GALNT2 expression is associated with glucose control and serum metabolites in patients with type 2 diabetes

Aims Aim of this study was to investigate in type 2 diabetes whether expression level of GALNT2, a positive modulator of insulin sensitivity, is associated with a metabolic signature. Methods Five different metabolite families, including acylcarnitines, aminoacids, biogenic amines, phospholipids and sphingolipids were investigated in fasting serum of 70 patients with type 2 diabetes, by targeted metabolomics. GALNT2 expression levels were measured in peripheral white blood cells by RT-PCR. The association between GALNT2 expression and serum metabolites was assessed using false discovery rate followed by stepwise selection and, finally, multivariate model including several clinical parameters as confounders. The association between GALNT2 expression and the same clinical parameters was also investigated. Results GALNT2 expression was independently correlated with HbA1c levels (P value = 0.0052), a finding that is the likely consequence of the role of GALNT2 on insulin sensitivity. GALNT2 expression was also independently associated with serum levels of the aminoacid glycine (P value = 0.014) and two biogenic amines phenylethylamine (P value = 0.0065) and taurine (P value = 0.0011). The association of GALNT2 expression with HbA1c was not mediated by these three metabolites. Conclusions Our data indicate that in type 2 diabetes the expression of GALNT2 is associated with several serum metabolites. This association needs to be further investigated to understand in depth its role in mediating the effect of GALNT2 on insulin sensitivity, glucose control and other clinical features in people with diabetes. Supplementary Information The online version contains supplementary material available at 10.1007/s00592-024-02280-7.


Introduction
Insulin resistance is a common pathogenic ground for many highly prevalent diseases.These include atherogenic dyslipidemia [1], type 2 diabetes, obesity, hypertension [1], Managed By Massimo Federici.and closely related cardiovascular disease and renal dysfunction [1,2], all major causes of morbidity and mortality worldwide [3].Unraveling the intimate molecular signature of insulin signaling would contribute to understanding the pathogenesis of all the above-mentioned diseases and is therefore urgently needed.
Thanks to the recent advances in bioinformatics and technology, measuring hundreds or thousands of metabolites in biological samples has unraveled specific signatures related to altered metabolic states, including insulin resistance, type 2 diabetes and obesity [28].
We investigated whether GALNT2 expression is characterized by a specific metabolic signature.In details, five different metabolite families were investigated, including acylcarnitines, aminoacids, biogenic amines, phospholipids and sphingolipids.

Participants
The study cohort consisted of 70 patients with type 2 diabetes (according to the American Diabetes Association 2003 criteria), belonging to the Gargano Mortality Study 2 (GMS, [29]) including individuals recruited from 2008 to 2010 at the Endocrine Unit of Fondazione Istituto di Ricovero e Cura a Carattere Scientifico "Casa Sollievo della Sofferenza" in San Giovanni Rotondo.Our 70 study patients were randomly selected among those whose RNA sample at recruitment was available.The study protocols and the informed consent procedures were approved by the local Institutional Ethic Committee.

GALNT2 expression levels
Gene expression levels were measured in peripheral white blood cells by using Gene Expression Assay on Demand Kit Reagents (Applera Life Technologies, Carlsbad, CA), by means of RT-PCR as previously described [24].Expression levels of GALNT2 were calculated by using the comparative DCT method normalizing, the amount of GALNT2 was normalized to GAPDH, B actin and 18S considered together (geometric mean) [30] and related to a control RNA as calibrator (2 −ΔΔCT ).

Statistical analysis
Patients' baseline characteristics are reported as mean ± SD, or median and interquartile range for continuous skewed variables (|skewness| > 1) and frequency and percentage for categorical variables.Values of serum metabolites below the limit of detection have been replaced by the limit of detection itself.
For pre-processing of data, normal distribution and skewness were tested in all metabolites and covariates.Since the metabolites' distributions were skewed, all concentrations were log 2 transformed and standardized.The 2 −ΔΔCT data of GALNT2 expression levels were also standardized.Clinical parameters with percentage of missing value less than 5% (i.e., BMI 1.4% and HbA1c 2.8%) were imputed with random forest method [31].
The association between GALNT2 expression and serum metabolites within each metabolite family (i.e., acylcarnitine, amino acids, biogenic amines, glycerophospholipids and sphingolipids), was firstly assessed in a univariate model by using false discovery rate (FDR) to take into account multiple comparisons and then in a multivariate model including age, sex, smoking habits, BMI, HbA1c, diabetes duration, eGFR, and anti-hypertensive, anti-hyperglycemia and lipidlowering therapies.Finally, in order to minimize potential multicollinearity issues, metabolites that were independently associated with GALNT2 expression, entered jointly a stepwise selection (SSE criterion: p value of the F-statistic to enter and to remove term to the model less than 0.05 and greater than 0.10 respectively).A P value < 0.05 was considered statistically significant.
All analyses were performed using SAS 9.4 (SAS Institute, Cary, NC) and Matlab R2022-Statistics and Machine Learning Toolbox (The MathWorks, Inc., Natick, MA).

Study patients
Clinical features as well as diabetes duration and ongoing treatments of the 70 study participants with type 2 diabetes are reported in Table 1.

GALNT2 and metabolites
Five out of 188 metabolites measured (i.e., carnosine, DOPA, dopamine, nitrotyrosine, cis-4-Hydroxyproline) were excluded from the analyses because their value was below the detection limit in > 80% samples.
All five metabolites remained significantly associated with GALNT2 mRNA levels in a multivariate model that included age, gender, smoking habits, BMI, HbA1c, diabetes duration, eGFR, and current treatments (P values = 0.0027, 0.000049, 0.013253, 0.035 and 0.00067, for glycine, asparagine, ADMA, phenylethylamine and taurine, respectively), thus indicating that their correlation with GALNT2 expression is independent of the most important clinical variables (Fig. 1B).After a stepwise (forward-backward) analyses aimed at taking into account the correlations among metabolites from the same family, the aminoacid glycine (P value = 0.0014) and two biogenic amines phenylethylamine (P value = 0.0065) and taurine (P value = 0.0011) remained associated with GALNT2 expression levels.

Discussion
Our study investigated whether in patients with type 2 diabetes GALNT2 expression is characterized by a specific signature belonging to several metabolite families, including acylcarnitines, amino acids, biogenic amines, glycerophospholipids and sphingolipids.We also investigated the association between GALNT2 expression and several clinical variables.Firstly, GALNT2 expression was independently and negatively correlated with HbA1c levels, a finding that may well be secondary to the reported positive effect of GALNT2 on insulin sensitivity [6].This link is also suggested by the negative association between GALNT2 and BMI, which however did not survive a multivariable model comprising several additional clinical variables.The expression of GALNT2 was also independently associated with serum levels of the aminoacid glycine and arginine and the biogenic amines phenylethylamine, taurine and ADMA.When collinearity within the same metabolite family was taken into account, only glycine, taurine and phenethylamine remained associated with GALNT2 expression levels.Interestingly, the association between GALNT2 expression and HbA1c was not modified taking into account these three latter metabolites, thus suggesting they do not mediate the positive effect of GALNT2 on glucose control.
Previous studies have highlighted that glycine, is consistently and negatively associated with reduced insulin Fig. 1 Associations between GALNT2 expression levels and serum metabolites.β (per 1 SD increase GALNT2 expression) and 95% CIs were estimated in univariate (A) and in multivariate (B) regression models, adjusting for age at recruitment, sex, smoking habit, BMI, HbA1c, eGFR, diabetes duration, and ongoing treatment sensitivity [33,34], impaired glucose homeostasis [33,[35][36][37][38][39] and liver steatosis [38].In addition, low glycine levels have been reported to predict prospectively the development of type 2 diabetes [36,37,[39][40][41].Also plasma taurine is reduced in subjects with metabolic syndrome [42], diabetes [43,44] and obesity [45,46] as well as in obese animals [47].All these previous evidences on the role of glycine and taurine make our present correlative findings compatible with the belief that GALNT2 is involved in insulin sensitivity and resistance [6].On the other hand, we acknowledge that their interpretation is not straightforward.Indeed, if the positive effect of GALNT2 on insulin sensitivity were mediated by the above-mentioned metabolites, one would expect an association with glycine and taurine in the opposite direction to that observed in our study (i.e., positive rather than negative correlation).This makes unlikely that glycine or taurine mediate the positive effect of GALNT2 on glucose control as also suggested by the observation that the association between GALNT2 and HbA1c does not change much after adjusting for these two metabolites.It can therefore be hypothesized that the counterintuitive associations we here report represent a homeostatic mechanism in which GALNT2 upregulation acts as a fine tuner to counteract insulin-resistance induced (or simply marked) by low levels of glycine and taurine.Conversely, no published data are available on circulating phenethylamine levels in different conditions related to metabolic abnormalities.Interestingly, fecal phenethylamine levels, derived from bacterial fermentation of amino acids in the gut, are correlated positively with glucose intolerance and negatively with improved dietinduced insulin sensitivity [48] while urinary phenylethylamine levels were higher in obese women as compared to their normal/underweight counterparts [49].These reports suggest that phenethylamine also plays a role in several clinically relevant insulin resistance phenotypes.Unfortunately, it is not known whether and how fecal and urinary phenethylamine levels are correlated with serum levels, thus making difficult the interpretation of the positive association we observed between GALNT2 expression and circulating phenethylamine.In all, we do acknowledge that the associations of GALNT2 expression levels with HbA1c and several circulating metabolites may imply more than a single and unambiguous interpretation and, consequently, does not allow, yet, to define a clear metabolic signature linking GALNT2, circulating metabolites and clinical features related to insulin resistance.Among limitation of our study, we do recognize that expression data in peripheral white blood cells may not mirror those of other tissues, including the most important ones for glucose homeostasis maintenance.On the other hand, this cell model has been successfully used in cis-eQTL, trans-eQTL analyses from the eQTLGen consortium (https:// www.eqtlg en.org/) aimed at understanding the genetic architecture underlying complex traits including insulin resistance-related abnormalities [50].Furthermore, we recognize that the small sample size of our study impacts statistical power, thus making it possible that we missed additional associations between GALNT2 expression and circulating levels of other metabolites as well as subtle effects of the associated metabolites on the role of GALNT2 on HbA1c and other clinical features (i.e., false-negative results).
In conclusion, our data indicate for the first time that in type 2 diabetes the expression of GALNT2 is associated with several serum metabolites.This association needs to be further investigated.To understand in depth its role in mediating the effect of GALNT2 on insulin sensitivity, glucose control and other clinical features in people with diabetes.If our current findings are confirmed and deepened by other studies to gain a better comprehension of the molecular effects of GALNT2 on insulin sensitivity, this will likely become instrumental in the discovery of hitherto unknown pathogenic nodes that can be targeted with new therapies in patients with insulin resistance and related anomalies.

Table 1
Clinical characteristics of study patients (n = 70)