Keywords

Obesity; fructose; sucrose; food intake

Overview

The obesity epidemic that we are experiencing in the United States is a crisis that affects all Americans. In general, obesity does not discriminate among ages, races, classes, or genders. In 1985, the rate of obesity across our country was less than 14% but by 2014 the rate was 35% (Ogden et al., 2014). Furthermore, obesity is quickly becoming a leading cause of major health problems and death. The research community, along with policy makers and health-care officials are grappling to find the causes of this sudden increase in Americans’ girth. Several researchers have targeted sugar—in particular, high-fructose corn syrup (HFCS)—as a potential accomplice in the obesity epidemic. Supporters have pointed out that fructose has become more prevalent in our diets over the past century. In 1900, the average fructose intake was 15 g/day. Fructose was consumed mainly through fruits and vegetables, which have the added benefit of fiber (Bray, 2010). However, as of 2010, fructose consumption had risen to 73 g/day and was being consumed in highly processed forms.

This research is further complicated by the varied forms of sugar in today’s Western diet. Sucrose, more commonly known as table sugar, is a disaccharide composed of a fructose molecule bonded to a glucose molecule. As a consequence, 50% of sucrose is fructose. HFCS is derived from corn sugar rich in glucose; the corn sugar is processed to increase fructose concentrations, which results in the much sweeter HFCS. Interestingly, HFCS can be produced to different sweetness by increasing the fructose concentrations. The most commonly used form of HFCS is 45% fructose followed by the less common 85% fructose. Therefore, HFCS is erroneously targeted as the sole cause of obesity. Rather, we believe fructose containing sugars, specifically in processed foods, may be the culprit of sugar-induced obesity. Further, while glucose and fructose are not typically used to sweeten foods and beverages alone, researchers still use the individual monosaccharides as treatment to better understand the molecular contributions of each sugar component.

One particularly attractive hypothesis linking obesity to fructose consumption is that increased fructose intake can disrupt normal liver metabolism and lead to an increase in hepatic lipogenesis. We will subsequently refer to this link as the hepatic lipogenesis hypothesis. A second but equally attractive hypothesis that links obesity to fructose is that monosaccharides quickly induce increases in both hepatic and adipose intracellular glucocorticoids that then promote increased fat accumulation. Finally, a third line of research that links obesity to fructose intake comes from recent findings from our laboratory that overnight access to fructose suppresses hypothalamic peptides that are involved with the regulation of normal hunger and satiety. In this chapter, we will review the data from inquiries into each set of experiments that link the intake of sugar—fructose in particular—to physiological conditions that favor the development of excess body fat. The review presented here is not exhaustive but puts our contributions in context so as to make the point that there is no one simple answer to the question, “How does an increase in sugar consumption promote obesity?”

The Lipogenesis Hypothesis

APOC3 and Hypertriglyceridemia

Large amounts of fructose in the diet can lead to HTG in both humans and laboratory rodents (Bocarsly et al., 2010; Teff et al., 2004; Teff et al., 2009). There are several plausible mechanisms that link fructose to HTG. We have recently examined the role of apolipoprotein C-III (APOC3) in promoting HTG in less than 24 h (Castonguay and Campbell, 2014). A lipoprotein, APOC3 is expressed in the liver of humans and rodents and is one of the most abundant apolipoproteins in plasma, with an average concentration of about 12 mg/dL. Plasma APOC3 concentrations are positively correlated to plasma triglycerides and VLDL triglycerides. In addition, transgenic mice that overexpress the APOC3 gene exhibit hyperlipidemia. Conversely, mice with suppressed APOC3 were protected from hyperlipidemia (Jong et al., 1999).

We tested various types of fructose-containing sugars (fructose, sucrose, and HFCS) with the intent of replicating our earlier observations that overnight access to fructose can promote HTG as well as examine the role of APOC3 in promoting HTG (Castonguay and Campbell, 2014). Briefly, 40 rats were randomly assigned to five weight-matched groups (n = 8). Rats assigned to the first group were given ad libitum access to control diet (Harlan rodent diet 7012). Rats assigned to the remaining four groups had ad libitum access to the control diet as well as ad libitum access to one of the following solutions: 16% weight/volume (w/v) fructose, 16% w/v glucose, or 16% w/v sucrose. Presented in Table 4.1 is a comparison of several measured endpoints. Note that only the rats that were fed fructose containing sugars differed from controls in circulating triglycerides under our experimental conditions.

Table 4.1

Average Insulin, Triglycerides, Glucose, and APOC3 in Plasma Following 24-h Sugar Access

Group	Insulin (ng/mL)	Triglycerides (mg/dL)	Glucose (mg/dL)	APOC3 (ng/mL)
Control	4.2a ± 0.6	73.4c ± 20.7	150.0a ± 2.0	321.9a ± 53.7
Fructose	4.5a ± 0.9	311.6a ± 41.6	155.7 a ± 2.9	412.7a ± 87.4
Glucose	4.4a ± 1.1	132.1bc ± 20.6	147.2a ± 3.4	208.4a ± 57.0
HFCS	4.2a ± 0.9	286.2a ± 31.6	148.8a ± 3.6	243.9a ± 64.4
Sucrose	4.5a ± 0.7	257.4ab ± 30.3	149.9a ± 2.1	333.8a ± 71.1

Group means ± SEM are presented here for plasma insulin, triglycerides, glucose, and APOC3. Superscripts with different letters indicate a significant difference at the p<.05 level. Refer to Campbell and Castonguay (2014) for further details.

Taken from Castonguay, T.W., Campbell, E.S., 2014. Fructose intake and circulating triglycerides: an examination of the role of APOC3. J. Diabetes Obesity 1, 1–7.

Since pure fructose is the most lipogenic of all the sugars, consumption of pure fructose was expected to elicit the greatest perturbation in APOC3 messenger RNA (mRNA) and protein. This was not the case. We found that sucrose- and glucose-fed groups had similar significant positive fold changes in APOC3 gene expression at 2.68 and 2.59, respectively, when compared to controls. Interestingly, HFCS had a positive fold change of 2.40 but failed to differ significantly from controls. Fructose consumption elicited a negative fold change at 0.86 but again this difference was not statistically significant different from control. Refer to Fig. 4.1.

Contrary to other studies that suggest APOC3 is responsible for HTG, we believe that there may be an alternative mechanism to elicit HTG in 24 h or less that has not yet been identified. This result suggests that the ratio of other lipoproteins to APOC3 may be the key to understanding how fructose consumption results in HTG. Clearly, more research is needed.

Dietary Sugars and Glucocorticoids: Additional Evidence Linking the Two to Obesity

London and Castonguay (2011) examined the acute effects of ad lib access to 16% solutions of sucrose (S16), fructose (F16), or glucose (G16) and chow and water. Diets high in fructose but not glucose or sucrose increased 11β-HSD1 mRNA within 24 h in liver and adipose by greater than two- and threefold, respectively (p ≤ 0.05). After 1 week, hepatic 11β-HSD1 mRNA and protein were suppressed by >60% in all sugar-fed groups, a phenomenon not previously reported in the absence of obesity. Sucrose- and fructose-fed rats had higher plasma triglycerides than did control or glucose-fed rats at both 24 h and 1 week (p ≤ 0.02), which is consistent with previously reported effects of fructose on lipid metabolism (refer to Fig. 4.2A–F).

Dietary fructose increased 11β-HSD1 mRNA in liver (p < 0.05, Fig. 4.2A) and mesenteric adipose (p = 0.05, Fig. 4.2B) within the first 24 h of exposure when compared to the mean levels of the control, S16, and G16 groups. Continued access to a fructose solution resulted in two outcomes: the suppression of hepatic 11β-HSD1 mRNA (p < 0.05, Fig. 4.2A) and an increase in 11β-HSD-1 mRNA in mesenteric adipose 11β-HSD-1 mRNA (p < 0.05, Fig. 4.2B). After 1 week of exposure to the experimental diets, 11β-HSD1 mRNA in liver was suppressed by greater than 60% in all of the sugar-fed groups compared to the control (p < 0.05, Fig. 4.2B). In mesenteric adipose, the increases in 11β-HSD1 mRNA in the sugar-fed groups at 1 week ranged from two- to sixfold that of the control group, yet mean levels of the S16 and G16 groups did not attain statistical significance.

After 24 h or 1 week of exposure to the experimental diets, there were no differences in mean H6PDH mRNA levels in liver or mesenteric adipose in comparison to control levels (Fig. 4.2C and D). All three sugar-fed groups had mean hepatic acetyl-CoA carboxylase (ACC) mRNA levels more than twice that of their respective controls at both 24 h and 1 week, although not all of these differences achieved statistical significance (Fig. 4.2F). After 24 h, mean hepatic ACC mRNA levels were significantly higher in the F16 and G16 groups (p < 0.05; Fig. 4.2F), and after 1 week the mean hepatic ACC mRNA level was increased in the G16 group compared to controls (p < 0.05; Fig. 4.2F). No changes in ACC mRNA levels were observed in mesenteric adipose.

Fructose increased 11β-HSD1 protein expression in liver after 24 h (p < 0.05, Fig. 4.2E), which was the same trend observed in 11β-HSD1 mRNA in the fructose-fed group. After 1 week, mean hepatic 11β-HSD1 protein levels were suppressed in the S16 and G16 groups (p < 0.05, Fig. 4.2E). Mean hepatic 11β- HSD1 protein of the F16 group was ~25% that of the control and approached being statistically significance versus the control (p = 0.06, Fig. 4.2E). There were no changes in hepatic H6PDH protein levels after 24 h of exposure to the experimental diets. After 1 week, the mean hepatic H6PDH protein levels were lower in all sugar-fed groups when compared to the mean level of the control group (p < 0.05, Fig. 4.2E).

One of the most important observations of the present work was that fructose, but not glucose, promoted increased 11β-HSD-1 message within 24 h of initial access. We can only speculate that the transient increases in mesenteric adipose and liver followed by suppression in hepatic message that was observed 1 week later was due to the fact that fructose, unlike glucose, bypasses the key regulatory step in glycolysis that otherwise limits flux through the cycle. Mechanistic studies aimed at a better understanding of how unregulated glycolytic activity can affect the expression or activity (or both) of 11β-HSD-1 and H6PDH are clearly warranted. One likely explanation is that the accumulation of glycolytic products triggers an acute inflammatory response and that increased local cytokine production in these key metabolic tissues impacts the transcriptional regulation of these genes.

Clearly, the postingestive effects of fructose change some of the fundamental controls of intracellular glucocorticoid regulation in both liver and adipose. Before this work, changes in 11β-HSD1 had been observed in several models of human and animal obesity, but it remained a question whether high-sugar diets can initiate changes in 11β-HSD1 or whether changes in 11β-HSD1 were the effect of increased adiposity caused by a high-sugar diet. The diet-induced obesity paradigm, as opposed to the use of a genetic obesity model, has enabled us to separate cause and effect to address this question. Fructose causes a disruption in the controls of intracellular glucocorticoid concentrations such as increased adipose 11β-HSD1 mRNA and protein. Similar disruptions have been repeatedly associated with the development of obesity.

Sugar Solutions and the Hypothalamus

Despite the extensive behavioral examinations of the rat’s avidity for sugar solutions, relatively little work has been focused on the impact of sugar intake on the central mechanisms controlling intake. For example, fructose can upregulate fatty acid amide hydrolase, an enzyme involved in the degradation of hypothalamic endocannabinoids, as well as other enzymes involved in the synthesis of endocannabinoids (Erlanson-Albertsson and Lindqvist, 2010). In addition, sugar solutions can have an effect on the release of dopamine in the nucleus accumbens, the brain’s so-called reward center (Avena et al., 2008). In addition to dopamine release, sugar consumption can alter receptor gene expression in reward areas of the brain. Rats with intermittent sugar and chow access also have decreases in dopamine receptor D2 mRNA in the nucleus accumbens compared with ad libitum chow controls (Spangler et al., 2004). Sucrose intake can influence D2R density specifically in subregions of the striatum (Bello et al., 2002) Given the dearth of work relating sugar to changes in the central nervous system as well as our observations on how brief access to fructose can double circulating triglycerides and liver and adipose 11βHSD-1 expression, we next turned our attention to the neural controls of food intake using the same experimental design where rats are given access to a dilute sugar solution overnight and compared with rats that had free access to food and water but no sugar solution.

The availability of a number of new gene-screening tools has given us the opportunity to monitor a large number of genes simultaneously. One such tool is a qPCR array developed by SABioscience (Gaithersburg, MD). In our first uses of this technology, we chose to screen the RNA extracted from the hypothalami of rats fed water, chow diet, and a 16% fructose solution. Controls were fed water and chow diet only. The rats were maintained on their respective diets overnight and then killed. At the time of sacrifice, brains were dissected and flash frozen at –80°C for subsequent analyses. Frozen brains were sectioned using an IEC Minot Custom Microtome (Damon/IEC Division) and 10 consecutive 20-μ-thick tissue hypothalamic region slices were sectioned. See Colley et al. (2012) for more details.

Dopamine receptors 1a and 2 and neuropeptide Y (NPY) were upregulated in the hypothalami taken from the fructose-fed group compared to control. Galanin, Brs3, agouti-related protein (AGRP), INSR, Gh1, Trhr, Atrn, NMB, tumor necrosis factor (TNF), and Thrb were downregulated in the fructose-fed group. Presented in Table 4.4 are the results of comparisons between fructose-fed and control-fed groups that were particularly pronounced. The PCR Array analyses revealed a number of genes that were either dramatically upregulated or silenced by overnight access to fructose. The data reported here include the observation that fructose can upregulate genes in the hypothalamus associated with the dopaminergic pathways. Smith and his group reported similar conclusions: hypothalamic dopamine plays an integral role in the control of sucrose intake (Simansky et al., 1985; Smith et al., 1987; Weatherford et al., 1990). In our preliminary array scan, we observed that overnight access to fructose led to significant upregulation in mRNA for both dopamine 1 and 2 receptors, consistent with their conclusions using microdialysis or dopamine antagonists.

The significance of this observation is that both fructose and sucrose have an effect on this dopaminergic system despite their separate metabolic fates. It is tempting to conclude that sensory properties associated with the intake of both sugars are determining this component of the response to the sugars. By contrast, the suppression of genes in the hypothalamus that are related to insulin (GLP1r, INSR, AGRP, and MCRh1 to name a few) suggests that other processes are involved. Presumably some of these endpoints are also part of the regulatory system involved in promoting increased intake, while others may be more involved in adjusting chow intake subsequent to the influx of calories from fructose. Only a more in-depth side-by-side comparison of the effects of different sugars will permit attribution to which of the changes reported here are specific to fructose and which are common to sugars.

In a follow-up study male, Sprague-Dawley rats were given access to food, water, and one of five different sugar solutions for 24 h, after which blood and tissues were collected. Access to the fructose solution (as opposed to other sugars that were tested) resulted in a doubling of circulating triglycerides. Glucose consumption resulted in upregulation of seven satiety-related hypothalamic peptides whereas changes in gene expression were mixed for remaining sugars. Also, following multiple verification assays, six satiety-related peptides were verified as being affected by sugar intake. These data provide evidence that not all sugars are equally effective in affecting the control of intake.

As encouraging as these results were, we were left with several questions that needed to be answered before going further. In particular, although we measured changes in whole hypothalamus, we still had no insight into which hypothalamic pathways were affected. As a consequence, we have subsequently replicated our earlier findings that overnight access to sugar solutions can affect hypothalamic gene expression but that some of these changes were specific to particular regions of the hypothalamus (Zhao et al., 2015a; Zhao and Castonguay, 2016; Zhao et al., 2015b).

Different hypothalamic structures and regions influence hunger and satiety. More than 60 years ago, Stellar proposed that the ventromedial nucleus of the hypothalamus and the lateral hypothalamic (LH) area acted together to control food intake. The dual-center hypothesis was one of the most studied theses in 20th-century neuroscience (Stellar, 1954). The paraventricular nucleus (PVN) was added to this mix later, noting that there were differences in metabolic and behavioral controls of hunger (Weingarten et al., 1985). Many neuropeptides synthesized in these hypothalamic regions play critical roles in energy maintenance. Accordingly, Zhao et al. examined how these neuropeptides were affected by different sugars in three hypothalamic regions: the paraventricular hypothalamic nuclei, the ventromedial hypothalamus (VMH), and the lateral hypothalamus (Zhao et al., 2015a).

Sprague-Dawley rats were provided with 24-h access to 15% solutions of glucose, fructose, sucrose, or HFCS and then killed. Portions of the PVN, VMH, and LH were then dissected. Expression of several neuropeptides in these tissues, all of which were previously shown to be influenced by free access to sugar solutions using PCR array, was subsequently measured. Of the four sugar solutions tested, only fructose decreased expression of cholecystokinin significantly and only in the PVN. Other differences between sugar-fed groups included the observation that glucose- and sucrose-fed rats significantly increased the expression of TNF-α only in the PVN and fructose and sucrose fed rats had decreased growth hormone in the VMH.

Zhao et al. went on to quantify the effects of access to different sugar solutions on the expression of hypothalamic 11b HSD-1 (Zhao et al., 2015b), finding that 11β-HSD1 was abundantly expressed in the hypothalamus. Specifically, 11β-HSD1 was mostly expressed in the LH. The remaining two hypothalamic regions (PVN and VMH) also produced 11β-HSD1 (see Fig. 4.3).

None of the sugars used had a significant effect on 11β-HSD1 expression in the PVN, VMH, or LH when compared to controls that were fed chow only. Interestingly, HFCS promoted an increase in 11β-HSD1 expression when compared with glucose- and sucrose-fed groups in both the PVN and VMH.

Sugars and the Hypothalamus: Evidence from Humans

Recently, we have begun using functional magnetic resonance imaging to gather further details about how sugars can and do affect the hypothalamus. Smeets et al. (2005b) has reported that oral consumption of a 25-g or 75-g glucose solution elicited a 1–2.5% signal decrease in the hypothalamus shortly after consumption. Furthermore, the 75-g dose induced a greater reduction in hypothalamic activity compared to the 25-g dose. Smeets et al. (2005a) also demonstrated that glucose infusion does not decrease hypothalamic activity to the same magnitude as oral glucose, thus the change in hypothalamic activity is only partially attributed to blood glucose concentration. When Page et al. (2013) compared glucose to fructose, hypothalamic response to both solutions was quite different. Fructose increased hypothalamic activity, while glucose decreased hypothalamic activity by the same magnitude from the baseline. Fructose is rapidly and efficiently taken up by the liver after digestion, leaving little available to reach the brain. As a consequence, the effects of fructose are most likely a secondary response to fructose metabolism.

We set out to identify the effects of HFCS on hypothalamic activity via cola. HFCS comes in many glucose-to-fructose ratios. The most common is 45:55 glucose to fructose. Given the opposing effects of glucose and fructose, hypothalamic activity could remain unchanged when administered simultaneously. However, glucose acts directly on the brain in conjunction with insulin, whereas fructose does not. As a result, we predicted hypothalamic response would reflect the effects of the glucose.

Correlational analyses were used to analyze the relationship between hypothalamic response and circulating metabolites. Hypothalamic response as measured by average hypothalamic signal intensity was positively correlated with fasting triglycerides as well as fasting insulin but not fasting glucose. Refer to Table 4.4. Higher-fasting triglycerides and insulin were associated with higher hypothalamic signal intensity, thus the hypothalamus was less responsive to treatment. High-fasting triglycerides and insulin are indicators of metabolic dysfunction, thus the hypothalamic response is partially inhibited by errors of metabolism.

Our results suggest that circulating triglycerides and circulating insulin influence hypothalamic responses. Insulin binds to many neurons within the hypothalamus as a modulator of intake, including NPY and AGRP neurons. Triglycerides are digested into free fatty acids that might indirectly modulate hypothalamic activity by regulating leptin activity. Glucose acts on the glucose-sensing neurons of the hypothalamus. However, the exact function of these neurons has not been determined. The locations of these neurons are most dense in the lateral hypothalamus. Unfortunately, the lateral hypothalamus may be poorly represented due to scanning limitations.

Body mass index (BMI) does not correlate with hypothalamic signal intensity. This result may be due to BMI’s inability to reliably measure metabolic dysfunction. While a BMI greater than 30 is the definition of obesity, BMI is a ratio of height to weight but does not reliably measure body composition or unhealthy fat deposits such as abdominal obesity that are better indicators of metabolic dysfunction.

Conclusions and Summary

The body of literature that has been presented in this review is only a fraction of the work that surrounds the question “Does fructose induce obesity?” Our answer to this question is “Fructose can.” Here we have presented three different approaches to answering the question, and all three sets of experiments provide evidence that fructose consumption is “not simply sugar,” but rather that fructose can promote physiological changes in brain, adipose tissue, and liver that are all conducive to the increased deposition of body fat. Fructose intake changes circulating lipids, promotes an increase in intracellular glucocorticoid concentrations in liver and adipose, and causes changes in genes that control hunger and appetite. One question that has been resolved by some of this work has to do with answering the question of whether or not it was the obesity induced by fructose access that was responsible for the changes in gene expression and intracellular steroids. Our overnight access paradigm has offered a resolution to this question: it is the sugar that is inducing these changes, long before excessive adiposity takes place.

Finally, we would like to echo the comment made by London several years ago (London and Castonguay, 2011). We advocate that mechanistic studies be conducted that are aimed at a better understanding of how unregulated glycolytic activity can affect the expression or activity of 11β-HSD-1 and H6PDH. Future research should be conducted that provides an explanation of how the accumulation of glycolytic products triggers an acute inflammatory response and that increased local cytokine production in these key metabolic tissues impacts the transcriptional regulation of these genes.