Mechanisms of weight loss after obesity surgery

Obesity surgery remains the most effective treatment for obesity and its complications. Weight loss was initially attributed to decreased energy absorption from the gut but have since been linked to reduced appetitive behaviour and potentially increased energy expenditure. mechanisms associating rearrangement of the gastrointestinal tract with these metabolic outcomes include central appetite control, release of gut peptides, change in microbiota and bile acids. However, the exact combination and timing of signals remain largely unknown. In this review, we survey recent research investigating these mechanisms, and seek to provide insights on unanswered questions over how weight loss is achieved following bariatric surgery which may eventually lead to safer, nonsurgical weight-loss interventions or combinations of medications with surgery. 21 . A recent study investigated the expression of hypothalamic NPY and AgRP in obese mice, following RYGB and compared the results to a weight-matched model. During the first two post-operative weeks, when the peak weight loss was observed, hypothalamic Agrp and Npy gene expression did not increase compared to mice undergoing sham surgery, suggesting that compensatory hunger signals in the RYGB mice were not activated. In contrast, when the same amount of weight loss was achieved by caloric restriction in a different group of mice, increased expression of Agrp and Npy was observed. Of note, Pomc expression was not altered to a similar degree as Agrp , indicating that RYGB suppresses the adaptive hunger response triggered by weight loss 22-24 . Similarly, VSG does not change Npy and Agrp gene expression in obese rats 4 weeks after surgery 25 . A study that compared VSG and AGB-treated obese rats six weeks after surgery showed that the hypothalamic expression of Npy was significantly lower and the expression of Pomc was significantly higher in the VSG group 26 . Given the similar post-operative time points, any discrepancies between these studies’ findings on Agrp, Npy and Pomc may be due to rodent strain, differences, diet type and length of exposure, and variations in surgical technique. RYGB and VSG, i.e. during the acute phase of negative energy balance 45, 46 , but these findings have not been replicated in animal models of RYGB during the stable energy balance phase 47 . The valid measurement of the consummatory reward value of taste is challenging in humans as it relies entirely on the use of indirect measures like visual analogue scales (VAS). Studies using VAS after RYGB surgery have shown discrepant results 44, 48 . There is more consistency in the rodent literature, in which orofacial responses, a good marker of consummatory responses, increase for low concentrations of glucose and decrease for high concentrations of glucose after RYGB 49, 50 . The third domain of taste function is termed digestive preparation and salivation is a marker of this reflex response to tastants. Rates of salivation correlate with the rewarding aspects of the tastant and people with obesity demonstrate higher salivation rates to normal-weight controls 51 . Attempts have been made to measure salivation rates after obesity surgery but with mixed results 52 . Our group’s experience with measuring salivation rates in humans is that the available methodologies suffer from low reproducibility (unpublished data). fact dieting and obesity surgery-induced changes in body weight are triggered by different mechanisms 112 . Discrepancies exist regarding the postoperative timing of bile acid increase, as some studies report an acute effect 109 whether others observe a gradual increase 1 year following surgery 113, 114 . of FGF21 after surgery remain more controversial between different possibly because circulating concentration and sensitivity shown to weight which can differ widely 99, 115-117 . revealed no differences between sham VSG mice at after indicating TGR5 maintain loss and fat mass reduction after VSG 123 . A possible mechanism of this is a TGR5-driven mitochondrial separation and turnover of white adipose tissue to beige, as administration of TGR5-selective bile acid mimetics to thermoneutral housed mice led to the appearance of beige adipocyte markers and an increase in mitochondrial content 124 . However, not all studies report a reduction of weight loss following VSG and RYGB in TGR5 knockout mice 125, 126 . A possible explanation for this is the rate of regain following and as a the type and length of exposure to high-fat diet in pre-operative mice. Most studies investigating the role of


Introduction
Obesity surgery over the past six decades has been successful not only in providing a means of achieving substantial weight loss but also in giving us many novel insights on the pathophysiology of obesity. Obesity surgery was first described in the 1960s, when it was observed that patients with sub-total gastrectomy for cancer lost a considerable amount of weight 1 . Several modifications to the technique led to the first laparoscopic gastric bypass in 1994 2 , and the establishment of the three techniques most widely-used in clinical practice today.
The two main approaches that are currently performed widely are Roux-en-Y Gastric Bypass (RYGB) and Vertical Sleeve Gastrectomy (VSG). RYGB involves the creation of a small gastric pouch (~30 mL) that is anastomosed to the proximal jejunum, which has been transected at 30-75 cm from the ligament of Treitz, to form the "alimentary limb". The continuity of the intestine is restored via a jejuno-jejunal anastomosis between the alimentary limb and the excluded biliopancreatic limb approximately 75-150 cm distal to the gastrojejunostomy 3 . As a result, food bypasses most of the stomach, the entire duodenum, and the proximal jejunum.
VSG involves dividing the stomach along its vertical length to create a sleeve and removing ∼75% of its volume 4 . Although decreasing in popularity, the adjustable gastric banding (AGB) involves placing a silicone ring around the proximal stomach, bellow the gastroesophageal junction. The ring pressure is adjusted through fluid injected or withdrawn from a subcutaneous port 5 .
Efficacy is not the same among procedures, as RYGB and VSG cause more weight loss compared to AGB 6 . Patients benefit not only form weight loss, but more vitally A c c e p t e d M a n u s c r i p t from improvements in glycaemic control 7 , reduced cardiovascular morbidity and mortality 8 and reduced incidence of cancer 9 . All three procedures cause no mechanical restriction with little or no macronutrient malabsorption. Instead, weight loss is due to changes in the physiology of body weight regulation.
In this review, we will explore the biological mechanisms underpinning weight loss. .
We will not discuss the mechanisms underlying glycaemic/metabolic improvements as they fall outside the already wide scope of this review. The impact of obesity surgery on metabolism appears to be predominantly because of the substantial and sustained weight loss, but given the large number of mechanisms which are not weight loss related we expect that the beneficial metabolic outcome at individual level may be a composite of the weight loss together with non-weight loss related mechanisms.
We will focus on mechanistic studies in humans and animal models focusing on RYGB, VSG and AGB, as they are the most commonly performed operations. While animal data may not always apply to humans, they also raise new questions that can be answered in humans and answer questions that cannot be answered in humans.
Eating behaviour i.

Reduction in energy intake
The setpoint theory supports the notion that an individual's body weight trajectory during life is predominantly influenced by their genetic make-up, which interacts with non-biological factors (e.g., social, psychological) to determine the final phenotype 10 .
Any weight loss below or above the set-point is perceived as an alarm signal by the A c c e p t e d M a n u s c r i p t areas of the brain that regulate energy intake and expenditure, such as the hypothalamus and brainstem 11 . These areas are located in the subcortical areas of the brain involved in automated function like respiration or body temperature. The hypothalamus and brainstem receive continuous and highly accurate humoral and neural signals from adipose tissue, stomach, intestine and pancreas regarding body energy stores and acute energy intake respectively. Upon weight loss, these messengers change and 12 signal depletion of body energy stores which is disadvantageous from an evolutionary perspective. The final common pathway of this mechanism is the defence of the individual's body weight setpoint through an increase in hunger and reduction in satiety which trigger the executive function areas located in the cortical areas of the brain to seek and consume food.
A good example of how this system is activated is intentional weight loss through caloric restrictive diets. People on severe caloric restriction frequently report a decrease in hunger and increase in satiety during the acute phase of negative energy balance. However, the vast majority find it difficult to maintain the weight they have lost when it plateaus during the stable energy balance phase. This is despite the cortical areas of the brain that control dietary restraint working intensely to maintain the body weight lost. The increase in hunger and decrease in satiety signalled by the hypothalamus/ brainstem results in an increase in caloric intake which eventually leads to the regain of weight lost and in many cases the establishment of a new setpoint which is higher than the original baseline 12 .
Repeated cycles of this process increase body weight setpoint, making it progressively harder to achieve sustained weight loss 13 . Consequently, any successful weight loss and maintenance therapy should be sophisticated enough, A c c e p t e d M a n u s c r i p t from a biological perspective, to counteract this elaborate body weight regulation system.
Obesity surgery has proven to be biologically very sophisticated and is thus an effective therapy. Similar to caloric restriction during the acute negative balance phase, patients after surgery report a decrease in hunger and increase in satiety 14 .
The key difference between dieting and obesity surgery is that after surgery, the body weight setpoint is reduced by approximately 20-30% 15 . Manipulation of the stomach and the small intestine result in favourable changes in humoral and neural signals from the gut to the brain that are conducive to the maintenance of this newly established body weight setpoint.
The comparison of patients' reports and actual weight during the plateau phase of weight loss during dieting vs. obesity surgery is intriguing. Even after surgery, patients report an "alarming" increase in hunger and decrease in satiety during the stable energy balance phase and indeed this translates in both higher energy intake during meals and an increase in meal frequency 16 . Yet, body weight increases only marginally and never reaches the pre-operative value in the majority of cases. Whilst at this new set point, the intensity of the internal feelings of hunger and satiety might return to almost pre-operative levels, altered signalling from the gut acts continuously to reduce total energy intake during a 24-hour period in order to robustly defend the new normal 12 .
Patients losing weight through pharmacotherapy (e.g. with glucagon-like peptide 1 (GLP-1) receptor agonists) report very similar changes in their appetite during the acute and chronic phase of their weight loss journey 17 . The only difference is that A c c e p t e d M a n u s c r i p t the effect size of pharmacotherapy is lower than that of surgery, and that is because medications change only one or few of the signalling pathways in the appetite centres of the brain.
Weight loss after the biliopancreatic diversion further highlights that the mechanisms through which these operations work are physiological and not "cognitive" in nature.
This procedure is the most effective operation for weight loss, but rarely performed these days due to the associated severe nutritional complications. The very long intestinal bypass in this procedure results in frank macronutrient malabsorption and weight loss. The brain appetite centres rapidly detect this and compensate by increasing hunger. Patients after the biliopancreatic diversion commonly consume more calories compared to before their operation. However, even this hyperphagia is not enough to compensate for the severe loss of calories through the gut which is therefore the dominant mechanism causing weight loss.

Neural correlates of reduction in energy intake
The hypothalamus is a critical brain area that controls energy intake and expenditure via two sets of antagonistic neurons: agouti-related peptide (AgRP) neurons to promote hunger and pro-opiomelanocortin (POMC) neurons to promote satiety 18 ( Figure 1 A c c e p t e d M a n u s c r i p t These findings from animal models support the observations from humans in that the direction of change in expression of neuropeptides in the hypothalamus and brainstem after RYGB and VSG is opposite to dieting and favour the maintenance of a lower body weight set point.

Neural signalling
The mechanism of action of AGB is thought to be exclusively through vagal signalling. Injection of fluid through the subcutaneous port increases the extraluminal pressure on vagal afferents, sending an anorexigenic signal to the brainstem, even in the fasting state 30 . This mechanism is further exaggerated through the increase in fundal intra-luminal pressure exerted by the consumption food, leading to early satiety during a meal. It is common for healthcare professionals to inject progressively more fluid in the AGB in patients not losing enough weight. This eventually leads to mechanical restriction and vomiting. This is a preventable complication that should be avoided, and instead an early decision should be made to remove the AGB in patients who do not respond. More patients do not respond to the AGB compared to RYGB/VSG 31 because the AGB activates only one signalling system to the brain, as opposed to the plethora of anorexigenic signals after RYGB/VSG. A study in rats suggested that signals carried by vagal afferents from the mid and lower small intestine contribute to the early RYGB-induced body weight loss and reduction of food intake 32 . Disruption of vagal afferents and/or efferents takes place during RYGB and VSG surgery; whether this affects appetite and postoperative weight loss remains unclear. Some studies suggested that vagal sparing surgical technique affects body weight loss in rodents, and therefore the vagal nerve A c c e p t e d M a n u s c r i p t should be preserved during the gastric bypass operation 33,34 . However, there are limited data on the role of vagus nerve dissection in RYGB and VSG with regards to body weight in humans 35 .

ii. Food selection
After RYGB and VSG surgery, but not AGB, some patients also change their food selection 36 . This includes a shift in preference from energy-dense sweet and fatty food to less energy-dense options. The majority of research in this area has used indirect measures of behaviour, e.g. questionnaires, food diaries and verbal report at recall sessions. Whilst these have suggested that the reduction of the consumption of energy dense food might be an additional weight loss mechanism after surgery, they have also demonstrated large variations in response and substantial heterogeneity in findings 37 . This is particularly noticeable in the longer-term measurements of eating behaviour, 5-10 years after surgery when any early changes in macronutrient selection tend to dissipate.
Only a small number of studies have used direct measurements of eating behaviour, i.e., observing the participant's choices during an ad libitum meal or an eating behaviour task. The best evidence so far suggests that patients who lost more weight were those who consumed a lower percentage of fat and low-glycaemic index foods, and higher percentage of protein as a proportion of their total daily caloric intake 38 .
A c c e p t e d M a n u s c r i p t The reduction in the rewarding properties of food is one of the mechanisms that underpins the changes in food selection ( Figure 2). This mechanisms has been investigated using functional neuroimaging. Functional Magnetic resonance imaging (MRI) and Positron Emission Tomography (PET) studies provide information both with respect to the direction of change and the areas of the brain reward system that correlate with changes in observed or reported eating behaviour. Notwithstanding discrepancies between studies, there is some agreement that there is a reduction in the activation of brain areas that respond to the involved cues with rewarding properties (e.g. food pictures) after RYGB and VSG 39, 40 . The effect size of this reduction is more pronounced after RYGB compared to VSG 41 . Gut hormones are mediators that underlie this observation, as the blockage partly reverses the reduction in activation of these brain regions 42 . This is in line with animal and human data demonstrating that gut hormones such as glucagon-like peptide-1 (GLP-1) and peptide YY (PYY) do not just reduce hunger and increase fullness, but reduce the rewarding properties of food through their direct action on their receptors in brain reward areas 43 . It should be noted that functional neuroimaging findings should be interpreted with some caution as they only measure neural correlates of eating behaviour, not behaviour itself. The available paradigms also do not allow enough granularity as to whether measured brain responses to food pictures reflect appetitive or consummatory behaviour.
Altered taste function is another mechanism underlying the changes in food selection after RYGB and VSG. With regards to the sensory domain of taste, acuity for sweet taste is heightened only in the early post-operative period. 44  Neural signalling also contributes to changes in the rewarding value of fat and sugar after surgery. This was investigated in obese rats undergoing RYGB as they were found produce less of the fat-satiety molecule oleoylethanolamide in the small intestine, and this effect was associated with vagus nerve-driven increases in dorsal striatal dopamine release 53 . Inhibition of local oleoylethanolamide, vagal, and dorsal striatal dopamine-1 receptor signalling was inhibited, the beneficial effects of RYGB on fat intake and preferences was reversed.
Post-ingestive neural signalling, in the form of what is widely known as dumping syndrome, may contribute to the underlying reductions in high-glycaemic index or fatty food after RYGB, and less so after VSG. Patients report unpleasant sensations A c c e p t e d M a n u s c r i p t of nausea, sweating and dizziness early after consumption of sugary or fatty foods, which in some people may result in conditioned taste avoidance 54 . During this learning process, these unpleasant symptoms are presumably generated through osmotic shifts between the intestine and circulation, and altered neural signalling.
These symptoms are usually associated with the ingestion of specific foods. This does not lead to the complete extinction of these foods from regular consumption, i.e., aversion, but rather their avoidance. Thus the foods remain pleasant to the subject but only when consumed in smaller quantities 54 . It should be noted that dumping syndrome is not present in all patients after RYGB and it may indeed be the case that its impact dissipates over time. This might be due to intestinal adaptation that continues to take place for years after surgery. Dumping is less common after VSG and AGB 55 , operations not involving duodenal bypass.
Overall, the available data suggest that changes in food selection take place in a proportion of people after RYGB and VSG, but not after AGB. In the former, this mechanism could compliment the reduction in hunger and increase in satiety to cause additional weight loss. Whether this mechanism persists over time or dissipates following intestinal adaptation remains uncertain. The process of learning to avoid foods that generate unpleasant post-ingestive effects has a greater impact than taste function in shaping food preferences after surgery. Some of the above unresolved questions could be answered using residential stays in facilities that allow human eating behaviour to be as close to normal as possible. Such experiments could be conducted both early and late after surgery and complimented by studies in animal models of surgery.
A c c e p t e d M a n u s c r i p t

b. Energy Expenditure
Enhanced energy expenditure after obesity surgery may be a contributing mechanism to weight loss. Resting energy expenditure has been measured in humans following RYGB, and most recent studies using indirect calorimetry show resting energy expenditure to either decrease within the first post-operative year [56][57][58] , remain stable 59 or even slightly increase 60 . These changes are reported to be highly dependent on organ-tissue body composition as RYGB patients maintain a larger high-metabolic rate organ mass than non-operated controls 59 . Moreover, the acute weight loss following obesity surgery was found to affect the accuracy of energy expenditure predictive equations 61 .
A small number of studies used 24-hour indirect calorimetry, a method that is optimal for measuring substrate oxidation because each subject can freely move, consume meals, and engage in physical activity. One study reported that diet-induced energy expenditure in patients 20 months after RYGB was increased, which resulted in an increased contribution to total energy expenditure over 24 hours from an average 12.9 cal/min/kg to 14.7 cal/min/kg, when corrected for total tissue mass, including total adipose tissue, lean body mass, bone mineral density and bone mineral content 62 . Another study reported no changes in 24-hour or diet-induced energy expenditure 11 weeks after RYGB, although this was not corrected for total tissue mass 63     is elevated in people with obesity 103 , and obese mice are insensitive to exogenous FGF21 administration, suggesting that obesity is an FGF21-resistant state 104 .
However, FGF21 sensitivity is restored in humans following weight loss 105 . Although not directly correlated with obesity, FGF21 variants are associated with increased sweet consumption, as plasma FGF21 levels increase acutely after oral sucrose ingestion. This indicates that FGF21 could influence eating behaviour 106 .
Total bile acids and FGF19 increase after RYGB and VSG in humans and rodents 107 . Specifically, glycine-conjugated serum bile acids increased acutely following RYGB in humans, while both conjugated and unconjugated bile acids increased after VSG, an effect not replicated in an unoperated calorie-restriction control group 108,109 . The bile acid increase is sustained five years after surgery, with higher levels associated with greater weight loss, and lower total cholesterol 110

c. Gut microbiota
Gut microbiota have a vital role in both energy harvesting and energy expenditure.
They can metabolize indigestible complex carbohydrates by fermentation, leading to the production of short-chain fatty acids, as well as control the absorption of nutrients 127,128 . Gut microbiota also play a role in the thermogenic capacity of brown adipose The mechanism via which obesity surgery achieves weight loss may include changes in gut microbiota. Transfer of gut microbiota from RYGB-treated mice to non-operated, germ-free mice resulted in weight loss and decreased fat mass in the recipient animals when compared to recipients of microbiota induced by sham surgery 135 . In rats transplanted with the RYGB-microbiota, this decrease in adiposity and body weight was not associated with a change in food intake, further suggesting that the RYGB-associated gut microbiota either increase energy expenditure or have reduced ability to harvest energy from nutrients 135  A c c e p t e d M a n u s c r i p t Although human studies 137,138 have reported differences in gut microbiota postoperatively, the extent of these changes varies. This could be due to patient inclusion criteria, such as glycemia state and medication, but also diet, and type of procedure. However, studies in humans consistently demonstrate an increase in gut microbiota diversity, spatial organization and stability, and specifically Proteobacteria, after RYGB (Table 1). Gut microbiota diversity is a measure of how many different species exist and how evenly distributed they are in the gut community, and low diversity is a sign of dysbiosis 139 . Some studies also reported a decrease in Firmicutes and Bacteroidetes phyla in humans and rats post RYGB 136,140,141 . Increase in gut microbiota diversity, stability and resilience is important, as a large number of associations between gut microbiota and adipose tissue gene regulation as early as three months after surgery 136,140,142 have been reported, further demonstrating that gut microbiota may play a direct role in the control of adiposity by regulating lipid metabolism. Moreover, gut microbiota lead to the production of short-chain fatty acids, which stimulate GLP-1 secretion via free fatty acid receptor-2 (FFAR2), and therefore may also reduce energy intake 143 .
A decrease in Proteobacteria was recorded in patients following VSG 144 and AGB 145 . This differential effect between VSG and RYGB could be a result of duodenal exclusion in RYGB, as duodenal-jejunal bypass with minimal gastric resection in obese rats increased microbial richness and abundancy when compared to rats treated with GLP-1R agonists 146,147 . This has also been observed in humans following treatment with the endoscopic duodenal-jejunal bypass liner 148 149 .
Comparison of AGB, pharmacologically induced weight loss and RYGB demonstrated that at similar weight loss, the greatest alteration in gut microbiota diversity occurred after RYGB 145,150 . A c c e p t e d M a n u s c r i p t Despite the positive effect on weight loss through a combination of mechanisms discussed above, RYGB is unable to fully reverse the decreased gut microbial gene richness and compositional changes observed in patients with obesity 151 .
Interventions such as faecal transplantation from lean donors to patients with obesity revealed that weight-lowering beneficial effects are linked to changes in plasma metabolites and driven by baseline faecal microbiota composition 152 . Moreover, gut microbiota diversity alteration accelerates post-dieting weight regain 153 , suggesting that microbiome-targeting approaches may help enhance weight loss after surgery or prevent weight regain.

Genetics and Obesity Surgery
Patient selection for surgery ("personalized medicine") may provide an additional refinement for existing procedures and could lead to the identification of genes or pathways which might provide useful therapeutic targets. Candidate gene studies have explored roles for the melacocortin-4 receptor (MC4R) 154 , revealing greater weight loss in patients whose obesity is in part driven by mutations in this gene. A more recent genome-wide association study (GWAS) 155 has reported 17 single nucleotide polymorphisms (SNPs) in weight loss two years post RYGB, implicating roles for the 5-hydroxytriptamine receptor 1A and other genes. Whether the strength and number of these associations is substantial enough to provide predictive power is unclear.
A c c e p t e d M a n u s c r i p t

Conclusion
The anatomical manipulations during the most frequently used obesity surgery procedures cause weight loss through changes in the biology of the gut. Altered signalling from the gut to the brain, the organ responsible for the disease of obesity, facilitate reductions in energy intake and in some people changes in food selection.
Increased or unaltered energy expenditure in the context of weight loss may also contribute to the defence of a new body weight set point. The precise mechanisms underlying these profound changes are not completely understood. Unravelling of the elusive physiology of the gut after surgery will help optimise surgical procedures, develop non-surgical therapies, address weight regain after surgery, but also understand the pathophysiology of the disease of obesity itself.  Graphical Abstract: Representation of the main physiological mechanisms underlying weight loss following Vertical Sleeve Gastrectomy (VSG) and Roux-n-Y Gastric Bypass (RYGB). Abbreviations: GLP-1, glucagon like peptide-1; PYY, peptide YY.