to differ between an exercise-induced energy deficit and a diet-induced energy deficit.
The observation that GP declined in response to an exercise-induced energy deficit is consistent with studies investigating negative energy balance due to restricted energy intake. Indeed, a decline in glucose production has been observed by Hellerstein et al.  in response to prolonged starvation (60 hours), likely due to reduced availability of alanine and lactate to the liver . This finding is also consistent with our observation that insulin levels declined in response to the exercise-induced energy deficit. We also observed maintenance of GP and insulin levels when energy balance was preserved despite increased exercise and total energy expenditure. This was also predictable, given that endogenous glucose production is maintained during periods of energy balance .
The observation that the high protein group was capable of preserving GP is suggestive that dietary protein supplementation may not only help with preservation of lean mass during periods of energy restriction [8–10], but may also have utility for preserving metabolic flexibility. These findings may also have implications for overweight individuals who purposely induce an energy deficit via an increase in exercise to achieve weight loss; however, future research is needed to confirm this hypothesis.
Regardless of energy or protein intake, fgng appears to have increased while GB appears to have decreased in response to an increase in energy expenditure (Table 3) although analysis of the confidence intervals is not entirely supportive (Figure 2). The pattern for increased fgng observed in this study is consistent with Bergman et al. , who reported an increase in gluconeogenesis in response to nine weeks of unaccustomed endurance training. Our findings are not surprising, given that glycogen stores likely decline over time in response to the demands of an unaccustomed increase in exercise thus placing greater reliance on fgng to maintain GP. This hepatic auto-regulation has been demonstrated consistently in the literature, for example, when glycogen stores are reduced in response to prolonged starvation, fgng increases in an attempt to maintain GP . Further, Staehr et al.  reported that when galactose was given to stimulate GB, gluconeogenesis declined in a compensatory manner and GP was maintained. The current study provides further evidence of hepatic auto-regulation in terms of glucose metabolism.
Although it is plausible that depleted glycogen stores may have been the primary factor in the apparent reduction in GB, lower plasma glucagon concentrations may also have played a role. Indeed, the DEF-MP group experienced a significant decrease in fasting glucagon in response to increased energy expenditure. This result is consistent with the findings of Tremblay et al.  and Oppert et al.  who found that fasting glucagon levels tended to decrease following 100 days of a 1000 kcal/d exercise-induced energy deficit. Further, hepatic glucagon sensitivity increases in response to unaccustomed endurance exercise , such that less glucagon is needed to evoke a given level of GP; and, fasting may elicit a similar response . Moreover, the maintenance of plasma glucagon levels over time for the DEF-HP group may be attributed to the higher protein content of the diet. Indeed, fasting plasma glucagon has been shown to be higher in energy-balanced individuals who habitually consume high-compared to normal-protein diets (~1.9 g/kg/d and ~0.75 g/kg/d, respectively, for 6 months), which was accompanied by a concomitant rise in gluconeogenesis in the high protein group .
Despite the differential effects of diet on GP, we did not observe differential effects on plasma glucose over time. Statistical analysis of the glucose values on day 4 and day 12 indicated that glucose fell over time; independent of group. However, the change-over-time scores visually suggest that the high protein group trend towards a greater decline over time. With that being said, the magnitude of the change over time is small. Moreover, the statistical tests were inconsistent with 2-way repeated measures ANOVA indicating a change across time; whereas the 95% confidence intervals for change scores of the diet groups overlap with zero change, suggesting that the real response may or may not have been different from the null. Our conclusion is that the glucose data should not be over-interpreted.
Generalizing study results to females should be done with caution, since the current study included males-only. Additionally, the methodology used to measure fgng should be taken into consideration when interpreting this data. The contribution of gluconeogenesis to endogenous glucose production observed in this study at baseline (i.e. prior to the increase in exercise energy expenditure) is similar to other studies employing the MIDA technique to estimate gluconeogenic flux [11, 32]. Differences between MIDA and other methods for estimating fgng have been presented in the literature [33, 34]. Briefly, fgng values determined via MIDA tend to be lower compared to nuclear magnetic resonance and liver biopsy methods, since MIDA involves the calculation of GB from measurements of GP and fgng whereas the other methods ascertain fgng via the difference between the measured GP and net change in glycogen . Nevertheless, the present investigation observed an increase in fgng in response to the increase in energy expenditure, which appears to be an adaptation to the unaccustomed increase in exercise.