We thoroughly characterized the association patterns of galectin-1 and galectin-3 to established and exploratory markers of metabolic disease in a community-based cohort. This allowed us to identify both clear distinctions as well as striking similarities between the two blood-based biomarkers in different aspects of metabolism. While galectin-1 was associated with several markers of adiposity on both an anatomic and proteomic level, galectin-3 showed no such tendency. Conversely, the two galectins were associated with different insulin sensitivity markers. Galectin-1 was associated with C-peptide and the Matsuda index, while galectin-3 was associated with fasting insulin in the fully adjusted models. Both had negative associations with fasting glucose, and positive associations with cholesterol, fatty acid, and triglyceride metabolism. Similar associations on the protein level also reflected this, with identified associated proteins including FABP4 and 5, as well as the LDL-receptor. A potential metabolic role of these galectins is interesting, in particular since galectin inhibitors are studied in clinical trials (1).
There was a clear distinction between the two galectins when comparing the association profiles and the Imiomics analysis for galectin-1 and galectin-3 on measures of obesity and adipose tissue distribution. Plasma levels of galectin-1 were closely associated with all adiposity variables, both in the subcutis and viscera. While ectopic fat deposition in the liver and pancreas were also significantly associated, this association disappeared after BMI adjustment, suggesting that these were indirect associations secondary to BMI. Several studies have previously reported close associations between galectin-1 and adipose tissue measures (7, 8, 30), and mechanistic studies in animal models suggest a functional role for galectin-1 in adipocyte handling of lipids (9, 11). Several of the previously proposed mechanisms from animal models also fit with the observations we find in the proteomic analysis, suggesting that galectin-1 may also play a role in adipose tissue organization in humans (9, 31). It is possible that galectin-1 interacts with leptin, FABP1, -4 and − 5, TGF-β1, and perilipin-1 and − 3 to modulate whole body lipid storage as indicated by our measurements on BMI, body fat, and Imiomics. The absence of associations between circulating galectin-3 levels and adipose tissue deposition was unexpected, and it could be that galectin-3 has a more prominent role in other tissues than the adipose tissue. As an example, pharmacological galectin-3 inhibitors are currently under evaluation in clinical trials for the treatment of non-alcoholic steatohepatitis (32).
Several reports have previously indicated the association of galectin-1 with glucose and insulin and a functional role is suggested in mechanistic studies in animal models (7–9). Here, galectin-1 was associated with all glucose homeostasis variables except for end-OGTT glucose value before BMI adjustment. The closest associations were seen for C-peptide and the insulin resistance measures, HOMA, and the Matsuda index, suggesting a role involving insulin resistance rather than glucose or insulin itself. As these associations were mitigated or lost after adjustments for BMI, it is possible that the functional role of galectin-1 lies mainly in adipose tissue metabolism as suggested previously (9, 11). The association with C-peptide could also indicate a role of galectin-1 on insulin release in the pancreas, as recently proposed by a study in mice (33). Galectin-3 was associated with insulin and glucose in the fasted state and did not associate with the direct measures of insulin resistance in fasting or during the OGTT. Galectin-1 has previously been associated with an improved glucose uptake, independent of insulin secretion. The inverse association observed for both galectins with fasting glucose may be explained by such a mechanism (34, 35). This could also explain the lack of association with the 2-hour glucose measurement. In contrast to the 2-hour glucose value, fasting glucose levels are closely dependent on hepatic glucose production, which is regulated by the fasting insulin levels. Notably, galectin-1 was associated with C-peptide suggesting an insulinotropic action of galectin-1. A direct effect of the galectins on the hepatocyte is thus another possibility for the inverse association with fasting plasma glucose. The general lack of associations between galectin-3 and markers of insulin resistance was surprising, as several studies have previously found associations in experimental studies in animal models (5, 36). It is well-known that the result of any study on insulin resistance depends on the definition of the study outcome, and that there are currently many different definitions of the term “insulin resistance” itself (37). The different outcomes for galectin-1 and − 3 on markers of insulin resistance could indicate that they act on different metabolically active tissues.
Several proposed ligands are shared between galectin-1 and galectin-3, although sometimes these common ligands are identified independently in separate studies (38, 39). Our study measuring both galectins in the same cohort supports such concept. This should also be considered as it could indicate some overlapping functionality of the two galectins. This notion is further supported by our observation of similar associations with proteins in triglyceride metabolism and the LDL-receptor. Although these galectins have an overlapping interaction profile in vivo and in vitro (40), their global expression is somewhat different. Galectin-1 protein expression is highest in adipose tissue, muscles, and tissues present in females while galectin-3 expression is highest in the gastrointestinal tract, lungs, skin, kidneys, and bone marrow (https://www.proteinatlas.org/, last accessed on 2022-02-15).
In line with this concept, the associations seen with glucose and lipid metabolism in the targeted metabolomics analysis further support an overlapping functionality between the two galectins. Similarities in cholesterol, fatty acid, and triglyceride markers were almost identical, with few exceptions including free-cholesterol and HDL-metabolites, where statistical significance was not matching. It was interesting to find similar associations with both LDL cholesterol and triglycerides for the galectins. BMI is normally more closely associated with TG than with LDL. Nonetheless, both LDL and triglycerides are known to associate with abdominal fat deposition, and a positive energy balance (41, 42). It is thus possible that our data indicate that both galectin-1 and galectin-3 are involved in human lipid metabolism, which is further supported by experimental studies on galectin-1 in adipose tissue (11) and galectin-3 in non-alcoholic steatohepatitis (12, 32). This also aligns with reports that galectin-1 and galectin-3 may act through the PPAR-γ pathway, although in different tissues (11, 12). Additionally, both galectin-1 and galectin-3 were associated with IL-10 and TNF signaling. These pathways are well known to be involved in obesity-related inflammation and lipid metabolism (43, 44).
In our non-targeted metabolomics analysis, the two galectins presented associations with metabolites in distinctly different pathways. Galectin-1-associated metabolites were found in pathways related to histidine and cysteine metabolism. Both cysteine and histidine play a role in galectin-1 function, and it is possible that these pathways are involved in galectin-1 activity (45, 46). Pentose and glucuronate interconversion and the TCA cycle pathways are related to carbohydrate metabolism, possibly relevant to our clinical variable observations. Pentose and glucuronate interconversions, histidine metabolism, and ascorbate and aldarate metabolism are metabolic pathways previously reported to be altered together in an animal model of renal failure, as well as an inflammatory state in the lung (47, 48). Whether the known association between galectin-1 and inflammation is related to these metabolic pathways remains to be determined in future studies. Galectin-3 was related to sphingolipid metabolism. This observation can be explained by the reported capacity for galectin-3 to interact with glycosphingolipids, e.g., during endocytosis (49). Sphingolipid metabolism and GPI-anchor biosynthesis pathways have previously been reported to be altered together in a study of a lipid-lowering drug on high-fat-fed mice (50). These metabolic pathways may be related to our other observed associations between galectin-3 and lipid metabolism, including fatty acids, cholesterols, and FABP4 and − 5.
There are some limitations to consider in this study. The POEM cohort consists almost exclusively of white participants, and it is possible that other results would be found in participants with other ethnicities. We did not stratify any analysis by sex, except the Imiomics, because of the sample size. Any sex-specific differences will therefore not be seen in this study. However, the stratified analyses that we were able to perform did not indicate any major differences between males and females, and all analyses were adjusted for sex to limit any potential bias. The close association between galectin-1 and adiposity also complicates the analysis of other variables, in turn related to obesity. There is an overt risk in over-adjusting the linear models, as galectin-1 can be an agent in obesity released from the adipocytes to regulate metabolic actions, and adjusting for BMI might then introduce bias to the analysis. To mitigate this, we present associations with clinical variables before and after adjustment for BMI in this report.
Taken together, we show that while galectin-1 and galectin-3 reveal distinctly different associations with obesity and adipose tissue distribution, they also present very similar associations with markers of glucose and lipid metabolism, including cholesterol, fatty acids, and triglycerides. Our observations suggest that galectin-1 and − 3 may have overlapping metabolic functions, but are utilized in different tissues. This study indicates a more explicit role for galectin-1 in human adipose tissue and highlights the importance of considering general galectin effects in the study of individual galectins.