Bariatric surgery has been established as the most effective strategy for inducing sustained weight loss and enhancing metabolism to manage morbid obesity and its systemic co-morbidities. These procedures improve longevity and quality of life and lead to remission of type 2 diabetes mellitus (T2DM) and reduction in cardiovascular risk, often independent of weight loss [1]. The physiological effects of bariatric surgery include the BRAVE steps (bile flow alteration, reduction of gastric size, anatomical gut rearrangement and altered flow of nutrients, vagal manipulation and enteric gut hormone modulation) with the associated modulation of the gut microbiome and a multitude of downstream physiological and disease-modifying effects [2, 3]. However, the precise mechanism behind bariatric surgery remain poorly understood.

MicroRNAs are a family of endogenous, non-coding RNAs that regulate gene expression at the post-transcriptional level. These small, evolutionarily conserved RNA transcripts bind to target messenger RNA (mRNA) transcripts leading to mRNA cleavage or translational repression [4]. This enables each of the approximately 3000 distinct human microRNA sequences to fine-tune or silence several mRNA targets involved in a diverse array of cellular pathways. Thus, microRNAs form a complex network of gene ‘super-regulators’ controlling virtually every biological process. As such microRNAs have been implicated in the progression of obesity and its many co-morbidities. MicroRNAs regulate adipogenesis, insulin secretion, glucose uptake and lipid metabolism and many other biological processes linked with obesity [5].

While microRNAs act at the subcellular level, they are also found in the circulation within exosomes or complexed with high-density lipoproteins or argonaute proteins [6]. The mechanisms whereby microRNAs appear in the circulation and the biological implications of circulation expression are not understood. However, microRNAs are stable in circulation, not readily digested by ribonucleases and able to withstand repeated freeze-thaw cycles allowing for long-term storage of circulatory fluids without compromising the microRNA integrity [7]. They can also be measured with speed, ease and sensitivity through quantitative PCR and other platforms. This has given microRNAs rich clinical and therapeutic potential as novel biomarkers [8]. Recent studies have reported circulating microRNAs as biomarkers for cancer, obesity, diabetes and many other disorders [7, 9, 10].

Recently, single post-surgical timepoint studies have shown that bariatric surgery modulates circulating microRNA expression [9, 11, 12]. The objective of the present work was to complete a longitudinal repeat sampling study in which patients were their own preoperative controls, to assess the temporal effect of bariatric surgery on circulating microRNA expression profiles and to identify potential microRNA biomarkers.

We have shown for the first time that RYGB persistently and progressively modulates circulating microRNA expression in a temporal manner with an increasing departure in profile from the preoperative obese state. In total, RYGB led to the differentiated expression of 48 circulating microRNAs typified by the selective, time-dependent differential expression of specific microRNAs. A temporary, immediate response is followed by a longer-lasting late response.

The circulating microRNAs whose expression is modified following RYGB are predicted to regulate cell growth, inflammation and neurological processes, and at the later months receptor pathways involved in gut hormone signaling, lipolysis and diabetic kidney disease. Although pathway analysis was largely predicted computationally, many of the circulating microRNAs whose expression is modified by RYGB have been experimentally linked with the metabolic syndrome. MiR 93, shown here to be decreased in the first few months following surgery, inhibits glucose transport and is overexpressed in insulin-resistant women with polycystic ovary syndrome [19]. MiR 27, another member of the early bariatric microRNA response, is a reported promoter of adipogenesis [20]. MiR 15a, decreased an astonishing 85 fold 1 year following surgery, also promotes adipogenesis [21] and miR 192 has a central regulatory role in diabetic nephropathy [22]. The altered expression of these microRNAs following RYGB may explain some of the mechanisms behind the health benefits associated with bariatric surgery.

The early postoperative months demonstrated the fewest changes in circulating microRNA expression, but the number increased longitudinally. Interestingly, as illustrated by the PCA scores plot of circulating microRNA profiles (Additional file 2: Figure S2), at 1 month following RYGB the microRNA circulating profile shifts from the preoperative profile, before a pronounced and gradual trajectory shift in the opposite direction from 3 months onwards. A possible explanation is that the initial physiological response to bariatric surgery was homeostatic as the body attempted to maintain the status quo. There is increasing evidence to suggest microRNAs help to maintain system robustness, fine-tuning or ‘buffering’ gene expression in response to subtle internal or external stimuli [23]. However, once a tipping point has been reached this homeostatic safety mechanism was adjusted and reset, moving the individual physiologically and phenotypically further from the morbidly obese state. In essence, this progressive timeline shift of circulating microRNA expression reveals a view of the physiological trajectory of bariatric surgery.

Many of the circulating microRNAs whose expression has been found to be altered by RYGB in the current study have also been reported as circulating biomarkers of T2DM, including of miR 15a, miR 192, miR 130b and miR 125b [24–26]. The loss of circulating miR 126 in particular is reportedly a strong selector for T2DM [25]. MiR 126 expression increased over 10 fold in each of the first 4 postoperative time points relative to preoperative levels, although none of the increases reached statistical significance (Additional file 3: Table S1). Importantly, fasting blood glucose levels positivity correlated with circulating miR 320a (Additional file 3: Table S2), consistent with previous findings [26] and supporting its candidacy as a novel diabetic biomarker. Circulating expression of miR 320a also correlated negatively with weight loss. In total, circulating levels of 9 microRNAs significantly correlated with fasting blood glucose levels. Correlations between circulating microRNAs levels and age were also found, supporting previous studies that show age impacts microRNA levels [27].

Although previous studies [9, 11, 12] have reported that bariatric surgery leads to changes in the expression of circulating microRNAs, these were time snapshots and lacked the temporal dependencies we have uncovered in the present longitudinal study. Lirun et al. looked at the effect of Roux-en-Y gastric bypass on circulating microRNAs 3 months postoperatively in diabetic patients, divided into a low BMI and high BMI preoperative groups [11]. They reported that 39 microRNAs were differentially expressed after surgery across both groups. Both Lirun et al. and our study reported the downregulation of let 7, miR 93 and miR 106b expression 3 months postoperatively, while the decrease in miR 16 expression at 3 months in Lirun et al. was only significant in our study after 9 months. Interestingly, they reported differences in the postoperative changes in microRNAs in the low and high BMI groups, suggesting that changes in microRNAs following surgery may depend on preoperative conditions.

Two further studies assessed circulating microRNA expression before and 12 months after RYGB. Ortega et al. [9] reported a significant modulation of 14 circulating microRNAs including decreases in miR 125b and miR 16 and an increase in miR 221, mirroring our findings (miR 211 increased across all postoperative timepoints, albeit not significantly). However, they also reported significant increases in the circulating levels of miR 130b and miR 21. The expression of both microRNAs were significantly decreased in our study 12 months following RYGB. Ortega et al. were also unable to detect a change in miR 15a following surgery. In another study, Hubal et al. reported the modified expression of 168 exosome-derived circulating microRNAs 1 year following RYGB and illustrated a correlation between altered microRNAs and improvements in insulin resistance [12]. Filtering for microRNAs with validated targets in insulin signaling or that correlated with changes insulin resistance, they identified several surgery-responsive microRNAs, including miR 125b and miR 122. However, it is important to reiterate that each of these previous studies measured circulating microRNAs at a single postoperative timepoint and are therefore merely snapshots of a dramatically altering physiology. Our longitudinal study emphasizes that the post-bariatric surgery circulating microRNAome and associated changes of physiology are both dynamic.

In rats, circulating levels of miR 122 decrease 60 fold following RYGB compared to sham-operated animals [15]. MiR 122 is a liver microRNA responsible for regulating lipid metabolism [28] and its decrease leads to increased glucose transportation, accelerated glycolysis and the inhibition of gluconeogenesis following RYGB [15]. This is consistent with the recent discovery that RYGB reprograms intestinal glucose metabolism by increasing glycolysis and glucose uptake [29], and suggests miR 122 could contribute to this effect. The present study also found a consistent decrease in circulating miR 122 in humans following RYGB from 3 months onwards, although this did not achieve statistical significance (Additional file 3: Table S1). Other microRNAs that decreased following RYGB in rats include miR 93, miR 30e and miR 320 [15] and we report significant decreases in the human orthologs of these microRNAs in our study.

The role of microRNAs following bariatric surgery remains relatively unexplored and offers potential insights into the dynamic physiological changes that follow surgical intervention. Our study, as with the few that have preceded it, was limited to a small number of participating individuals and should be considered a pilot study at this stage. Additionally, not all subjects provided samples at all post-operative timepoints. Nevertheless, our study confirms the proof of principle that analysis of circulating microRNAs offers insight into the dynamic physiological changes associated with bariatric intervention. Well-designed, multi-center studies with larger diabetes subsets, non-bariatric obese and lean control groups and more robust clinical follow-ups and endpoints would provide a clearer picture into the role microRNAs play following bariatric surgery. Although the field is still in its infancy, bariatric surgery has been consistently shown to significantly alter circulating microRNA levels. MicroRNA expression profiles offer a wealth of biological information and mechanistic insights, as the changes in the regulation of several core physiological processes are related to the expression of the microRNAs that regulate them. MicroRNAs can act as mediators or effectors in positive and negative feedback loops in wider regulation networks. Their presence in circulation in exosomes and high-density lipoprotein or argonaute complexes that can be taken up in an active form by recipient cells is also suggestive of a potential tissue-to-tissue communicative role and that microRNAs may have hormonal as well as biomarker potential [30].

We have shown here that RYGB fundamentally alters microRNA expression in the circulation in a temporal manner with an increasing departure from the preoperative state. These microRNAs correlate with the beneficial physiological and metabolic fluxes seen after these operations and they therefore represent potentially novel post-surgical biomarkers of surgical outcomes. There is ongoing debate on updating the criteria for bariatric surgery, in recognition of its ability to improve health beyond weight loss. MicroRNA expression profiles could potentially add another facet for consideration in the decision-pathway for bariatric operations and offer additional opportunity for developing the next generation of combined multi-modal metabolic interventions to enhance the treatment outcomes for obesity and its associated disorders.