Differences in splicing factors may predict type 2 diabetes remission in the CORDIOPREV study

Summary

Alternative splicing is a post-transcriptional process resulting in multiple protein isoforms from a single gene. Abnormal splicing may lead to metabolic diseases, including type 2 diabetes mellitus (T2DM). To identify the splicing factor expression that predicts T2DM remission in coronary heart disease (CHD) patients, we identified newly diagnosed T2DM at baseline (n = 190) from the CORDIOPREV study. Patients were classified as Responders (T2DM remission during 5 years without antidiabetic drugs) or non-Responders. Baseline dysregulation in 5 splicing factors (MBNL1, RBM5, hnRNP G/RBMX, CD44, NT5E) distinguished Responders from non-Responders. Adding these factors to clinical variables [AUC = 0.67], insulin resistance, and beta-cell indexes [AUC = 0.76], improved T2DM remission prediction [AUC = 0.80]. Cox regression analysis showed those with higher remission scores had a 2.63-fold increased remission probability. To conclude, a set of splicing factors that contribute to predicting T2DM remission in patients with CHD has been identified. Further research is needed to elucidate these findings’ clinical relevance.

Graphical abstract

Highlights

• •Splicing factor expression may predict T2DM remission, supporting personalized CHD treatment
• •It contributes valuable insights into dietary approaches to reduce complications
• •The score based on splicing factors shows precision in predicting T2DM remission
• •It represents an advancement toward tailored and precise clinical management of T2DM

Cardiovascular medicine; Public health; Human metabolism

Introduction

Type 2 diabetes mellitus (T2DM) is recognized as a major global public health concern.¹ This chronic metabolic disease, characterized by a persistent state of hyperglycemia and glucose intolerance, leads to the development of serious health complications, including microvascular complications (e.g., nephropathy, neuropathy, and retinopathy),^2,3,4 macrovascular complications with a 2- to 3-fold increased risk of developing cardiovascular disease,⁵ carotid disease,⁶ heart failure,⁷ and non-alcoholic fatty liver disease.⁸ For this reason, several clinical studies have focused on T2DM prevention and treatment, offering the opportunity to develop a series of guidelines on managing T2DM.^9,10

Of note, back in 2016 the World Health Organization (WHO) discussed for the first time the potential of T2DM reversal within its global report on diabetes,¹¹ whereas more recently (in 2019), Diabetes UK suggested that T2DM may be metabolically reversible through low carbohydrate diets, very low-calorie diets, exercise, and bariatric surgery.¹² This approach of reversing T2DM has also been proposed as the primary clinical goal in T2DM patients.¹³ The American College of Lifestyle Medicine has recently (in 2020) considered that T2DM remission can be achieved through intensive lifestyle interventions as the cornerstone of medical care in T2DM patients.¹⁴ In fact, in the context of the CORDIOPREV (CORonary Diet Intervention with Olive oil and cardiovascular PREVention) study, we have identified a set of newly diagnosed T2DM patients who achieved remission after the long-term consumption of a healthy dietary intervention (a Mediterranean or a low-fat dietary pattern), without the use of pharmacological treatment or weight loss.^{15,16,17,18,19,20} Briefly, baseline metabolic characteristics related to better beta-cell functionality and lower hepatic insulin resistance,¹⁵ specific metabolomics profile,^16,17,18 microbiome composition,¹⁹ or even epigenetic factors as miRNAs expression²⁰ were found to be responsible for the T2DM remission observed in the CORDIOPREV study.

Interestingly, in the context of identifying predictive molecular mechanisms and markers, we have previously found that splicing machinery alterations were associated with the risk of T2DM development in the CORDIOPREV population.^21,22 Specifically, alternative splicing is a post-transcriptional process that involves removal of intron and exon ligation to generate multiple mRNA transcripts, thus resulting in different functional protein isoforms encoded by a single gene.²³ These isoforms have distinct temporal and spatial roles, linking alternative splicing to both normal biological activities and the development of diseases.²⁴ Normal splicing process is necessary for maintaining cellular homeostasis; however, a dysregulation of this process is strongly related to the development of many pathologies, such as cancer and neurodegenerative and metabolic diseases, including T2DM.^25,26,27 Key features of diabetes, such as insulin resistance and impaired glucose metabolism, have been associated with alternative splicing patterns. Although the role of specific isoforms is still underexplored, genes such as CD44, a gene involved in cell communication and adipose tissue inflammation,^28,29 and NT5E, which plays a key role in the insulin secretion,³⁰ have been studied for their alternative splicing isoforms in diabetes.

Alternative splicing is catalyzed by the splicing machinery, which is composed of the spliceosome, a large and dynamic ribonucleoprotein complex, whereas approximately 200 pre-mRNA binding proteins are associated with the spliceosome, thus affecting its activity.³¹ Most studies have been performed in the setting of cancer, showing the impact of aberrant mRNA splicing on potent oncogenes and tumor suppressors.³² However, emerging data suggest that aberrant mRNA transcripts may contribute to essential phenotypes related to specific expression patterns, in particular in T2DM patients with coronary heart disease (CHD),²¹ suggesting the relation of alternative splicing with insulin-mediated glucose metabolism in T2DM.^25,26,27

Based on the above, we aimed to identify whether the pattern of expression of certain splicing factors could predict T2DM remission after the long-term consumption of a Mediterranean or a low-fat diet in CHD patients.

Results

Baseline characteristics of the participants

When we compared the Responders with the non-Responders groups (Table S1), we found that non-Responders had higher baseline BMI, WC, body weight, glucose and insulin levels, HOMA-IR and HIRI, as well as lower ISI and DI compared with Responders (all p < 0.05).

Expression of splicing factors is different between Responders and non-Responders

The expression pattern of several splicing factors was dysregulated in the PBMCs of Responders compared with non-Responders at baseline (Figure 1A). In particular, dynamic qPCR array showed that PBMCs of Responders exhibited markedly higher levels of three pre-mRNA binding proteins: MBNL1 (p = 0.017), RBM5 (p = 0.017), and hnRNP G/RBMX (p = 0.029); and two distinct protein isoforms: CD44 (p = 0.033) and NT5E (p = 0.046) compared with non-Responders (Figure 1A).

Figure 1.

Splicing factors expression levels in Responders/non-Responders

(A) Significant differences in the expressions between Responders and non-Responders.

(B) Variable of Importance in Projection (VIP) scores obtained from Partial Least-Squares Discriminant Analysis (PLS-DA) of all the splicing factor studies. Data are represented as mean ± SEM.

VIP score of PLS-DA showed that MBNL1, RBM5, hnRNP G/RBMX, CD44, and NT5E were the best score factors capable to discriminate Responders from non-Responders (Figure 1B).

Expression of splicing factors was associated with changes in insulin levels

In the total population, baseline expression levels of MBL1, hnRNP G/RBMX, and CD44 were inversely related to changes (between pre- and post-intervention) in plasma fasting insulin levels (p = 0.017, 0.048 and 0.003, respectively). We did not observe any association between baseline expression of the different splicing factors and changes in glucose levels (p > 0.05) (Table S2).

Expression of splicing factors is associated to the risk of T2DM remission

The probability of T2DM remission according to the baseline expression levels of splicing factors is shown in Figure 2. Patients with higher expression levels of MBNL1 [odds ratio (OR) 1.50, 95% confidence interval (CI) 1.05–2.14], RBM5 (OR 1.49, 95%CI 1.07–2.07), hnRNP G/RBMX (OR 1.43, 95%CI 1.03–1.98), CD44 (OR 1.43, 95%CI 1.02–1.99) and NT5E (OR 1.36, 95% CI 1.00–1.83) had an increased probability of T2DM remission.

Figure 2.

Odds Ratios (OR) and 95% confident intervals (CI) of type 2 diabetes mellitus (T2DM) remission in CHD patients with the highest expression levels of the 5 splicing factors

Alterations in the expression of splicing factors may predict T2DM remission

We performed an ROC curve analysis to evaluate the potential for splicing factors to classify the patients in Responders and non-Responders. First, based on the classic predictors of T2DM remission (i.e., BMI, age, HDL-c, and triglycerides), we observed an area under the curve (AUC) of 0.67 (95% CI, 0.59–0.75) (Figure 3A). To improve this model, we added insulin resistance and beta-cell function indexes to the ROC analysis. Our results showed an AUC of 0.76 (95% CI, 0.69–0.83) (Figure 3B). Finally, we carried out the ROC curve analysis based on splicing factors (MBNL1, RBM5, hnRNP G/RBMX, CD44 and NT5E) with the highest AUC (0.80, 95% CI, 0.73–0.87) (Figure 3C). The 3 models were significantly different (p = 0.002) through a DeLong Test. We also carried out additional ROC curve analyses based on those splicing factors with a >1.5 VIP score (RBM5, MBNL1, hnRNP G/RBMX, CD44, NT5E, SLU7, EIF4A3, BRBMS1, RBM6 and MAGOH) (Table S3).

Figure 3.

Remission of type 2 diabetes mellitus (T2DM) assessed by ROC curve models based on clinical variables, indexes and splicing factors

(A) Model based on clinical variables, including BMI, age, HDL-C and triglycerides.

(B) Model based on insulin resistance and beta-cell function indexes added to clinical variables.

(C) Model based on the 5 splicing factors (MBNL1, RBM5, hnRNP G/RBMX, NT5E, and CD44) added to clinical variables and indexes. Test Delong = 0.002.

T2DM remission score based on splicing factors

To evaluate the probability of remission, we performed a T2DM remission score based on splicing factors (MBNL1, RBM5, hnRNP G/RBMX, CD44 and NT5E) (Figure 4). We then classified the population according to tertiles of the score and carried out a Cox regression analysis using as a reference the tertile with the lowest probability of T2DM remission (tertile 1 = lowest score). Finally, the hazard ratios (HRs) of the analysis were assessed. Half of the patients with the highest score (highest expression levels, tertile 3) remitted from T2DM after 49 months of follow-up, with a 2.63-fold (HR 2.63, 95%CI 1.28–5.41) increased probability of T2DM remission than those with a low score (lower expression levels) when adjusted by covariates.

Figure 4.

Probability of type 2 diabetes mellitus (T2DM) remission by a T2DM remission score based on 5 splicing factors

The analysis was performed using a Cox regression curve by tertiles of the T2DM remission score and adjusted by significant variables at baseline. Red line indicates low score (T1), blue line indicates medium score (T2), and green line indicates high score (T3). Abbreviations: HR (Hazard Ratio); CI (Confidence Interval).

Discussion

The present study examined whether dysregulation of the expression of certain splicing factors could predict T2DM remission in patients with CHD, thus further promoting the understanding of the underlying molecular mechanisms. We showed that patients with T2DM remission (Responders), after the long-term consumption of a healthy dietary intervention (a Mediterranean or a low-fat diet), presented at baseline with higher PBMCs expression levels of 5 splicing factors (three pre-mRNA binding proteins: MBNL1, RBM5, hnRNP G/RBMX, and two distinct protein isoforms, one from CD44 and another from NT5E) compared with non-Responders. These findings highlight the potential existence of an association between dysregulation in the splicing pattern with the probability of T2DM remission. We also found that the addition of these 5 splicing factors to clinical variables, insulin resistance, and beta-cell function indexes, improved the ability to differentiate between Responders and non-Responders during the 5 years of follow-up. Finally, we developed a T2DM remission score based on these 5 splicing factors and showed that CHD patients with the highest score (i.e., highest expression levels of these 5 splicing factors) exhibited a 2.63-fold increased probability of T2DM remission compared with those with the lowest score (i.e., lowest expression of the 5 splicing factors).

In recent years, a growing body of evidence supports an association between metabolic disease and alterations in splicing processes.^{21,22,25,26,33,34} Two previous reviews summarized the biological importance of the aberrant splicing of some genes linked to obesity and insulin resistance.^25,26 In the context of the CORDIOPREV study, we found that the expression of key splicing factors at baseline (i.e., RNU2, RNU4 and RNU12)²¹ and the change after 3-year (i.e., SPFQ, RMB45 and SRSF6)²² could early predict the development of T2DM, regardless of the type of the healthy diet. In the present study, we identified that specific alterations in the expression of splicing factors may also contribute to the prediction of T2DM remission in CHD patients. The specific genes identified in the present study (i.e., MBNL1, RBM5, hnRNP G/RBMX, CD44, and NT5E) differ from the splicing factors previously associated with T2DM development. In newly diagnosed T2DM patients, dysregulation of glucose homeostasis is already stablished,¹⁵ which may account for the differences in splicing patterns observed compared to non-T2DM patients studied previously.^21,22 Such findings may help to achieve a deeper understanding of the mechanisms underlying the phenotypic flexibility associated with T2DM remission, thus potentially facilitating the early identification of patients likely to attain T2DM remission and also the development of more “personalized” and effective therapeutic strategy.

We described herein a dysregulation in the expression of 3 pre-mRNA binding proteins associated to the spliceosome (i.e., MBNL1, RBM5 and hnRNP G/RBMX) and 2 protein isoforms (i.e., CD44 and NT5E). Some of these factors are primarily related to tumor progression. Specifically, RMB5, a pro-death tumor suppressor gene in cancer cells is highly related to injured brain^35,36 and lung³⁷ and breast cancer,³⁸ whereas high expression of hnRNP G/RBMX (a heterogeneous nuclear ribonucleoprotein that has a role in alternative splicing) correlates with favorable cancer outcomes due to its tumor suppressive effects.^39,40 The role of these factors in diabetes pathophysiology remains largely unknown and, thus, dysregulation in the expression levels of these 2 pre-mRNA binding proteins, in the context of T2DM remission, would provide insights into the complexity of the molecular mechanisms related to metabolic flexibility. However, MBNL1 has been previously associated with T2DM pathogenesis.^41,42 Briefly, MBNL1 can induce insulin receptor pre-mRNA exon inclusion^43,44 thus potentially modulating beta cell survival, whereas the upregulation of MBNL1 expression has been linked to a reduction in the development of diabetic microvascular complications, such as diabetic kidney disease.^41,42

Regarding the protein isoforms, little is known about how splicing process could affect physiological mechanisms in specific diseases.²⁵ In the present study, we observed a dysregulation in the expression of CD44 and NT5E protein isoforms between Responders and non-Responders; in particular, patients with T2DM remission had higher expression levels than those who did not achieve remission. In this context, Fadista J et al. showed that NT5E downregulation was associated with impaired insulin secretion.³⁰ Interestingly, NT5E gene encoding CD73, an enzyme involved in renal vascular injury in the setting of T2DM. Thus, NT5E gene expression may represent a traceable marker of diabetic complications in some studies.^45,46 On the other hand, an expression-based genome-wide association study revealed that CD44 gene is implicated in the molecular pathogenesis of T2DM.²⁹ Briefly, in both diabetic mouse models and humans, Kodama et al. found that the CD44 was implicated in the development of adipose tissue inflammation and insulin resistance, which are common pathophysiological mechanisms seen in the early stages of T2DM.²⁹ In line with the above, the present analysis reported an inverse association between baseline expression levels of splicing factors (MBL1, hnRNP G/RBMX, and CD44) and changes in insulin levels, suggesting that alterations in alternative splicing, that downregulates these splicing factors, could be acting at insulin receptor-mediated pathways. In this way, this dysregulation of the alternative splicing would be affecting disease susceptibility genes that may contribute to T2DM pathogenesis.

Taking all the above into consideration, the present findings may suggest a predictive role of certain splicing factors in the evaluation of T2DM remission probability. Furthermore, the present study has shed some light into the development of efficient dietary strategies to promote T2DM remission and, thus, reduce the risk of diabetic complications. Additionally, we developed a T2DM remission score, based on splicing factors, with a high precision capability of predicting T2DM remission.

In conclusion, we identified a set of splicing factors (i.e., MBNL1, RBM5, hnRNP G/RBMX, CD44, and NT5E) that may be able to evaluate the probability of T2DM remission. To the best of our knowledge, there are no previous studies on splicing factors in relation to predicting T2DM remission. Therefore, our findings “open an opportunity window” in terms of screening strategies and preventive therapies in the context of T2DM remission. This obviously should be further investigated.

Limitations of the study

The present study has a few limitations. First, the analyses of the expression patterns of splicing factors were conducted in the heterogeneous PBMC population, but not in specific cell types or tissues. While this approach provides a broader view of splicing factor dynamics, it does not capture the specific splicing events occurring in endocrine tissues directly relevant to the pathogenesis of T2DM, such as pancreatic islets. Dysregulation in the splicing pattern could be present not only in circulating PBMCs but may also operate in cell types from other tissues and organs, which is an avenue worth exploring. Second, there was the inability to conduct an independent replication of the findings due to limited availability of appropriate data. To support our conclusions, we performed additional analyses, including logistic regression and ROC curves, which further reinforce the predictive value of the biomarkers. Nonetheless, the lack of independent validation should be considered when interpreting the results. Finally, the generalizability of our findings is limited by two factors: 1) the absence of distinction between tertiles 2 and 3 in the T2DM remission score suggests a plateau effect in biomarker expression levels and 2) the study was conducted exclusively in patients with CHD, which restricts the applicability of the results to other populations.

Resource availability

Lead contact

Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Jose Lopez-Miranda (jlopezmir@uco.es).

Materials availability

All materials reported in this paper will be shared by the lead contact upon request.

Data and code availability

• All data reported in this paper will be shared by the lead contact upon request (jlopezmir@uco.es.).

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

The CORDIOPREV study is supported by the Fundación Patrimonio Comunal Olivarero (Cordioprev-CEAS, 1/2016 to J.L.-M.). The main sponsor was not involved in the design or carrying out the study, and its participation was limited to funding and providing the olive oil used in the study. This study also received research grants from Ministerio de Ciencia e Innovación (AGL2012-39615, AGL2015-67896-P and PID2019-104362RB-I00 funded by MCIN/AEI/1.0.13039/501100011033 to J.L.-M.), from Consejería de Salud-Junta de Andalucía (PC-0283-2017 to E.M.Y.-S.) and FIS (PI18/01822 and PI21/00383 to E.M.Y.-S.), integrated into the framework of the National Plan for Scientific Research, Technological Development and Innovation 2013–2016, co-financed by the Instituto de Salud Carlos III (ISCIII) of Spain and also by the Directorate General for Assessment and Promotion of Research and the EU’s European Regional Development Fund (FEDER). A.O.-R. and H.B. were the recipients of Sara Borrell grants (CD21/00099 and CD22/00053, respectively) from the ISCIII. The funding bodies had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. We would like to thank the EASP (Escuela Andaluza de Salud Publica), Granada (Spain), for carrying out the randomization process in this study. The CIBEROBN is an initiative of the Instituto de Salud Carlos III, Madrid, Spain. The authors declare no competing interests. The authors read and approved the final manuscript.

Author contributions

A.O.-R. and J.D.T.-P. contributed equally to this work. E.M.Y.-S. and J.L.-M. contributed equally to this work. A.O.-R. and J.D.T.-P. wrote the first draft of the manuscript. A.O.-R. and A.P.-H. carried out the analysis. J.D.T.-P., A.P.A.-d.L. and O.A.R.-Z. collected the data and performed the classification of participants. A.O.-R., A.P.-H., H.B., and M.E.G.-G. performed the experiments. A.P.A.-d.L. and J.D.T.-P. performed the medical revisions of participants and clinical databases. A.O.-R., J.D.T.-P., H.B., A.L.-M., N.K., R.M.L., E.M.Y.-S., and J.L.-M. interpreted the data and contributed to the discussion. N.K., R.M.L., J.D.-L., P.P.M., E.M.Y.-S., and J.L.-M. contributed to the writing of the manuscript and revised it critically for important intellectual content. All authors read and approved the final manuscript. J.D.-L. and J.L.-M. conceptualized the study and designed the research methodology. J.D.-L. and J.L.-M. are the guarantors of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Biological samples
CORDIOPREV study	Delgado-Lista J. et al.47 and Delgado-Lista J. et al.48	NCT00924937
Chemicals, peptides, and recombinant proteins
Flavored glucose load (Trutol 75)	Thermo Fisher	401009P
Critical commercial assays
Direct-zol RNA kit	Zymo Research	R2055
First Strand Synthesis Kit	Thermo Fisher	K1612
Preamp Master Mix	Fluidigm	100–5580
Exonuclease I	New England Biolabs	M0293L
GE 96.96 Dynamic Array	Fluidigm	BMK-M-96.96
Oligonucleotides
See the Table S4 with designed primers to perform the microarray	This paper	N/A
Software and algorithms
StepOne™ Real-Time PCR system software v.2.3	Applied Biosystems	https://www.thermofisher.com/es/en/home/technical-resources/software-downloads/StepOne-and-StepOnePlus-Real-Time-PCR-System.html
Real-Time PCR Analysis Software v.4.7	Fluidigm	https://www.standardbio.com/products/software
STATA v.12	StataCorp	https://www.stata.com/install-guide/windows/download/
MetaboAnalyst Software v.4.0	McGill University	https://www.metaboanalyst.ca/
Other
UV-visible spectrophotometer	Thermo Fisher	NanoDrop2000 spectrophotometer
Biomark HD System	Fluidigm	N/A

Experimental model and study participant details

Study population

The current work was performed within the framework of the CORDIOPREV study (Clinicaltrials.gov NCT00924937), a prospective, randomized, controlled trial including 1002 patients with CHD, who followed one of two different healthy dietary patterns (a Mediterranean diet and a low-fat diet) for 7 years. The patients were recruited from November 2009 to February 2012, mostly at the Reina Sofia University Hospital (Cordoba, Spain), but other hospitals from the Cordoba and Jaen provinces (Spain) were also included. Inclusion and exclusion criteria have been detailed previously.⁴⁷ Briefly, patients were eligible if they were aged 20 to 75 years, with established CHD but without clinical events in the last 6 months, were willing to follow a long-term monitoring study, and had no other serious illnesses. All patients gave written informed consent to participate in the study.

The trial protocol and amendments were approved by the local ethics committees, following the Helsinki declaration and good clinical practices. The results of the main objective of the CORDIOPREV study have been published elsewhere.⁴⁸

The sample size and power calculation for CORDIOPREV study have been calculated on the following assumptions: an incidence rate in the control group (Low Fat) of 4 events/100 person-years that will amount to 24.9% of absolute cumulative incidence after 7 years, a hazard ratio of 0.7 and a statistical power of 80%, with two tailed alpha = 0.05. Under these assumptions, the required sample size was 491 patients in each of the two groups. The procedure of randomization was carried out so that the assignment to both diets is well-balanced. The randomization was based on the following variables: sex (male, female), age (under and over 60 years old) and previous myocardial infarction (yes, no). With this distribution, eight distinct groups were created, with all the possible combinations of the above factors, and eight different blocks were created to assign the diets (en-bloc randomization). The process of randomization was performed by the Andalusian School of Public Health. The procedure for assigning a diet was as follows: when there was a candidate for randomization, the study dietitians phoned the person in charge of the study in the Andalusian School of Public Health, which communicated the assigned diet to the dietitian. The dietitians were the only members of the intervention team to be aware of the dietary group of each participant. The School of Public Health gave the head of the dietary staff weekly reports with the progress of randomization, and the assigned diets were crossed between each other to validate the correctness of each assignment.

The CORDIOPREV-DIRECT is a study that included all newly diagnosed T2DM patients at the beginning of the CORDIOPREV study, according to the American Diabetes Association (ADA) diagnosis criteria,⁴⁹ who had not been receiving glucose-lowering treatment (190 out of 1002 patients). Of these, seven patients could not be included due to the inability to perform the diagnostic test. Thus, a total of 183 T2DM patients were evaluated in this sub-study. T2DM remission was defined as glycemia below the diabetic range for at least 2 consecutive years (Hemoglobin A1c, HbA1c <6.5%, fasting plasma glucose <126 mg/dL, and 2h plasma glucose after 75g in the oral glucose tolerance test, OGTT <200 mg dL⁻¹) without the use of diabetes medication to lower blood glucose levels.⁵⁰ Patients were tested yearly for follow-up and classified as Responders or non-Responders at the 5^th year of the study. From the 183 patients included in the present sub-study, 73 patients (60 males and 13 females) reverted from T2DM during the 5 years of dietary intervention without the use of antidiabetic medication (i.e., Responders), while 110 (92 males and 18 females) did not achieve diabetes remission at the end of the follow-up period (i.e., non-Responders).

RNA extraction, quantification and reverse transcription

Direct-zol RNA kit (Zymo Research, Irvine, CA, USA) was used to isolate total RNA from PBMCs following manufacturer’s instructions. UV-visible spectrophotometer (NanoDrop2000 spectrophotometer Thermo Fisher) was used to quantify the extracted RNA. The quality of the RNA samples was reflected in the ratio of the absorbance obtained at 260 nm/280 nm, which should be within a range of 1.8–2.0 indicating high purity of RNA. Random hexamer primers were used to reserve transcribed 1 μg of RNA (RT) to cDNA using with the First Strand Synthesis Kit (Thermo Fisher).

Realtime qPCR and customized qPCR dynamic array based on microfluidic technology

A 96.96 Dynamic Array (Fluidigm, San Francisco, CA) using a microfluidic-based technique for gene-expression analysis was implemented to determine the expression of 96 transcripts in 96 samples, simultaneously. Specific primers for human transcripts including splicing factors (components of spliceosome and pre-mRNA binding proteins, $n=$72) and protein isoforms ($n=$21), three housekeeping genes [beta-actin (ACTB), glyceraldehyde-3-Phosphate dehydrogenase (GAPDH) and Hypoxanthine phosphoribosyltransferase (HPRT)] were specifically designed with Primer3 software and StepOne Real-Time PCR system software v2.3 (Applied Biosystems, Foster City, CA) (Table S4). The selection of splicing machinery components in this study was based on two key criteria: (1) the critical role of specific spliceosome components in the splicing process, including core spliceosome elements, and (2) their established involvement in regulating splicing variants associated with the pathophysiology of T2DM, as demonstrated by the splicing factors included in this analysis.

Preamplification, exonuclease treatment and qPCR dynamic array based on microfluidic technology were implemented following manufacturer’s instructions using the Biomark System and the Real-Time PCR Analysis Software (Fluidigm) as previously published in detail.^21,22

Method details

Dietary intervention

Patients were randomized into two different healthy dietary patterns: a Mediterranean diet, with a minimum of 35% of calories from fat (22% monounsaturated, 6% polyunsaturated, <10% saturated), and a maximum of 50% carbohydrates; and the low-fat diet recommended by the National Cholesterol Education Program and the American Heart Association, comprising <30% total fat (<10% saturated, 12–14% monounsaturated, and 6–8% polyunsaturated), 15% proteins, and a minimum of 55% carbohydrates. In both diets, the cholesterol content was adjusted to <300 mg/day. Patients received the same intensive dietary counseling and were monitored by nutritionists, internists and cardiologists. Details about diets and randomization have been previously reported.⁴⁸

Anthropometric and biochemical measurements

After a 12-h overnight fast, patients were admitted to the laboratory for anthropometric and biochemical measurements [body weight, body mass index (BMI), waist circumference (WC), systolic blood pressure, diastolic blood pressure, HDL-cholesterol (HDL-c), LDL-cholesterol (LDL-c), triglycerides, highly sensitive C-reactive protein (hs-CRP), glucose, insulin and HbA1c].

Anthropometric parameters were measured by trained dietitians using calibrated scales (BF511 body composition analyzer/scale, OMROM, Japan) and a wall-mounted stadiometer (Seca 242, HealthCheck Systems, Brooklyn, NY).

Venus blood samples were obtained from participants in tubes containing EDTA after a 12-h overnight fast. Isolated peripheral blood mononuclear cell (PBMCs), lipid variables, serum insulin, and plasma glucose were determined as previously reported.^51,52

Determination of insulin resistance, and beta-cell function indexes

OGTT, insulin resistance, and beta-cell function indexes were previously reported.^51,52 In brief, patients underwent a standard OGTT analyzed by the Matsuda and DeFronzo method⁵³ at baseline and year-to-year during the follow-up period. After a 12-h overnight fast, blood was sampled from a vein before the oral glucose intake (0 min) and again after a 75g flavored glucose load (Trutol 75; Custom Laboratories, Balti-more, MD, USA). Blood samples were taken at 30, 60, 90, and 120 min to determine glucose and insulin concentrations. OGTT provided the data needed to calculate the homeostatic model assessment of insulin resistance (HOMA-IR), the insulin sensitivity index (ISI), the insulinogenic index (IGI), the hepatic insulin resistance index (HIRI), the Muscle Insulin sensitivity index (MISI) and the disposition index (DI) as previously reported.^51,52

Quantification and statistical analysis

The application of preprocessing methods and the construction of classification and clustering methods were performed to estimate the relevance of specific factors.

All statistical analyses were performed using STATA version 12 (StataCorp, College Station, TX, USA) except the clustering analyses which were performed with MetaboAnalyst Software v.4.0 (McGill University, Quebec, Canada).

Normality was assessed by Shapiro-Wilk. All the tests were two-sided, and the significance level was set at α = 0.05. We used the mean and standard error of the mean (mean ± SEM) for continuous variables, and percentages for categorical variables. Student’s unpaired test was used for comparison between Responders and non-Responders, resulting in a significant p-value <0.05 after correction for false discovery rate (FDR) correction.

Variable importance in projection (VIP) score of partial least squares discriminant analysis (PLS-DA) were measured to evaluate the contribution of splicing factors to the model in order to distinguish the Responders from the non-Responders.

Linear and logistic regression models were performed to assess the relationship between the expression levels of the splicing factors with clinical parameters and the probability of T2DM remission, respectively.

Receiver operating characteristic curve (ROC) analysis was run for assessing the potential for splicing factors to classify the population in Responders and non-Responders. We performed 3 models: the first based on clinical variables (BMI, age, HDL-c and triglycerides), the second model with added insulin resistance and beta-cell function indexes (HOMA-IR, ISI, IGI, HIRI, MISI and DI) and finally we included the splicing factors.

In order to evaluate the probability of T2DM remission using the factors of the splicing, we calculated a T2DM remission score. We performed a generalized linear model analysis, and we multiplied the z value of each splicing factors by the expression value of factors in all subjects. Finally, we added the five factors to obtain a single value per subject. We classified the patients in tertiles of the T2DM remission score and we performed a Cox regression analysis.

Confuse variables included in the adjusted models were those with significant differences at baseline: body weight, BMI, WC, HbA1c, glucose, insulin, HOMA-IR, ISI, HIRI and DI.

Additional resources

Trial Registration: CORonary Diet Intervention With Olive Oil and Cardiovascular PREVention (CORDIOPREV). Clinicaltrials.gov NTC00924937. https://clinicaltrials.gov/ct2/show/NCT00924937.

Published: December 4, 2024