|Home | About | Journals | Submit | Contact Us | Français|
Whether intramuscular triglyceride (IMTG) concentration or flux is more important in the progression to type 2 diabetes is controversial. Therefore, this study examined IMTG concentration, as well as its fractional synthesis rate (FSR), in obese people with normal glucose tolerance (NGT; n = 20) vs. obese people with prediabetes (PD; n = 19), at rest and during exercise. Insulin action and secretion were assessed using an intravenous glucose tolerance test. [U-13C] palmitate was infused for 4 h before and throughout 1.5 h of treadmill walking at 50% VO2max. IMTG concentration was measured by gas chromatograph/mass spectrometer, and FSR by gas chromatography–combustion isotope ratio mass spectrometer, from muscle biopsies taken immediately before and after exercise. Basal IMTG concentration was higher (43 ± 5.7 vs. 27 ± 3.9 mg/mg dry weight, P = 0.03) and FSR trended lower (0.23 ± 0.04 vs. 0.32 ± 0.05/h, P = 0.075), as did insulin action (Si; 2.9 ± 0.43 vs. 3.3 ± 0.35 × 10−4/mU/ml, P = 0.07), in PD vs. NGT. IMTG concentration did not change significantly during exercise, but was no longer different in PD vs. NGT (45 ± 7.7 vs. 37 ± 5.8 mg/mg dry weight, P = 0.41). IMTG FSR suppressed during exercise in NGT (−81% to 0.06 ± 0.13/h, P = 0.02), but not PD (+4% to 0.24 ± 0.13%/h, P = 0.95). Palmitate oxidation was similar during rest (P = 0.92) and exercise (P = 0.94) between groups, but its source appeared different with more coming from muscle at rest and plasma during exercise in NGT, whereas the converse was true in PD. Altogether, higher basal IMTG concentration that is metabolically inflexible distinguishes obese people with PD from those with NGT.
The global burden of type 2 diabetes continues to rise. National statistics estimate that roughly 25 million Americans currently have diabetes, and 57 million are at high risk because they have prediabetes (PD) (1). Although two-thirds of people with diabetes are overweight or obese (2), only 2–13% of those who are simply obese will ever acquire diabetes (3), whereas up to 70% of those with PD (isolated impaired fasting glucose, isolated impaired glucose tolerance, or both impaired fasting glucose + impaired glucose tolerance (4)) will acquire the disease (5). Implied are more severe adverse changes in tissue-specific insulin action at any given level of body fatness in those most predisposed. Therefore, understanding and exploiting differences between simple obesity and PD may be key in understanding who will acquire diabetes and who will not.
Accretion of lipid into skeletal muscle appears an early and key alteration in the development of type 2 diabetes (6), thus deserves further consideration as a potentially distinguishing feature between simple obesity and PD. Repeated observations using different techniques have noted a positive linear relationship between intramuscular triglyceride (IMTG) concentration and insulin resistance (7–9). Nevertheless, the causal relationship between the two is unconfirmed. For example, high IMTG concentration in trained athletes (10), as well as in transgenic mice overexpressing diacylglycerol acyltransferase-1 (11), is associated with enhanced, not diminished, insulin sensitivity. This apparent conflict has led to speculation about the role of IMTG synthesis, rather than total concentration, as a link to insulin resistance (12). Specifically, low IMTG synthesis rates may lead to production of insulin-desensitizing intermediates (11,13,14). Surprisingly, however, neither IMTG synthesis nor degradation have been directly measured in people with chronic insulin resistance.
Therefore, the aims of this study were the direct measurement of (i) IMTG concentration and (ii) fractional synthesis rate (FSR), as possible distinguishing features between simple obesity and PD. All measures were made before and after an acute bout of exercise as IMTG FSR has been shown to fully suppress during exercise in healthy humans (15), thus a failure to do so would demonstrate a defect in the dynamic regulation of the IMTG pool. We hypothesized that in people with PD vs. simple obesity (i) IMTG concentration would be higher, (ii) IMTG FSR would be lower, and (iii) IMTG FSR would not change in response to acute exercise.
Research volunteers were solicited from the general population by campus e-mail, postings, and health fairs held in the Denver, Colorado metro area. A total of 39 healthy, sedentary (<90 min/week planned activity), nonsmoking men and postmenopausal women between the ages of 45 and 70 years were placed into 1 of the 2 groups based on two 2-h 75-g oral glucose tolerance tests, separated by 1 week: a control group with normal glucose tolerance (NGT; n = 19; fasting glucose <5.6 mmol/l and 2-h oral glucose tolerance test <7.8 mmol/l), or a group with PD (n = 20; fasting glucose 5.6–6.9 mmol/l, and/or 2-h oral glucose tolerance test 7.8–11.1 mmol/l) (ref. 16). The protocol was approved by the Colorado Multiple Institutional Review Board before recruitment, and Informed consent was obtained from all participants according to the principles of the Declaration of Helsinki as revised in 2004.
This was estimated from dual-energy X-ray absorptiometry.
Oxygen consumption and carbon dioxide production were measured using indirect calorimetry (model 2900; Sensormedics, Yorba Linda, CA) during treadmill walking. To ensure that the each subject had reached his/her VO2max, the following criteria must have been met. (i) Plateau in oxygen consumption at increasing workload, (ii) respiratory exchange ratio >1.10 at maximal exercise, and (iii) achievement of age-predicted maximum heart rate.
A modified frequently sampled intravenous glucose tolerance test was performed as previously described by Bergman et al. (17). Insulin sensitivity (Si) and secretion (acute insulin response (AIR) and the disposition index (DI) (Si × AIR = DI)) were calculated using the MINMOD computer program (Millenium version; MINMOD, Los Angeles, CA).
Subjects were fed a control diet for 3 days before admission to the General Clinical Research Center for study. The control diet was isocaloric (calculated as 1.4 × (372 + (23.9 × fat-free mass))), using the fat-free mass measured by dual-energy X-ray absorptiometry. The diet composition was standardized as: 30% fat (saturated, polyunsaturated, and monounsaturated fats in a 1:1:1 ratio), 15% protein, and 55% carbohydrate.
Subjects were fasted overnight (~12 h) and were admitted to the General Clinical Research Center at 07:30 on the morning of the IMTG synthesis study. Upon admission, an intravenous catheter was placed in an antecubital vein for infusion, and sampling catheter was placed in a dorsal hand vein of the contralateral arm. For all blood samples, the heated hand technique was used to arterialize the blood. Background sampling began 30 min after sampling catheters had been placed. A baseline blood sample was drawn for determination of circulating hormone and substrate concentrations (catecholamines, insulin, glucose, c-peptide, glucagon, free fatty acids (FFAs), glycerol, and lactate). Following the baseline blood draw, a continuous infusion of [U-13C] palmitate (Isotec, Miamisburg, OH) bound to human albumin was initiated at 0.0174 mmol/kg/min and continued throughout the study. Subjects remained semirecumbent for 4 h to allow for tracer incorporation into the intramuscular lipid pools. Blood samples were taken for hormone and substrate concentrations (as above) during the final 30 min of the 4-h rest period. Following the rest period, a vastus lateralis skeletal muscle biopsy was performed using the Bergstrom technique (18). Muscle was immediately flash frozen in liquid nitrogen and stored at −80 °C until dissection and analysis. Subjects were then escorted to the treadmill where they walked at 50% VO2max for 90 min. [U-13C]palmitate infusion rate was doubled during was exercise to minimize changes in isotopic enrichment. Indirect calorimetry was performed three times during the exercise bout to confirm the intensity and measure whole-body substrate oxidation using standard equations (19). Four blood samples for tracer determination, as well as hormone and substrate concentrations (as above) were taken during the final 30 min of the 90-min exercise period. A second muscle biopsy was taken from the contralateral leg immediately after exercise.
All samples were stored at −80 °C until analysis. Radioimmunoassay was used to determine insulin and glucagon (Linco Research, St Louis, MO), as well as c-peptide (Gamma counter; Diagnostic Products, Los Angeles, CA), concentrations. Standard enzymatic assays were used to measure glucose (COBA-Mira Plus; Roche Diagnostics, Mannheim, Germany), lactate (Kit #826; Sigma, St Louis, MO), glycerol (Boehringer Mannheim Diagnostics, Mannheim, Germany), and FFA (NEFA Kit; Wako, TX). Epinephrine and norepinephrine concentrations were measured using high-performance liquid chromatography (Dionex, Sunnyvale, CA).
Skeletal muscle samples were dissected free of extramuscular fat on ice as described by Guo et al. (20). Muscle (~70 mg) was lyophilized, added to 1 ml iced MeOH along with internal standards of tripentadecanoic acid and dipentadecanoic acid, and homogenized (Omni TH; Omni International, Marietta, GA). Total lipids were extracted (21) and then added to solid-phase extraction columns (Supelclean LC-NH2, 3 ml; Supelco Analytical, Bellefonte, PA) to isolate FFAs, and IMTG as described by Kaluzny (22). The FFA fraction was methylated using 0.5 ml 2% sulfuric acid, and heated at 100 °C for 1.5 h. The IMTG fraction was converted to a fatty acid methyl ester by transmethylation using sodium methoxide. Stable isotope ratios of 13C in fatty acid methyl esters were measured using a gas chromatography–combustion isotope ratio mass spectrometer system ( ermo Electron, Bremen, Germany). Enrichment was calculated based on a standard curve of known enrichments, and corrected for variations in abundance (23). Concentration and composition analysis was performed on an HP 6890 GC with a 30-m DB-23 capillary column, connected to an HP 5973 MS. Peak identities were determined by retention time and mass spectra compared to standards of known composition.
Two milliliter of breath CO2 was transferred into a 20-ml exetainer for the measurement of 13CO2/12CO2 with continuous flow isotope ratio mass spectrometry (Delta V; Thermo Electron). Each sample was injected (1.2 μl/injection) in duplicate for isotope ratio analyses, with an average standard deviation for all injections of 0.0001 atom percent.
Methylation and extraction of plasma palmitate was performed as previously described (24). Samples were run on an HP 6890 GC with a 30-m DB-23 capillary column, connected to an HP 5973 MS. Enrichments were calculated based on a standard curve of known enrichments, and corrected for variations in abundance (23). Peak identities were determined by retention time and mass spectra compared to standards of known composition.
Frozen skeletal muscle samples were weighed, and homogenized on ice using a Kontes glass homogenizer (Kimble/Kontes, Vineland, NJ) in buffer previously described (25). Samples were agitated at 4 °C for 2 h, then spun at 16,000 g for 15 min to pellet insoluble protein. Supernatant was saved, and used to determine protein concentration (Calbiochem, San Diego, CA).
Forty micrograms of sample protein and an internal standard were run on an sodium dodecyl sulfate–polyacrylamide gel electrophoresis 8% Bis–Tris gel (Invitrogen, Carlsbad, CA), transferred to a polyvinylidene fluoride membrane, and blocked with 5% bovine serum albumin for 1 h at room temperature. Primary antibody incubations were performed in 5% bovine serum albumin overnight at 4 °C, and an horseradish peroxidase–conjugated secondary antibody was incubated for 1 h at room temperature. Enhanced chemiluminescence was used to visualize protein bands of interest. Intensity of protein bands was captured using an AlphaImager 3300 and quantified using FluorChem software (Alpha Innotech, San Leandro, CA). Protein bands were normalized to glyceraldehyde 3-phosphate dehydrogenase as a loading control, and then normalized to an internal standard that was run on each gel. Antihuman myosin A4.840 and A4.74 antibodies were purchased from the University of Iowa Hybridoma Bank (Iowa City, IA), anti-rabbit succinate dehydrogenase and peroxisome proliferator–activated receptor-α (Santa Cruz Biotechnology, Santa Cruz, CA), MAP4K4 (Abgent, San Diego, CA), IRS-1ser636 and IRS-1total (Cell Signaling Technology, Danvers, MA), PKC-ε (Cell Signaling Technology, Beverly, MA), and CPT-1 (Alpha Diagnostics International, San Antonio, TX) antibodies were commercially available. The rabbit anti-4-HNE antibody was a generous gift from Dr Peterson (University of Colorado–Denver). Secondary antibodies were from Bio-Rad (Hercules, CA).
IMTG FSR was calculated as previously described for use with stable isotopes (15), and is expressed in %/h to account for interindividual differences in IMTG pool size. The average enrichment of the skeletal muscle FFA pool during the 4-h equilibration period was used to represent the precursor pool from which triglyceride was synthesized (26). Specifically, IMTG FSR at rest was calculated according to the published methods of Guo et al. (27):
whereas IMTG FSR during exercise was calculated as described by Sacchetti et al. (15):
where EIMTGpalm(t1) and EIMTGpalm(t2) are enrichments of palmitate in the IMTG pool at baseline (after the 4-h tracer equilibration period) and during exercise, respectively. EIMTGpalm(t0) is the enrichment of background palmitate in the IMTG pool. EFFApalm(t1) and EFFApalm(t2) are the enrichments of skeletal muscle palmitate in the FFA pool at baseline and during exercise, respectively. Background enrichment of palmitate was determined as previously described (28).
where FFA represents concentration of individual FFA species in IMTG after transmethylation.
Palmitate rate of disappearance (Rd) and palmitate rate of oxidation were calculated using steady state kinetics and a whole-body estimate of carbon label retention as previously described (29). Calculation of palmitate oxidation rates were made using the methods of Wolfe et al. (29):
Published values for the acetate recovery factor (K) in obese humans at rest and during exercise (30), rather than individually measured values, were used. Palmitate incorporation rate into IMTG was calculated as the product of IMTG FSR and palmitate pool size in IMTG. Fat-free mass was used to extrapolate skeletal muscle palmitate IMTG storage to the whole body.
Testing of the data revealed a non-normal distribution; therefore, analyses were conducted on log-transformed values. Comparisons between groups at baseline was made using Student’s t-tests, whereas comparisons before and during exercise were made using two-way ANOVA (SPSS, Chicago, IL). Interactions within the data were examined and found only to be present for sex and IMTG concentration; therefore, the final ANOVA analysis for group differences in IMTG concentration was adjusted for difference in numbers of men and women between groups. All other data are unadjusted. The relationship between IMTG concentration and FSR was analyzed using a “best-fit” line. All data are presented as mean ± s.e.m. Statistical significance was set at P ≤ 0.05 and statistical trends are mentioned where P = 0.05–0.10.
Baseline subject demographics are summarized in Table 1. Briefly, male sex and a positive family history of type 2 diabetes were more prevalent in the PD group, whereas the NGT group had more women and people of nonwhite ethnicity. Otherwise, age, BMI, percent body fat, and VO2max were not significantly different between groups.
A nonsignificant trend toward lower insulin action (Si) in PD was observed (2.9 ± 0.43 vs. 3.3 ± 0.35 × 10−4/mU/ml, PD vs. NGT, P = 0.07), while insulin secretion (AIR) was significantly lower in PD (AIR 154 ± 24 vs. 434 ± 67 mU/ml, PD vs. NGT, P < 0.01). The composite of insulin action and secretion (DI) was also lower in PD (347 ± 57 vs. 1,281 ± 170 × 10−4/min, PD vs. NGT, P < 0.01).
Fasting and 2-h glucose concentrations were higher in PD vs. NGT, by study design. Otherwise, concentrations of substrates and hormones were similar between groups at baseline (Table 2). Glucose concentration decreased during exercise in the PD group, whereas insulin concentration decreased in the NGT group. FFAs, glycerol, epinephrine, and norepinephrine concentrations increased in both groups during exercise (P < 0.05 vs. rest for all comparisons; Table 2), but were not different between groups.
When expressed in absolute terms, IMTG FSR was not different in PD vs. NGT at baseline (191 ± 23 vs. 210 ± 21 mg/mg dry weight/h, P = 0.56) or during exercise (60 ± 109 vs. −12 ± 112 mg/mg dry weight/h, P = 0.61). However, IMTG concentration was significantly higher in PD vs. NGT (P = 0.03; Figure 1a). A nonsignificant trend for a lower relative IMTG FSR (0.23 ± 0.04 vs. 0.32 ± 0.05/h, PD vs. NGT, P = 0.075; Figure 1b) was observed in PD at baseline. No differences in IMTG concentration (45 ± 7.7 vs. 37 ± 5.8 mg/mg dry weight, PD vs. NGT, P = 0.41) or FSR (0.24 ± 0.13 vs. 0.06 ± 0.13/h, PD vs. NGT, P = 0.27) were seen after exercise. However, people with NGT significantly suppressed IMTG FSR during exercise (P = 0.02; Figure 1b), whereas those with PD did not (P = 0.95). IMTG saturation (32 ± 1.2 vs. 33 ± 1.5%, PD vs. NGT, P = 0.94), as well as fatty acid species within IMTG (data not shown), were similar between groups. IMTG concentration and FSR were strongly inversely correlated in the group as a whole (R2 = −0.80, P < 0.001; Figure 2).
Respiratory exchange ratio, a measure of whole-body substrate use, was similar between groups at rest (0.78 ± 0.01 vs. 0.77 ± 0.01, PD vs. NGT, P = 0.53) and during exercise (0.78 ± 0.01 vs. 0.79 ± 0.01, PD vs. NGT, P = 0.63). Whole-body palmitate rate of disappearance (Rd), as well as its oxidation and incorporation into IMTG, is depicted in Figure 3. Results show that plasma palmitate Rd increases with exercise in both groups (P < 0.01), but is higher in NGT vs. PD at baseline and during exercise (P < 0.01 for both). Plasma palmitate oxidation also increased with exercise in both groups (P < 0.01), but was similar between the groups at baseline (P = 0.92) and during exercise (P = 0.94). Incorporation of palmitate into IMTG (Figure 3) was similar between groups.
Western blots were run to examine known mediators of IMTG (e.g., muscle fiber type) and processes that IMTG may indirectly influence (e.g., oxidative capacity, insulin signaling, lipid peroxidation). Protein abundance for type 1 and 2 muscle fibers, myosin A4.840 and A4.74, respectively, were similar between groups. Similarly, succinate dehydrogenase, MAP4K4, 4-HNE, CPT-1, IRS-1ser636, ratio of IRS-1ser636/IRS-1total, peroxisome proliferator–activated receptor-α, PCK-e were not different between groups (Table 3; blots not shown).
Type 2 diabetes is considered one of the greatest health crises of our time. People with PD carry significantly higher risk for diabetes than those with simple obesity (3,31), therefore, discerning metabolic differences may prove useful as therapeutic targets in diabetes prevention. This study examined the role of intramuscular lipid concentration and FSR as potentially distinguishing features between simple obesity and PD. Major findings from this study demonstrated higher basal IMTG concentration with synthesis rates that were metabolically “inflexible” in obese people with PD compared to a matched control group with NGT.
Over the past two decades, considerable attention has been paid to the frequent association between high IMTG concentration and insulin resistance (7,9,32,33). Nonetheless, a direct cause–effect relationship between the two is unlikely for several reasons. First, mounting evidence shows that IMTG concentration and insulin action can be dissociated. Such was the case in trained athletes (10), and endurance-trained formerly sedentary obese individuals (34), where insulin sensitivity was maintained in an environment of high IMTG concentration. In addition, transgenic mice overexpressing diacylglycerol acyltransferase-1 display high levels of IMTG with enhanced insulin sensitivity compared to their wild-type littermates (11). Second, no support is found in the literature for IMTG concentration directly causing insulin resistance. Rather, metabolites involved in the synthesis and/or degradation of IMTG may be more involved in mediating peripheral insulin action (35–37). Together, these data have led to speculation about the role of IMTG synthesis, rather than total concentration, as a link to insulin resistance (12). We believe that our data are the first to examine IMTG FSR in insulin-resistant humans. Our data suggest that obese individuals with PD tended to have lower insulin sensitivity and lower rates of IMTG synthesis than a matched group of obese individuals with NGT. Whether this relationship is a cause or a result of insulin resistance is unclear.
The relevance of IMTG synthesis to insulin action has been highlighted in several recent studies. Schenk et al. clearly demonstrated the prevention of fatty acid–induced insulin resistance following a single bout of exercise in humans that related to increased IMTG synthesis, decreased DAG and ceramide, but not decreased IMTG concentration (14). Increased insulin sensitivity has also been reported after exercise training, without changing IMTG concentration in obese individuals (34), presumably due to enhanced IMTG FSR. In rodents, 1 week of exercise training or diacylglycerol acyltransferase-1 over-expression increased IMTG synthesis and protected against fat-induced insulin resistance, likely from decreased DAG and ceramide concentration (11). Cell culture studies also suggest protection from lipotoxicity with increased IMTG synthesis (38). Therefore, high rates of IMTG synthesis may act indirectly by decreasing the concentration of insulin-desensitizing intermediates, such as diacylglycerol (DAG), long-chain Acyl CoA, and/or ceramide (35–37), that can modulate insulin action through activation of protein kinase C. Although no differences in PKC-ε were noted in this study, this measure was made in the basal (non-insulin-stimulated state) in two very closely matched groups, likely underestimating the indirect effects of IMTG FSR on insulin signaling and action.
Perhaps more noteworthy than the nonsignificant trend for higher IMTG FSR in NGT vs. PD in the basal state was the failure of IMTG FSR to suppress during exercise in the PD group. Sacchetti et al. reported a basal IMTG FSR of 3.5%/h that completely suppressed during exercise in a group of healthy, young, lean volunteers (15). First, basal IMTG FSR was ~0.3%/h in subjects in this study highlighting the dramatic effect of obesity and insulin resistance on this parameter. Second, the suppression of FSR during exercise was only seen in the NGT group, supporting the notion that the IMTG pool is metabolically inflexible in the progression toward diabetes. The latter contention is further supported by examination of palmitate handling. At baseline, palmitate Rd was higher in NGT. Given that IMTG concentration was lower and FSR trended higher in this group, one would expect palmitate oxidation to be higher in NGT vs. PD, but this was not the case. It is possible, however, that although overall oxidation rates were similar, the source of the palmitate oxidized was different with more coming from muscle in NGT and more from blood in PD. The higher basal palmitate Rd in NGT is likely a composite of higher oxidative and nonoxidative palmitate disposal into muscle in this group. Palmitate Rd increased in both groups during exercise, but its fate appeared to reverse from what is observed at baseline. The suppressed FSR and decreased palmitate incorporation into IMTG during exercise in NGT suggests that plasma-derived palmitate comprised the majority of fat oxidation under these conditions. In PD, maintenance of pre-exercise FSR and palmitate incorporation into IMTG implies that Rd during exercise may be a composite of greater cycling of palmitate through the IMTG pool. Together, failure to suppress IMTG FSR (during exercise or possibly postprandially) in PD may affect substrate trafficking, leading to accumulation of IMTG despite a low IMTG FSR in the basal state.
Known mediators of IMTG concentration (presumably by affecting FSR), such as training status (10), nutrition (39), and gender (40) were controlled for in our study. Similar amount of type 1 and 2 muscle fibers between the groups also makes this an unlikely contributor to differences in IMTG FSR before or during exercise (41). Furthermore, substrate availability was also not different either at rest or during exercise. Markers of other relevant biologic processes, such as inflammation (MAP4K4), oxidative capacity (succinate dehydrogenase), capacity for mitochondrial lipid transport (CPT-1), lipid peroxidation (4-HNE), and trafficking (peroxisome proliferator–activated receptor-a) implicated by previous investigations, were also unrevealing in this study and may reflect the close matching between groups.
There are several limitations of this study worth noting. In order to focus on metabolic differences between simple obesity and PD, no lean control group was studied. Although this likely diminished finding differences between PD and “normal,” the design was vital in differentiating high- vs. low-risk individuals who are often grouped together. It is also possible that the imbalance in the gender distribution between groups confounded our results. Previous reports of sex differences in lipid metabolism have largely attributed their findings to sex differences in insulin action (42,43). Of note, no sex differences in Si were noted in this study making the group difference in sex distribution less apt to explain our findings. In addition, people with impaired fasting glucose and impaired glucose tolerance were grouped together as “PD”, despite reports that their site of insulin resistance differs (44,45). In so doing, differences between NGT and PD may be underestimated. Lastly, a published, nonindividually measured, acetate recovery factor was used in the calculation of palmitate oxidation. This was done because IMTG dynamics, not palmitate oxidation, was the primary outcome of interest.
In conclusion, delineating differences between simple obesity and PD is vital in knowing who will acquire type 2 diabetes and who will not. This study examined the role of intramuscular lipid concentration and synthesis as potentially distinguishing features between simple obesity and PD. Major findings from this study demonstrated higher basal IMTG concentration that was metabolically “inflexible” in people with PD compared to a matched control group with NGT. Longitudinal studies are needed to determine whether alterations in the dynamics of IMTG synthesis can prevent diabetes in this high-risk group.
We owe the success of this work to the research subjects who volunteered their time to participate, the committed staff of the General Clinical Research Center, as well as to the National Institutes of Health, who funded this work (grant# NIH DK-064811, DK-059739, and RR-0036).
The authors declared no conflict of interest.