Fetuin-A inhibits functional maturation of NICCs
To assess whether fetuin-A impacts on functional maturation of beta cells, NICCs were maturated in HSA- or fetuin-A-containing medium. HSA at a concentration identical to that of fetuin-A (0.6 mg/ml) was used as control, since fetuin-A and HSA are plasma components sharing similar characteristics as Ca2+- and fatty acid-binding proteins [25, 26]. Successful maturation was confirmed by upregulation of INS, PDX1, NEUROD1, MAFA, GCK, SYT4 and SYT7 mRNAs, paralleled by the increased number of insulin- and PDX1-positive cells (maturation d5 vs d1; Fig. 1a–h). Fetuin-A prevented upregulation of the aforementioned beta cell markers, an effect unsusceptible to the TLR4 inhibitor CLI-095. However, CLI-095 inhibited fetuin-A-induced increase of IL1B mRNA (Fig. 1i). Fetuin-A also reduced the number of insulin/PDX1-positive cells as well as the protein amount of PDX1 (Fig. 1h, j, k).
Since HSA concentration was rather low and fetuin-A-containing medium was albumin-free, we assessed whether HSA per se interferes with NICC maturation. Neither omission of HSA nor the increase of HSA concentration (0.6 to 1.2 mg/ml) altered NICC maturation. Moreover, addition of HSA to fetuin-A (0.6 mg/ml each) did not counteract the inhibitory effect of fetuin-A (ESM Fig. 1a–g). Despite marked upregulation of functional beta cell genes, matured NICCs (HSA/maturation d5) acquired a modest GSIS (1.4-fold increase). Nonetheless, glucose responsiveness was significantly improved (twofold increase) in the presence of forskolin, an adenylate cyclase activator that increases the intracellular level of cAMP, bypassing membrane receptors (Fig. 1l and ESM Fig. 1h). Fetuin-A-cultured NICCs were glucose unresponsive even in the presence of forskolin. Of note, forskolin-elevated insulin secretion was glucose independent (Fig. 1l).
These results suggest that fetuin-A hinders functional maturation of neonatal beta cells in a TLR4-independent manner.
Role of TGFBR–SMAD2/3 signalling for NICC maturation
Postnatal maturation of beta cells is accompanied by upregulation of p16/Ink4a (also known as cyclin-dependent kinase inhibitor 2A [encoded by CDKN2A]), a cell cycle inhibitor under positive regulation of TGFBR–SMAD2/3. NICC maturation (d5 vs d1) was accompanied by CDKN2A upregulation, while both immature (d1) and mature (d5) NICCs displayed p16/Ink4a-positive nuclei (Fig. 2a,b). Upregulation of CDKN2A was mediated by TGFBR–SMAD2/3 signalling, as it was inhibited by SB431542 and augmented by TGFβ-1 (Fig. 2a).
TGFBR-dependent augmentation of CDKN2A expression during maturation is further corroborated by TGFβ-1 accumulation in culture medium, indicating constitutive expression and release of TGFβ-1 (Fig. 2c and ESM Fig. 1i). The secreted ligand can activate TGFBR with subsequent SMAD2/3 phosphorylation and CDKN2A upregulation (Fig. 2a–e). On the contrary, SB431542 reduced SMAD2/3 phosphorylation, i.e. the lower SMAD2/3 phosphoprotein-band disappeared, while the upper one was minimally affected (Fig. 2d,e). NEUROD1 and SMAD7 were revealed to be regulated by TGFBR–SMAD2/3, since SB431542 reduced both SMAD7 mRNA levels and the increase of NEUROD1 mRNA levels during maturation (Fig. 2f,g). SB431542 affected neither PDX1 nor INS expression (Fig. 2h,i). An active TGFBR–SMAD2/3 pathway is essential for beta cell maturation, as endorsed by the observation that SB431542 increased, while TGFβ-1 decreased, the mRNA and protein levels of aldolase B (ALDOB), a marker of functionally immature as well as of human diabetic beta cells (Fig. 2j–n) [27, 28].
These findings indicate that selective expression of beta cell markers during NICC maturation requires an active TGFBR–SMAD2/3 pathway.
Fetuin-A impairs TGFBR–SMAD2/3 signalling in NICCs
Next, we examined whether the effect of fetuin-A on NICC maturation was mediated by inhibition of TGFBR–SMAD2/3 signalling. Indeed, fetuin-A inhibited phosphorylation of SMAD2/3 (Fig. 3a–c). In accordance, fetuin-A recapitulated the effects of SB431542 on SNAI1 and cMYC expression, two known targets of SMAD2/3 (Fig. 3d–h). Furthermore, fetuin-A counteracted upregulation of UCN3, ABCC8, PCSK1 and G6PC2, an effect also exerted by SB431542 (Fig. 3i–l).
To gain insight into the common and distinct effects of fetuin-A and SB431542 on NICC maturation, and to identify SMAD2/3-regulated genes potentially affected by fetuin-A, we performed comparative bulk RNAseq-based transcriptome analysis of NICCs cultured with fetuin-A, HSA, or HSA + SB431542. Using +1 < Log2FC < −1 and p < 0.05 adjusted for multiple comparisons as thresholds, we found 882 up- and 328 downregulated genes following maturation (HSA; d5 vs d1) (ESM Table 4). In the fetuin-A-treated NICCs (Fet-A/d5 vs HSA/d5) 423 genes were up- and 1198 were downregulated, while in the presence of SB431542, 156 and 289 genes were up- and downregulated, respectively (ESM Tables 5 and 6). In addition to the shared beta cell markers NEUROD1, UCN3, ABCC8, PCSK1 and CASR (Fig. 4a), fetuin-A inhibited a larger set of genes than SB431542, including terms controlling differentiation (PDX1, MAFA, MAFB), insulin secretion (G6PC2, SLC2A2, SLC30A8, SYT4/7/13, membrane receptors) and adaptive proliferation (FOXM1, CDK1, CENPA/F/E, TOP2A) (Fig. 4a,b).
A comparison of the gene sets altered by SB431542 and fetuin-A revealed 172 common targets, suggesting that fetuin-A regulates these genes via inhibition of TGFBR–SMAD2/3 signalling (ESM Fig. 2a). In accordance with reduced SMAD2/3 phosphorylation, fetuin-A altered expression of typical SMAD2/3 targets, such as cMYC, SNAI1, TGFBI or POSTN (Fig. 4b). It is noteworthy that the effect of fetuin-A on gene expression was age-independent, although slightly reduced in NICCs from older piglets (12d vs 4d old animals; Fig. 4b).
To identify common contributors to the transcriptional phenotypes of fetuin-A- and SB431542-treated NICCs, the respective differentially expressed genes (DEGs) were subjected to upstream regulator analysis (Ingenuity). The analysis relies on coordinated changes which impact on downstream targets of these contributors, regardless of whether the contributor’s own expression is altered. z scores predict activation or inhibition of such regulators. While TGFβ1 emerged as the top ranking contributor, the analysis identified SMAD2/3, SP1, FOXM1 and MAFB as transcriptional regulators inhibited by both fetuin-A and SB431542 (Fig. 4c).
These results endorse inhibition of TGFBR–SMAD2/3-dependent transcription as an important contributor to the phenotype of fetuin-A-treated NICCs.
To reveal cellular processes potentially affected by the DEGs, a GO overrepresentation analysis was performed. The maturation-upregulated gene set was enriched for terms related to insulin secretion, membrane potential, cell adhesion and cell–cell communication (Fig. 5a and ESM Fig. 2b). The fetuin-A-downregulated gene set was enriched for GO terms attributed to hormone level, cell adhesion, cell division and ECM organisation (Fig. 5b and ESM Fig. 2c). Similarly, the DEGs downregulated by SB431542 were enriched for terms associated with ECM organisation and cell adhesion (Fig. 5c and ESM Fig. 2d). Several GO terms were common for maturation-, fetuin-A- or SB431542-altered gene sets, indicating that fetuin-A and SB431542 impact on identical cellular processes, such as ECM organisation and cell adhesion. The GO ranking suggested regulation of insulin secretion as most probable consequence of gene upregulation upon maturation, while ECM organisation, cell adhesion and cell cycle processes were consequences of fetuin-A-downregulated genes (ESM Fig. 2b,c).
Fetuin-A impairs NICC proliferation
TGFBR–SMAD2/3 signalling inhibits beta cell proliferation via induction of p16/Ink4a. While SB431542 reduced the mRNA level of CDKN2A (see Fig. 2a), the transcriptome analysis suggested a stimulatory effect of fetuin-A on CDKN2A-dependent transcription (see Fig. 4c). On the other hand, fetuin-A and SB431542 increased cMYC expression (see Fig. 3e–h), a proliferative gene under negative regulation of SMAD2/3. Therefore, we examined the effects of fetuin-A and SB431542 on beta cell proliferation in more detail.
In contrast to SB431542 and in spite of reduced SMAD2/3 phosphorylation, fetuin-A increased CDKN2A expression, an effect counteracted by the TLR4 inhibitor CLI-095 (Fig. 6a). Fetuin-A did not alter the cellular distribution of p16/Ink4a in beta cells (Fig. 6b). Confirming the RNAseq data (Fig. 4b), fetuin-A reduced expression of genes regulating beta cell proliferation, i.e. PRLR1, FOXM1, CENPA, TOP2A and CDK1 (Fig. 6c–g). Distinct to fetuin-A, SB431542 did not alter the expression of these genes. In line with this expression pattern, fetuin-A reduced (0.27 ± 0.08%), while SB431542 and prolactin increased (2.00 ± 0.21% and 2.49 ± 0.28%, respectively) the number of proliferating (i.e. Ki67-positive) beta cells (Fig. 6h–i).
These findings suggest that fetuin-A limits the proliferative expansion of beta cell mass. Indeed, the fetuin-A-treated NICCs displayed significantly lower numbers of beta cells (i.e. insulin-positive), in comparison with control NICCs (14.4 ± 1.2% and 22.3 ± 1.1%, respectively) (Fig. 6j). Despite increasing beta cell proliferation, SB431542 and prolactin had no effect on beta cell number (Fig. 6j).
Fetuin-A impairs TGFBR signalling and glucose responsiveness of adult human islets
In order to translate these results to humans we examined whether fetuin-A impacts on TGFBR–SMAD2/3 signalling in isolated islets from organ donors. In human islets cultured for 2 days with fetuin-A, basal and TGFβ-1-stimulated SMAD2/3 phosphorylation were inhibited (Fig. 7a–c). In accordance, fetuin-A reduced nuclear accumulation of SMAD2/3 in cultured islet cells (Fig. 7d). An Affymetrix-based transcriptome analysis of human islets cultured for 2 days with fetuin-A revealed RANBP3L as the top upregulated gene, while POSTN, CDK1, CENPF, TOP2A and TPX2 occupied prominent positions among the downregulated ones (Fig. 7f). Of note, the qRT-PCR analysis confirmed the fetuin-A-induced upregulation of RAN binding protein 3L (RANBP3L), a protein that regulates nuclear–cytosolic shuttling of SMADs (Fig. 7e, ). The fetuin-A-altered gene set was enriched in GO terms related to calcium transport and ECM organisation, and the GO ranking suggests calcium transport into the cytosol as functional consequence (Fig. 7g and ESM Fig. 3). Fetuin-A abolished glucose responsiveness of human islets (from 1.7-fold increase of secretion [p < 0.05] to 1.1-fold [p > 0.05]) by increasing basal secretion, an effect maintained in the islets co-treated with the TLR4 inhibitor CLI-095 (Fig. 7h). Notably, we found a negative correlation of islet beta cell area with the level of plasma fetuin-A in non-diabetic humans (p = 0.013; Fig. 7i–k and ESM Fig. 4).
These results indicate that fetuin-A impairs functional maturity (i.e. GSIS, and TGFBR signalling) in adult islets. In addition, fetuin-A might compromise the adaptive increase of functional beta cell mass.