5-HT 2 A and 5-HT 2 C receptors as hypothalamic targets of developmental programming in male rats

Although obesity is a global epidemic, the physiological mechanisms involved are not well understood. Recent advances reveal that susceptibility to obesity can be programmed by maternal and neonatal nutrition. Specifically, a maternal low-protein diet during pregnancy causes decreased intrauterine growth, rapid postnatal catch-up growth and an increased risk for diet-induced obesity. Given that the synthesis of the neurotransmitter 5-hydroxytryptamine (5-HT) is nutritionally regulated and 5-HT is a trophic factor, we hypothesised that maternal diet influences fetal 5-HT exposure, which then influences development of the central appetite network and the subsequent efficacy of 5-HT to control energy balance in later life. Consistent with our hypothesis, pregnant rats fed a low-protein diet exhibited elevated serum levels of 5-HT, which was also evident in the placenta and fetal brains at embryonic day 16.5. This increase was associated with reduced levels of 5-HT2CR, the primary 5-HT receptor influencing appetite, in the fetal, neonatal and adult hypothalamus. As expected, a reduction of 5-HT2CR was associated with impaired sensitivity to 5-HT-mediated appetite suppression in adulthood. 5-HT primarily achieves effects on appetite by 5-HT2CR stimulation of proopiomelanocortin (POMC) peptides within the arcuate nucleus of the hypothalamus (ARC). We show that 5-HT2ARs are also anatomically positioned to influence the activity of ARC POMC neurons and that mRNA encoding 5-HT2AR is increased in the hypothalamus of in utero growth-restricted offspring that underwent rapid postnatal catch-up growth. Furthermore, these animals at 3 months of age are more sensitive to appetite suppression induced by 5-HT2AR agonists. These findings not only reveal a 5-HT-mediated mechanism underlying the programming of susceptibility to obesity, but also provide a promising means to correct it, by treatment with a 5-HT2AR agonist.


Summary Statement
Maternal protein restriction increases fetal brain 5-HT and may contribute to changes in expression and function of hypothalamic 5-HT2C and 5-HT2A receptors in the offspring later in life.

Introduction
Nutrition during critical periods of development in early life can exert profound, long-term effects on susceptibility to obesity. For example, men exposed to the Dutch Hunger Winter in utero during early gestation had an increased risk of developing obesity as adults, while obesity rates were reduced amongst those exposed to famine during the last trimester of gestation and in early postnatal life (Ravelli et al., 1976). The early postnatal diet is also important. In randomized trials, full-term, low birth weight infants fed a growth-promoting nutrient-enriched formula had a higher fat mass at the age of 5-8 years than those fed standard formula (Singhal et al., 2010). Studies in rodents, showing that low birth weight followed by rapid postnatal growth is associated with increased adiposity, support findings in humans (Plagemann et al., 1992;Cottrell et al., 2011;Berends et al., 2013). Despite these robust associations, the molecular mechanisms mediating the interaction between early nutrition and obesity risk are still largely unknown.
It is widely acknowledged that the hypothalamus, which in humans develops primarily prenatally but in rodents postnatally, plays an important role in the programming of body weight (Grove et al., 2005;Horvath and Bruning, 2006;Glavas et al., 2007). To date major efforts have been directed at understanding the roles of leptin and insulin in this process (Bouret et al, 2010;Yura et al., 2005;Steculorum and Bouret, 2011). However, studies in our laboratory using the leptindeficient ob/ob mice demonstrated that leptin-independent mechanisms are likely to also program body weight (Cottrell et al., 2011).
The importance of the 5-hydroxytryptamine (5-HT; serotonin) system in the control of food intake and body weight has been recognised for many years and alterations in central serotonergic activity have been observed in obese humans, non-human primates and rodents (Mori et al., 1999;De Fanti et al., 2001;Sullivan et al., 2010). 5-HT, a potent anorectic signal, influences food intake in the mature brain by acting predominantly via 5-HT2CR to regulate the key energy balance mediator, pro-opiomelanocortin (POMC), within the arcuate nucleus of the hypothalamus (ARC) (Doslikova et al., 2013;Burke et al., 2014). Indeed, genetic disruption of the 5-HT2CR specifically expressed on POMC neurons programs hyperphagia and obesity when animals are fed a high-fat diet (Berglund et al., 2013). In addition to playing an important role in energy balance and body weight, 5-HT acts as a trophic factor during fetal brain development in rodents, regulating cell migration, proliferation, maturation and axonal growth (Whitaker-Azmitia et al., 1996;Gaspar et al., 2003). The long term effects of disturbances in the 5-HT system during early life on obesity risk are, however, not fully defined.
Here we investigated 5-HT system function and used an unbiased genome-wide profiling approach to identify differentially expressed targets in the hypothalamus of intrauterine-restricted rats that underwent postnatal catch-up growth (recuperated animals) in an effort to elucidate the molecular mechanisms mediating the effects of early nutrition on energy balance.

Growth trajectories, body weights and brain weights
At birth, offspring of mothers fed a low protein diet during pregnancy were significantly smaller than the control pups (P<0.001) ( Table 1A). The body weight of the recuperated pups caught up with that of control pups by day 14 and they weighed the same as controls at weaning (Table 1A).
Fractional growth rate did not differ between recuperated and control pups between birth and day 7 of life (0.16 ± 0.1 vs 0.17 ± 0.01); however recuperated offspring had significantly higher fractional growth rate between birth and days 14 and 21 when compared to controls (day 14: 0.31 ± 0.01 vs 0.24 ± 0.01, P<0.001; day 21: 0.38 ± 0.01 vs 0.29 ± 0.01, P<0.001). At 3 months of age body weights were similar between recuperated and control offspring. Brain weights were similar between the experimental animals and their age-matched controls at both 22 days and 3 months of age, in both absolute terms and relative to body weight (Table 1B). These findings demonstrate that maternal low protein diet decreases in utero growth of the offspring so that they have a lower birth weight.
These animals then undergo "catch up" growth such that by 22 days of age, they are the same body weight as the control offspring.

Maternal, placental and amniotic fluid 5-HT and tryptophan levels
Tryptophan is an essential amino acid and is required for 5-HT synthesis. A low protein diet would therefore be expected to yield low plasma tryptophan levels. We analyzed the levels of 5-HT and tryptophan in the dams, placenta and amniotic fluid to investigate whether 5-HT may be a programming factor that mediates changes in appetite and susceptibility to diet-induced obesity in recuperated animals. As expected, we observed reduced levels of tryptophan in the serum of low protein fed dams at E16.5 of pregnancy compared to control dams (158 ± 17 mmol.l -1 vs 220 ± 20 mmol.l -1 , P<0.05). Paradoxically, the serum 5-HT level was raised in low protein diet fed dams at this time-point (P<0.05; Figure 1A). A similar pattern was observed for tryptophan and 5-HT concentration in in the amniotic fluid at E16.5, with tryptophan levels being reduced in the amniotic fluid from low protein pregnancies (26.3 ± 3.2 mmol.l -1 vs 40.6 ± 3.4 mmol.l -1 , P<0.01, n=10-11 per group), whilst 5-HT levels were raised (65.0 ± 6.6 nmol.l -1 vs 48.2 ± 3.9 nmol.l -1 , P<0.05, n=10-11 per group). 5-HT levels were also significantly increased in E16.5 placentas from low protein pregnancies when compared to control pregnancies (P<0.01; Figure 1A). Furthermore, we observed a negative correlation between placental 5-HT level and maternal serum tryptophan (P<0.05; Figure 1B). There was a positive correlation between placental 5-HT and fetal brain 5-HT levels ( Figure 1C). These findings reveal that in response to significantly reduced dietary tryptophan, 5-HT is over-produced in the mother and in the intrauterine environment in which the fetus grows and develops.

Fetal, Neonatal, Weaning and Adult tryptophan and 5-HT levels
Likewise, in the neonatal brain, a decrease in tryptophan levels (67.9 ± 13.6 mmol.l -1 vs 140 ± 14.7 mmol.l -1 , P<0.01) was associated with an increase in 5-HT levels in the hypothalami of recuperated pups at birth (P<0.01; Figure 1A). 5-HT levels were also increased in the brains of E16.5 fetuses from low protein pregnancies ( Figure 1A). By weaning as well as at 3-months of age, hypothalamic 5-HT levels were normalized to control levels (P21: Recuperated 1347 ± 110 mmol.l -1 vs Control 1659 ± 131 pmol.g -1 n= 12 per group; 3 months of age: Recuperated 1414 ± 154 mmol.l -1 vs Control 1354 ± 85 pmol.g -1 n= 10 per group). These data indicate that the over-production of 5-HT in response to maternal low protein diet is normalized by weaning.

Effect of D-fenfluramine on food intake
We next assessed the significance of maternal diet-stimulated increased 5-HT during early development on the function of the adult 5-HT system. D-fenfluramine is a pharmacological probe of the endogenous 5-HT system that works by stimulating endogenous 5-HT release and blocking its reuptake. D-fenfluramine was prescribed until the 1990s for human obesity treatment since increasing endogenous 5-HT bioavailability reduces appetite and body weight (Lam et al., 2010;Marston et al., 2011). We administered D-fenfluramine into the third ventricle of rats at 3 months of age and observed the expected reduction in food intake in controls ( Figure 2). However, recuperated offspring had impaired sensitivity to 5-HT-stimulated (D-fenfluramine) food intake reduction (P<0.05) ( Figure 2). These results show that a low protein maternal diet followed by accelerated postnatal growth diminishes 5-HT mediated control of food intake in the adult offspring. Given the strong inverse correlation between 5-HT and appetite, these findings lead us to hypothesise that deregulated 5-HT system may contribute to the programmed increased food intake in these offspring when exposed to an obesogenic diet.

Fetal, neonatal and postnatal hypothalamic 5-HT2CR expression
5-HT is a neurotransmitter that communicates appetite-related signals primarily through the 5-HT2CR within the ARC. If 5-HT's effects on appetite are diminished in recuperated rats, this suggests a perturbation in signalling at the 5-HT2CR. We next probed the consequence of elevated 5-HT during development on the expression of the 5-HT2CR. Analysis of E16.5 fetal brain 5-HT2CR mRNA expression showed significantly reduced 5-HT2CR in the whole heads of fetuses from low protein pregnancies (P<0.05; Figure 3A). At birth 5-HT2CR was still reduced in the hypothalamus of the growth restricted pups at the mRNA (P<0.001; Figure 3A) as well as at the protein (P<0.001; Figure   3B,C) level. However, by 3 months of age no significant differences in 5-HT2CR mRNA levels were observed between the experimental groups ( Figure 3A). Protein expression was, however, still significantly reduced in the hypothalamus of recuperated animals at 3-months of age (P<0.05; Figure   Disease Models & Mechanisms • DMM • Advance article 3B,C). Given that perturbations of ARC 5-HT2CR function increases appetite and body weight (Berglund et al., 2013), these results thereby reveal a potential mechanism through which rapid postnatal growth is achieved in recuperated rats, which also could potentially contribute to increased susceptibility to diet-induced obesity in these animals.

Microarray analysis of LCM ARC from control and recuperated 3-month-old offspring
5-HT is a trophic factor during fetal brain development influencing both 5-HT circuits and the anatomical organisation of other systems (Whitaker-Azmitia et al., 1996;Gaspar et al., 2003). The ARC is a crucial regulator of energy balance. Having demonstrated that maternal diet induces changes in offspring 5-HT and disrupts 5-HT appetite system function, we next probed whether 5-HT disrupts other ARC genes and circuits regulating appetite. To achieve this, we performed LCM of the ARC of 3-month-old control and recuperated rats and subjected samples to microarray.

Validation of laser-capture micro dissection (LCM)
To ensure the specificity of the ARC dissection within the LCM material ( Figure 4), we analyzed the expression of two genes that are not expressed in the ARC but are expressed in the neighbouring hypothalamic nuclei. Neither PVN-specific Brn1 (Pou3f3) nor the ventromedial nucleus of the hypothalamus (VMN) specific Sf-1 were detected (data not shown). In addition, we examined the expression of four key genes involved in the regulation of energy balance in the ARC: POMC, AgRP, CART and NPY. We confirmed using Taqman RT-PCR that all of these genes were highly expressed in the dissected material (data not shown).

Analysis of microarrays using three different algorithms and pathway analysis
Following global transcriptional profiling, we analyzed the data using three different algorithms (1) GeneChip Operating Software (GCOS), (2) Robust Multi-array Averaging (RMA) and (3) GeneChip Robust Multi-array Averaging (GCRMA) to ensure maximum stringency and to reduce the number of false positives. The combined analysis, using a 1.3-fold cut off threshold and P<0.05, revealed that out of 31,099 genes analyzed, expression of 15951 genes was detected in the ARC. Of these 142 genes were up-regulated in the ARC of the recuperated animals, while 59 genes were down-regulated ( Figure 4A and 4B respectively). The top 25 up-and down-regulated genes in recuperated offspring are presented in Tables 2A and 2B. Htr2a, which codes for 5-HT2AR, was the gene with the highest (2.55-fold) increase. We further analyzed the microarray data using Ingenuity Pathway Analysis software to detect groups of functionally-related genes. The top three gene functions affected by early nutrition in 3-month-old male recuperated rats were cell cycle, connective tissue development and function, cellular growth and proliferation (Table 2C).

Validation of genes identified by microarray
Out of the seven genes chosen for validation, two genes had very low expression (Plau and Pde4a). Out of five genes remaining, differential expression of three genes (5-HT2AR, Khsrp and Retsat) was confirmed by RT-PCR, whilst two genes were not validated (Car8 and EEF2k) ( Figure 5A).
As the 5-HT2AR had the highest fold change in the microarray analysis (Table 2A), it was further investigated. The up-regulation of the 5-HT2AR in the ARC of 3-month-old recuperated animals (P<0.05) observed using RT-PCR was confirmed by in situ hybridization (P<0.05; Figure 5B).

Timing and location of increased hypothalamic 5-HT2AR expression
A similar pattern of 5-HT2AR expression to that seen in 3-month-old animals was observed at the age of 22 days. 5-HT2AR was significantly elevated in the ARC of recuperated rats at weaning (P<0.05; Figure 5C). The expression of 5-HT2AR was, however, not altered in neonatal ARC ( Figure 5C) or in E16.5 fetal brains (Low protein 0.933+0.035; Control 1.002+0.022). As in the ARC the expression of 5-HT2AR was also significantly increased in VMN of 3-month-old recuperated animals (P<0.05; Figure 5 D,E). However, in contrast to the ARC, no difference was seen in the 5-HT2AR expression within VMN between the two experimental groups at day 22 of life ( Figure 5E) suggesting that upregulation of 5-HT2AR within the VMN in the recuperated group is secondary to the up-regulation of this receptor in the ARC. Overall these findings reveal a change in 5-HT2AR receptor expression using an unbiased genome-wide profiling approach and thereby support the importance of maternal diet induced changes in the 5-HT system within the hypothalamus in disrupted offspring energy balance.

Neuronal co-localization of ARC 5-HT2AR and POMC
Within the ARC, 5-HT primarily influences energy balance via action at 5-HT2CR stimulating the activity of POMC neurons (Heisler et al., 2002;Xu et al., 2008;Berglund et al., 2013;Burke et al., 2014). Given that 5-HT2CR expression is reduced in rats exposed to a maternal low protein diet, we hypothesized that 5-HT2AR may therefore be up-regulated to compensate for this in an effort to normalise energy balance. To determine whether 5-HT2ARs are anatomically positioned to influence the activity of POMC neurons, we visualized co-expression using dual histochemical labelling. We determined that approximately 40% of the ARC POMC neurons express 5-HT2ARs in control rats ( Figure 5F). These data reveal a defined population of neurons within the ARC critically involved in body weight regulation that could be influenced by Gq-protein coupled 5-HT2ARs.

Disease Models & Mechanisms • DMM • Advance article
in adult 3-month-old recuperated and control offspring. We confirmed previous results showing that 5-HT2AR agonism significantly suppresses food intake (Fox et al., 2010). TCB2 also significantly decreased food intake in the recuperated offspring as shown by two-way ANOVA analysis (overall effect of TCB2 administration, P<0.001; Figure 6). However, the dose response curves between the control and recuperated groups were significantly different (P<0.05; Figure 6). Recuperated offspring were more sensitive than the controls to the centrally administered action of 5-HT2AR agonist TCB-2 (overall effect of early nutrition, P<0.05; Figure 6). These results reveal that over-expressed 5-HT2ARs in recuperated rats are functional and when pharmacologically stimulated, a greater net effect on appetite is achieved. Extrapolating these findings to endogenous 5-HT activity, as indicated by the D-fenfluramine results, suggests that the up-regulation of 5-HT2AR is insufficient to compensate for 5-HT2CR down-regulation in the 5-HT regulation of appetite.

Discussion
Evidence from epidemiological studies and animal models shows that suboptimal early nutrition affects susceptibility to obesity in later life. We and others have shown that rodents exposed to maternal protein restriction in utero and accelerated postnatal growth (recuperated offspring) are more susceptible to development of diet-induced obesity (Ozanne et al., 2004;Bieswal et al., 2006). In the current study we focused on 3-month-old recuperated rats fed a chow diet that are a similar weight to control offspring. This enabled us to investigate mechanisms mediating the effects of early nutrition on susceptibility to developing diet-induced obesity without the confounding effects of obesity itself. In recent years, leptin has been studied as a key factor involved in programming of obesity risk (Cripps et al., 2009;Bouret, 2004). However, we previously showed that programming of increased adiposity is, at least in part, leptin-independent (Cottrell et al., 2011).
In the current study, we investigated the effect of a maternal low protein diet during gestation on a factor directly impacted by reduced dietary protein, 5-HT. We observed that 5-HT levels were increased in the fetal brains and neonatal hypothalami of rats exposed to a low protein diet in utero.
As 5-HT is known as an essential growth and regulatory factor that drives the development and maturation of its own cellular network and related neuronal systems (Pino et al., 2004;Oberlander, 2012), an excess of 5-HT may lead to impaired interneuron migration and abnormal development of the serotonergic system itself and its target regions (Pino et al., 2004;Oberlander, 2012). Here we suggest that elevated hypothalamic levels of 5-HT in recuperated animals in early life cause developmental and functional alterations that increase susceptibility to obesity later in life.
We observed reduced maternal circulating tryptophan but increased 5-HT in pregnant dams fed a low-protein diet. This could be a consequence of (1) increased carbohydrate content in lowprotein diet, added to balance protein reduction to obtain an isocaloric diet and (2) the dams' increased food intake in both absolute and relative terms (Fernandez-Twinn et al., 2003). Such dietary intervention can affect maternal insulin levels. Low-protein dams were shown to be hyperinsulinaemic at day 14 of pregnancy (Fernandez-Twinn et al., 2003). As circulating tryptophan levels decrease in an insulin-dose dependent manner, increased insulin concentration in these dams could contribute to the reduced tryptophan levels (Fernandez-Twinn et al., 2003;Fukagawa et al., 1987). Another possibility is that the dietary intervention has altered the gastrointestinal microbiome of the dam. Diet is the key regulator of both taxa and microbiota production of metabolites (Scott et al., 2012). More than 90% of 5-HT in the body is synthesized in the gut and microbiota has been shown to play a critical role in regulating host 5-HT biosynthesis and 5-HT blood levels (Yano et al., 2015).
Several mechanisms may be involved in mediating the relationship between reduced maternal circulating tryptophan and increased 5-HT in the brains of their unborn offspring. In adult rats food deprivation leads to an increase in the concentration of circulating free fraction of tryptophan (FFT), the fraction that passes through the blood-brain-barrier (Knott and Curzon., 1972; Perez-Cruet et al., tryptophan-5-hydroxylase and synthesis of 5-HT have also been reported in intra-uterine growth restricted (IUGR) rat pups and infants (Kalyanasundaram, 1976;Hernandez et al., 1989;Manjerrez et al., 1996;Manjerrez et al., 1998). Up-regulation of FFT in IUGR was attributed to abnormal kinetics of tryptophan binding to albumin (Hernandez-Rodriguez et al., 2009). Reduced plasma levels of leucine, isoleucine and tyrosine which compete with tryptophan for entry into brain, may also contribute to an increase in transport of tryptophan across the blood-brain-barrier in IUGR pups (Roux and Jahchan, 1974).
In recent years, in both mice and humans, the placenta has emerged as an important source of 5-HT, that can directly impact on the fetal brain development . In our study the impact of maternal LP diet on 5-HT was more pronounced in the placenta and fetal brain than in the dam. This supports the proposal that, the placenta is cannibalized in response to maternal food deprivation to provide nutrients and substrates for the developing hypothalamus (Broad and Keverne, 2011). We observed an association between placental 5-HT and fetal brain 5-HT at E16.5, the time-point at which 5-HT fibres can be detected in the hypothalamus and both exogenous and endogenous sources of 5-HT contribute to fetal 5-HT synthesis (Aitken and Tork, 1988;. Increased 5-HT production within the placenta may contribute to raised 5-HT levels in fetal brains from low protein fed dams. Maternal tryptophan levels have also been reported to influence central 5-HT levels in the offspring and maternal peripheral 5-HT proved to be important for offspring neurodevelopment Cote et al., 2007). This suggests that both maternal and placental tryptophan/5-HT metabolism may be an important molecular pathway for the fetal programming of the 5-HT system and brain development.
Changes in 5-HT bioavailability not only impact on hypothalamic circuitry formation during development but also have been shown to influence appetite and body weight in the adult (Porto et al., 2009;Berg et al., 2013;Madden and Zup 2014). 5-HT primarily regulates energy balance via activity at the Gq-coupled receptor, 5-HT2CR. Here we report that elevated fetal 5-HT could potentially program decreased hypothalamic 5-HT2CR expression, although direct evidence for this relationship remains to be established. The reduction in 5-HT2CR may be an early and permanent developmental consequence of central hyperserotonemia. Reduced 5-HT2CR function is known to promote increased appetite and obesity. Specifically, genetic inactivation of Htr2c that codes for 5-HT2CRs, leads to lifelong hyperphagia and elevated body weight in mice fed standard chow diet (Tecott et al, 1995), while pharmacological activation of these receptors reduces food intake and decreases body weight (Heisler et al., 2002;Halford and Harrold, 2012;Burke et al., 2014). 5-HT2CRs are involved in mediating the action of d-fenfluramine, which stimulates the release of 5-HT and inhibits the re-uptake of 5-HT into nerve terminals (Vickers et al., 2001;Trifunovic and Reilly, 2006;Xu et al., 2010). Moreover, the subpopulation of 5-HT2CR expressed solely with POMC is sufficient to mediate D-fenfluramine and 5-HT2CR agonist mCPP appetite suppression (Xu et al., 2008). Following the administration of D-fenfluramine into the third ventricle we observed impaired sensitivity to 5-HT-stimulated food reduction in 3-month-old recuperated animals. As the functional capacity of 5-HT2CR is dependent upon availability of active membrane receptor pools, the impaired response to D-fenfluramine in recuperated animals is likely a consequence of the reduced 5-HT2CR protein expression in the hypothalamus of these animals.
We used LCM of the ARC combined with microarray analysis to identify differentially expressed genes that could be affected by early hyperserotonemia. The expression of KHSRP, which is involved in the control of mRNA decay (Gherzi et al., 2006;Ruggiero et al., 2007) and plays a key role in the translation of DNA damage signalling to miRNA biogenesis (Trabucchi et al., 2009;Zhang et al., 2011) was increased in 3-month-old recuperated offspring. Although recuperated animals have reduced longevity (Jennings et al., 1999), which is associated with increased oxidative stress and an impaired response to DNA damage Tarry Adkins et al., 2009) the functional significance of the increase in KHSRP in the brain of these animals remains to be determined.

5-HT2AR
is another gene that was increased in the 3-month-old recuperated offspring. The precise role that 5-HT2AR plays in food intake regulation has not been fully defined. 5-HT2AR was upregulated in the ARC of diet-induced obese rats (Park et al., 1999), while in humans, 5-HT2AR correlated positively with BMI and a polymorphism in the gene encoding 5-HT2AR has been associated with obesity (Rosmond et al., 2002;Erritzoe et al., 2009;Carr et al., 2013). However, knockout of 5-HT2AR in mice does not alter food consumption or body weight gain when mice are fed laboratory chow (Weisstaub et al., 2006), suggesting the presence of developmental compensatory mechanisms or that body weight gain only occurs when these mice are placed on an obesogenic diet. There is little information regarding the precise neuronal location of 5-HT2AR within the ARC.
Here we report that 5-HT2ARs, like 5-HT2CRs (Heisler et al., 2002;Lam et al., 2008), are anatomically positioned to influence the activity of the critical energy balance regulator POMC. The extent to which the activity of POMC neurons can be influenced by action at 5-HT2ARs and whether this mechanism is altered in recuperated offspring as well as direct evidence for linking fetal hyperserotonemia to alteration of 5-HT2AR in later life remain to be established.
As 5-HT2AR expression was increased in the ARC of recuperated pups at weaning and in adulthood but not in brains of fetuses or neonates, the differences in 5-HT2AR must be established during the period of catch-up growth, between day 3 and day 22 of life, much later then the observed differences in 5-HT2CR levels. Therefore, up-regulation of 5-HT2AR may act as a secondary, counter-regulatory response to impaired 5-HT2CR signalling. Counter-regulatory mechanisms have been reported among 15 existing 5-HT receptor subtypes as genetic alteration of one specific 5-HT receptor subtype can result in compensatory signalling via another, for example signalling through 5-HT1BR in the absence of 5-HT2CR (Heisler and Tecott, 2000;Dalton et al., 2004). It is possible that counter-regulatory up-regulation of 5-HT2AR in response to impaired signalling via 5-HT2CR only emerges when growth-restricted newborn pups are born into and exposed postnatally to an environment of plentiful food. Some effects of early alterations in 5-HT therefore, may only appear in the presence of a particular postnatal environment. This has been shown to be true in a study of selective serotonin reuptake inhibitor-exposed 3-month-old infants, showing altered HPA stress response patterns which only became apparent when the method of infant feeding was taken into consideration (Oberlander et al., 2008).
5-HT2AR stimulation via agonist administration produced a greater effect on appetite in recuperated rats, revealing that the over-expressed ARC 5-HT2ARs are functional. Since pharmacological stimulation of endogenous 5-HT was less effective in suppressing appetite in recuperated rats, these results suggest that the endogenous circuitry is insufficient to appropriately modulate 5-HT-regulated appetite. However, this endogenous programming change may be circumvented and the up-regulation of the 5-HT2AR may be capitalized upon through pharmacological treatment with a 5-HT2AR non-hallucinogenic agonist. The anorectic therapeutic profile could be potentially further improved by combining the 5-HT2AR agonist, with 5-HT2CR and 5-HT1BR agonists as the co-application of 5-HT2CR and 5-HT1BR agonists produced a significant increase in the activity of POMC in the ARC (Doslikova et al., 2013). Development of a non-hallucinogenic 5-HT2AR agonist may seem to be challenging however, studies showing that the glutamate mGlu2 receptor heterocomplex with 5-HT2AR, and not 5-HT2AR on its own, acts as a molecular target for the actions of hallucinogenic drugs provide a plausible avenue to explore in the development of therapeutically suitable5-HT2AR agonist (Moreno et al., 2012).
Although we concentrated on studying mechanisms that potentially could mediate the effects of early elevated 5-HT levels on obesity risk in later life, early perturbations in the 5-HT neurotransmitter system are known to have major implications for mental health and behaviour in childhood and adulthood. Increased anxiety-like behaviour in nonhuman primate offspring of dams fed high fat diet was associated with perturbations in 5-HT system during fetal life (Sullivan et al., 2010). Early hyperserotonemia, in particular, has been identified as a potential factor in the pathogenesis of Autism Spectrum Disorder and Schizophrenia (Chugani, 2004;Madden and Zup, 2014). In addition, both 5-HT2AR and 5-HT2CR that are expressed throughout the brain have been implicated in the pathophysiology of psychiatric disorders (Abramowski et al., 1995;Hoyer et al., 1986;Heisler et al., 2007;Mestre et al., 2013;Lyddon et al., 2013]. 5-HT2CRs for example, was shown to play a role in addiction and reward behavior via modulating dopamine transmission within the mesolimbocortical dopaminergic system (Katsidoni et al., 2011). Dysregulation of the reward-related neurotransmitter systems and behaviours could also contribute to the increased susceptibility to obesity in the recuperated offspring (Grissom et al., 2014) In summary, we report that elevated 5-HT in fetuses exposed to maternal protein restriction, may underlie a permanent reduction in 5-HT2CR expression and function and secondary counter-regulatory up-regulation of 5-HT2AR and increased sensitivity to 5-HT2AR agonist. Thus, our

Disease Models & Mechanisms • DMM • Advance article
results not only identify a molecular mechanism through which maternal diet may impair offspring energy balance but also point to a promising pharmacological strategy with 5-HT2AR agonist medication to correct this impairment.

Experimental groups and tissue collection
All procedures involving animals were conducted in accordance with the University of were dissected from brains of male offspring at P3. The body weight of the remaining pups was recorded at P7, P14 and P21. Following weaning at day 22, one male per litter was culled by a rising concentration of CO2 and the brain was dissected. After weaning, the remaining males were fed standard laboratory chow and body weight and food intake were recorded weekly. At 3 months of age males were culled and brains were collected. All the dissected brains and heads were frozen on powdered dry ice and were stored at -80°C until further processing. Amniotic fluid and placentas were harvested at E16.5 and serum was prepared from the dams and male pups at P3, weaning and 3 months of age, and stored at -80°C prior to the measurement of 5HT and tryptophan levels.

5-HT and tryptophan assay
Whole brains, placentas, amniotic fluid and serum were used for measurement of 5-HT (DRG International, Inc, New Jersey, USA) and tryptophan (Abnova, Heidelberg, Germany) by ELISA following the manufacturer's recommendation for sample preparation, acylation (5-HT) and derivatisation (tryptophan). For the measurement of 5-HT in tissue, hypothalamic blocks were excised from frozen brains and extracted as described (Huang et al., 2012). Briefly, samples were homogenised and deproteinised for 30 min in 0.2 N perchloric acid solution containing 7.9 mM Na2S2O5 and 1.3 mM disodium ethyleneamine-tetra-acetic acid. The homogenate was centrifuged at 10,000 g for 10 min at 4 0 C and the supernatant assayed for 5-HT.

Disease Models & Mechanisms • DMM • Advance article
Feeding studies Animals were cannulated as described previously (Stocker et al., 2012). Briefly a cannula was inserted into the third ventricle under a gaseous anaesthetic (isofluorane: Isoba, Schering-Plough Animal Health) using coordinates from the stereotactic rat brain atlas (Paxinos and Watson, 1998).
Its position was verified by a positive drinking response over 15 min to angiotensin II (20 gml -1 in 2.5 l). For measurements of acute effects of D-fenfluramine or 5-HT2AR agonism on food intake, 3month-old rats were individually housed, fasted for 4 h, dosed at the beginning of the dark period and re-fed. Peptide (250 nmoles) was given in 2.5 l saline. Animals were dosed using a Latin square design. Doses were separated by at least 4 days and normal feeding behaviour and body weight was restored prior to administration of the next dose. The specificity and doses of the D-fenfluramine (Tocris Bioscience, Bristol, UK), or the high affinity 5-HT2AR receptor agonist TCB-2 (Tocris Biosciences, Bristol, UK) were based on published data (Vickers et al., 1999;Vickers et al., 2001;Trifunovic and Reilly, 2006;McLean et al., 2006;Xu et al., 2010, Fox et al., 2010 and doses were optimised by ourselves.

Western blotting
Hypothalami were dissected from frozen brains according to landmarks: anterior to the optic chiasma; posterior to the mammillary bodies; lateral at the hypothalamic sulcus; and superior to the anterior commissure. Dissected hypothalami from day-3 and 3-month-old male offspring were homogenised in TK lysis buffer and Western blotting analysis was carried out as previously described . Anti-goat primary antibody against 5-HT2CR was purchased from Santa Cruz Biotechnology (cat sc-15081, lot D1114, Santa Cruz, USA) and 1:200 dilution was used. The antibody was validated in two previous studies (Bubar et al., 2005;Anastasio et al., 2010).

Laser-capture microdissection (LCM), RNA isolation and analysis
Hypothalamic sections of the ARC were prepared on a cryostat at 14m thickness from approximately -4.52 to -2.30 mm relative to Bregma (Paxinos and Watson, 1998) with the manufacturer's protocol. The quality and quantity of the RNA samples was determined using Agilent BioAnalyzer PicoChips (Agilent Technologies Inc, Santa Clara, CA, USA). Total RNA was isolated from E16.5 fetal heads and neonatal and adult (3 months of age) hypothalami as previously described (Zaibi et al., 2010) and analyzed using a NanoDrop ND1000 (Thermo Fisher Scientific, Delaware, USA).

RNA amplification
An ovation Pico RNA Amplification System (Nugen Technologies Inc, San Carlos, CA, USA) was used for the amplification of RNA destined for microarray analysis. RNA amplification of LCM ARC samples used to validate genes identified by microarray analysis was performed using a MegaScript T7 Amplification Kit (Ambion) in combination with the GeneChip sample CleanUp Module kit (Affymetrix). The use of a different method of RNA amplification enhanced the validation of the microarray data.

Microarray hybridization
The amplified RNA was used for gene expression profiling on Affymetrix Rat Genome 230 2.0 Arrays (Affymetrix Inc, San Carlos, CA, USA) using the Affymetrix GeneChip protocol to fragment and label the target, ready for hybridization to the arrays (Affymetrix 2004). GeneChip sequences were selected from GenBank, dbEST and RefSeq and the sequence clusters created using UniGene were then further refined by comparison with the publicly available assembly of the rat genome.
Microarray hybridization was carried out by Molecular Biology Services at the University of Warwick, using n=6 chips per group.

Microarray analysis and selection of the genes for validation
Raw image data files were converted to CEL and pivot files using Affymetrix GeneChip Operating Software. All downstream analysis of microarray data was performed using GeneSpring GX 12.0 (Agilent). The CEL files were used for the RMA (Irizzarry et al., 2003) and GC-RMA (Wu et al., 2004) analyses, while the pivot files were used for GCOS analysis. After importing the data, each chip was normalized to the 50 th centile of the measurement taken from that chip and all gene expression data reported as a fold-change from the control state. Genes were considered to be up-or downregulated if the 1.3-fold threshold was reached and P<0.05. Only genes that met the above criteria using three different algorithms: GCOS, RMA and GCRMA were taken forward for additional study.
The further selection of genes for validation was based on the function of the gene and the availability of suitable primers for validation. Functional analysis was performed using Ingenuity Pathway Analysis (Ingenuity Systems Inc).

In Situ Hybridization Histochemistry (ISHH)
Coronal sections (20 m) of frozen hypothalamic ARC, obtained using a cryostat, were thawmounted onto poly-L-lysine slides (Polysine, Menzel Glaser, Braunschweig, Germany). Ten sets of slides per animal were serially collected, with the first set of 10 beginning at approximately -4.52 and ending at -2.12 mm, relative to Bregma, according to the atlas of the rat brain (Paxinos and Watson).
For the sectioning of neonatal rat brains whole heads were used and the sectioning was guided by the neonatal rat brain atlas (Ramachandra and Subramanian, 2011). Slides were stored at -80 C until use.
A [ 35 S]-radiolabelled riboprobe targeting nucleotides 1700-1910 of the rat 5-HT2AR mRNA transcript was generated by PCR from whole rat brain cDNA. The 210 bp fragment was cloned into pCR-TOPO4 (Life Technologies, Paisley, UK). For antisense probe generation, Pst1 linearised recombinant plasmid was subjected to in vitro transcription using T7 polymerase in the presence of 35 S-labelled UTP, as per manufacturer's instructions (Ambion). The ISHH procedure used has been described in detail previously (Alon et al., 2009;Garfield et al., 2012). Autoradiographic images were quantified using Image Proplus software (Media Cybernetics, MD 20910). Standard curves were generated from 14 C autoradiographic microscales (Amersham, UK) and integrated optical density (IOD) and area of the hybridization signal were measured. For each animal, 3-5 sections of ARC (-4.52mm to -2.12 mm from bregma) and 2-3 sections for VMN (-3.60mm to -2.12mm from bregma) were analyzed. Average ARC and VMH 5-HT2AR mRNA expression was calculated.

Dual-label ISHH and immunohistochemistry (IHC)
To assess co-localisation of 5-HT2AR and POMC, brains taken from control rats, were first processed for detection of 5-HT2AR mRNA by ISHH, as described above. Following this, the tissue was washed in PBS before commencement of the IHC protocol to label α-melanocyte-stimulating hormone (MSH) protein using procedures previously described (Alon et al., 2009;Garfield et al., 2012). Briefly, sections were incubated in 0.3% H2O2 in PBS, then rinsed in PBS, blocked in 0.5% BSA/ 0.5% Triton-X 100 in PBS and left in blocking buffer containing rabbit α-MSH antibody (1/10,000; Chemicon, Millipore, MA, USA) overnight. Tissue was then washed in PBS and a biotinylated donkey anti-rabbit secondary antibody (Vector Laboratories, CA, USA) was applied at 1/1000 in blocking buffer. Sections were then washed in PBS, incubated in VectaStain ABC reagent, and following this, chromogenic detection was conducted using 3,3'-diaminobenzidine (DAB) reagent (Vector Laboratories). Sections were mounted onto superfrost slides, dried, then dipped in photographic emulsion (Kodak) and stored at 4°C for two weeks before being developed using Kodak developer and fixer. Double labelled cells were recorded if α-MSH-immunoreactive (IR) positive cell bodies were overlaid with a 35 S 5-HT2AR signal greater than 3X background.

Statistical analysis
Two-tailed unpaired Student's t-test was used for statistical analysis and the data are presented as mean±s.e.m. unless otherwise stated. Fractional growth rates were calculated using formula: fractional growth rate = (current -starting weight)/(period x starting weight). For microarray data p-values were calculated using a two-tailed t-test. All data were analyzed using GraphPad Prism (Graphpad, La Jolla, USA). Number (n) refers to number of litters used. P<0.05 was considered statistically significant.