Elevated fetuin-A levels, chemokines and islet-resident macrophages are crucial factors associated with obesity-mediated type 2 diabetes (T2D). Here, the aim of the study was to investigate the effect of MIN6 (a mouse insulinoma cell line)-derived fetuin-A (also known as AHSG) in macrophage polarization and decipher the effect of M1 type pro-inflammatory macrophages in commanding over insulin secretion. MIN6 and islet-derived fetuin-A induced expression of the M1 type macrophage markers Emr1 (also known as Adgre1), Cd68 and CD11c (Itgax) (∼1.8 fold) along with increased cytokine secretion. Interestingly, suppression of fetuin-A in MIN6 successfully reduced M1 markers by ∼1.5 fold. MIN6-derived fetuin-A also induced chemotaxis of macrophages in a Boyden chamber chemotaxis assay. Furthermore, high-fat feeding in mice showed elevated cytokine and fetuin-A content in serum and islets, and also migration and polarization of macrophages to the islets, while β-cells failed to meet the increased insulin demand. Moreover, in MIN6 culture, M1 macrophages sharply decreased insulin secretion by ∼2.8 fold. Altogether our results support an association of fetuin-A with islet inflammation and β-cell dysfunction, owing to its role as a key chemoattractant and macrophage polarizing factor.
Insulin resistance acts as the gateway to the development of insulin-dependent type 2 diabetes (T2D). The pancreatic islets remain engaged in increasing insulin secretion, better known as hyper-insulinemia, to compensate for impaired insulin signalling to maintain glucose homeostasis (DeFronzo et al., 2015). Overnutrition is one of the key factors behind insulin resistance and the progression of T2D. The pancreatic β-cells eventually succumb to the excess work load during sustained hyperinsulinemic state. The β-cells either undergo dysfunction or death. The failure of the β-cells has been a major topic of investigation ever since obesity and T2D has become endemic. Several factors have been identified for their role in loss of function and mass of β-cells. Intra-islet inflammation is one such factor that has gained lot of importance due to its role in β-cell death (Donath et al., 2013). A recent study has shown that there are two different subsets of islet-resident macrophages that tend to expand during obesity-mediated T2D and pose a threat to β-cell function. This study focusses on the origin of islet inflammation as one of the prospective processes in progression of T2D (Ying et al., 2019).
Macrophages are activated by various endogenous and exogenous stimuli, which produces an assembly of precisely activated cells, depending on the tissue and type of stimuli present in the local milieu of the organ (Gordon, 2003; Stout et al., 2005). Hence, the macrophage population consists of a mixture of cells. Macrophage polarization divides macrophages into M1, as pro-inflammatory or classically activated macrophages, and M2, as anti-inflammatory or alternatively activated, macrophages. Classical activation consists of inflammatory stimuli, such as IFN-γ, lipopolysaccharide (LPS) and TNF, which generate M1 macrophages with high capacity to produce powerful pro-inflammatory cytokines like interleukin (IL)-1β, TNF and IL-6. It must be noted that M2 macrophages produce anti-inflammatory mediators like IL-10 and TGF-β, and are associated with later stages of resolving inflammation (Martinez and Gordon, 2014). The shift in the polarity from M2 to M1 macrophage during obesity is usually characterized by expression of F4/80 (encoded by Emr1, also known as Adgre1), CD11b and CD11c (also known as Itgam and Itgax, respectively) (Lumeng et al., 2007; Zeyda and Stulnig, 2007). During islet inflammation, this shift in the polarity of islet macrophage population contributes to β-cell dysfunction in T2D mouse models (Eguchi et al., 2012; Jourdan et al., 2013; Westwell-Roper et al., 2014).
Various immune cells and even β-cells are guessed to be the cellular origin of islet inflammation (Donath et al., 2013). However, details of all the factors that trigger islet inflammation are still not identified. Recently some studies have shown that a high-fat diet (HFD) is one of the major reasons behind increasing islet inflammation. It has been demonstrated by Ying et al. (2019) that mice fed with regular chow diet have a subset of islet immune cells that are F4/80+/CD11c−, while obese mice fed with an HFD have a subset of islet immune cells that are F4/80+/CD11c+ (Ying et al., 2019).
Macrophages sense the changes in microenvironment and respond to the environmental cues through moderation of cell surface receptors. Toll-like receptors (TLRs), which are members of the pattern recognition receptor (PRR) family, are the main receptors to relay these environmental cues and bring about changes in phenotype of the macrophages (Stout et al., 2005). Diet-induced obesity is one of the reasons behind the supply of these environmental cues that bring about an increase in islet macrophage population (Calderon et al., 2015; Ying et al., 2019). In this context, the role of TLR4 in pancreatic β-cell dysfunction must be mentioned. Activation of TLR4 has a major contribution in β-cell failure and stimulation of isolated rodent islets by TLR4 ligands impair function and survival of β-cells (Amyot et al., 2012; Li et al., 2013; Nackiewicz et al., 2014; Osterbye et al., 2010; Vives-Pi et al., 2003). Interestingly, β-cells are a source of various inflammatory factors that aid in islet inflammation (Mahdi et al., 2012; Segerstolpe et al., 2016). Fetuin-A (also known as AHSG), a circulatory glycoprotein, is one of those factors and a known bio-marker of chronic inflammatory diseases (Ix and Sharma, 2010; Johnson and Olefsky, 2013; Mori et al., 2006; Pal et al., 2012). We have recently showed that fetuin-A is secreted from pancreatic β-cells, aggravates inflammation and β-cell dysfunction and death via the TLR4–nuclear factor NF-κB pathway (Mukhuty et al., 2017, 2021a,b). Fetuin-A forms complex with free fatty acids (FFAs), binds to TLR4, and activates TLR4 signalling, which plays a crucial role in lipid-induced insulin resistance (Pal et al., 2012). Another study by us has shown that fetuin-A stimulates macrophage chemotaxis and their polarization to the inflammatory type in adipose tissue (Chatterjee et al., 2013); however, the role of fetuin-A in islet inflammation and accumulation of macrophage in the islets has not been studied yet. Apart from fetuin-A, β-cells are also a source of MCP-1 (also known as CCL2), a well-known chemoattractant for macrophages (Piemonti et al., 2002; Yu et al., 2006). Hence, the β-cells are a hub of various inflammatory factors which can contribute to islet inflammation. As a matter of fact, it is assumed that islet inflammation mediates obesity-induced β-cell alterations (Aamodt and Powers, 2017). IL-1β is one such macrophage-derived pro-inflammatory cytokine that is over-secreted during obesity and causes insulin resistance in multiple organs including pancreatic islets (Stienstra et al., 2012; Youm et al., 2011). However, the crucial factors derived from β-cells during overnutrition that lead to macrophage migration to the islets and their polarization, remain poorly understood. The two most prominent gaps that remain in our understanding are the role of β-cell-derived fetuin-A in islet inflammation and the effect of M1 type macrophages in the impairment of β-cell function.
Here, we have assessed the role of HFD in islet inflammation and β-cell derived fetuin-A in macrophage migration and polarization. We have also demonstrated an intricate crosstalk between M1 macrophages and pancreatic β-cells focussing on its insulin secretory function.
Lipid-mediated fetuin-A secretion from MIN6 cells induces M1 markers in RAW 264.7 macrophages
In order to investigate the effect of palmitate and fetuin-A in macrophage polarization, RAW 264.7 macrophages were stimulated by a combination of palmitate (0.50 mM) and fetuin-A (100 µg/ml) for 24 h. ELISA analysis from the cell culture medium of the above treated cells indicated a significant elevation, by 20±1.5-fold (mean±s.e.m.), in the level of pro-inflammatory cytokines, namely IL-1β, TNF and IL-6 (Fig. 1A–C; P<0.05). To confirm whether palmitate plus fetuin-A influenced macrophage polarization, the RAW macrophages were immunostained with anti-CD11c antibodies and subjected to flow cytometry analysis to assess the shift in CD11c+ cells. The flow cytometry results indicated a distinct right shift in CD11c+ RAW 264.7 cells in palmitate plus fetuin-A-stimulated cells (Fig. 1D). Since, fetuin-A acts as a ligand for TLR4, we inspected the effect of fetuin-A via the TLR4 pathway. Interestingly, palmitate, when used in combination with fetuin-A, activated TLR4 and phosphorylation of the p65 subunit of pNF-κB [also known as RELA; pNF-κBp65 (Ser536)] was seen, which resulted in increase of NF-κB-mediated secretion of pro-inflammatory cytokines IL-1β, TNF and IL-6 (Fig. 1E; Fig. S3A; P<0.05). Paralleled with the induction of pro-inflammatory cytokines, the gene expression of general and M1 type macrophage markers significantly increased by 1.78±0.05-fold (mean±s.e.m.) due to the combination of palmitate and fetuin-A (Fig. 1F; P<0.05). However, palmitate and fetuin-A used alone in various doses for 24 h caused no significant stimulation in expression patterns of TLR4, pNF-κBp65(Ser536) and its downstream proteins IL-1β, TNF and IL-6, or in the gene expression of M1 markers (Fig. S1A–D). We treated MIN6 cells side-by-side with various doses of palmitate for 4 h to stimulate fetuin-A secretion, and used this palmitate conditioned medium from MIN6 cells to trigger the RAW macrophages. The amount of fetuin-A secreted from palmitate stimulated MIN6 was illustrated in Fig. 1G (P<0.01). Palmitate conditioned medium from MIN6 cells when used to stimulate RAW 264.7 cells elicited an increase in secretion of cytokines such as IL-1β, TNF and IL6 (Fig. 1H–J) as well as increased the expression of pNF-κBp65(Ser536) (Fig. 1L; Fig. S3B; P<0.05). There was also a distinct right shift in CD11c+ RAW 264.7 cells in the 0.50 mM palmitate conditioned medium-treated cell population and the higher concentrations as well (Fig. 1K). Gene expression analysis from these treated cells demonstrated 2-fold increase in expression of M1 type marker CD11c (Fig. 1M). To check the same in case of peritoneal macrophages, these cells were isolated from standard diet (SD)-fed mice and then treated with palmitate (0.25, 0.50, 0.75 and 1.0 mM) conditioned medium from isolated islets from SD mice. The cell culture medium of these peritoneal macrophages were then subjected to ELISA analysis for cytokines which supported the findings from cell line as evident from increased secretion of IL-1β, TNF and IL6 (Fig. 1N–P). There was also a 2-fold rise in gene expression of CD11c in the peritoneal macrophages treated with palmitate conditioned medium from islets (Fig. 1Q). Together these findings showed that palmitate plus fetuin-A or medium from palmitate-stimulated MIN6 cells (containing fetuin-A) can significantly alter macrophage polarity towards the M1 type.
Knockdown of fetuin-A in MIN6 cells reduces M1 markers in RAW 264.7 macrophages
The existence of a positive correlation between MIN6-derived fetuin-A and macrophage polarization was clear from the previous experiment. To understand whether fetuin-A secreted from MIN6 could regulate polarization of RAW 264.7 cells, we examined the expression of pro-inflammatory cytokines and M1 markers in RAW 264.7 cells treated with palmitate conditioned medium from fetuin-A siRNA-transfected MIN6 cells (fetuin-A KD). First, the suppression of fetuin-A by using specific siRNA in MIN6 was confirmed by quantitative (q)PCR analysis, which showed an almost 70% suppression of fetuin-A gene expression (Fig. S4A; P<0.01). ELISA analysis for fetuin-A from cell culture medium of palmitate treated fetuin-A KD MIN6 cells showed significant suppression in secretion of fetuin-A from MIN6 cells (Fig. 2A; P<0.05). Palmitate conditioned medium from fetuin-A KD MIN6 cells significantly reduced secretion of the pro-inflammatory cytokines IL-1β, TNF and IL-6 from RAW 264.7 cells, by 2–3-fold, compared to RAW 264.7 cells treated with fetuin-A-containing palmitate conditioned medium from control siRNA-transfected MIN6 cells (Fig. 2B–D; P<0.01 vs conditioned medium from control MIN6 cells; P<0.05 vs palmitate conditioned medium from MIN6 cells). The decrease in fetuin-A levels in the cell culture medium of palmitate-treated fetuin-A KD MIN6 failed to activate TLR4 in RAW 264.7 significantly; hence, the levels of IL-1β, TNF and IL-6 in cell culture medium did not increase owing to reduced expression of pNF-κBp65(Ser536) (Fig. 2E; Fig. S4B; P<0.05). In addition, the expression of general and M1 type macrophage markers significantly decreased in RAW 264.7 cells treated with palmitate conditioned medium from MIN6 cells (Fig. 2F; P<0.05 vs conditioned medium from control MIN6 cells; P<0.05 vs palmitate conditioned medium from MIN6 cells). Next, we subjected RAW 264.7 macrophages treated with palmitate conditioned medium from fetuin-A KD MIN6 cells to flow cytometry after immunostaining with anti-CD11c antibodies. The flow cytometry results also supported the previous results, since there was a clear left shift in CD11c+ cells in RAW 264.7 macrophages treated with palmitate conditioned medium from fetuin-A KD MIN6 cells, which was significantly distinguishable from the right shift in RAW 264.7 cells treated with fetuin-A-containing palmitate conditioned medium from MIN6 cells (Fig. 2G). These results imply that if fetuin-A is silenced in MIN6 cells, then even palmitate conditioned medium from these cells cannot polarize the macrophages significantly.
MIN6-derived fetuin-A stimulates macrophage migration
We performed experiments using a Boyden chamber system chemotaxis assay (Fig. 3A) to examine whether fetuin-A secreted from MIN6 cells could act as a chemoattractant for macrophage infiltration in the islet microenvironment. Under inflamed condition, β-cells may release fetuin-A, MCP-1 and certain pro-inflammatory cytokines concomitantly. ELISA analysis for IL-1β, TNF and MCP1 from cell culture medium of MIN6 cells showed a rise in their level during palmitate stimulation (Fig. S5A–C; P<0.01). Hence, we immunodepleted fetuin-A, MCP-1, IL-1β and TNF to check the alterations in macrophage migration. Addition of fetuin-A significantly increased macrophage migration, compared to that of MCP-1 or IL-1β plus TNF (Fig. 3B–E). Fetuin-A secreted from palmitate-stimulated MIN6 cells also had significant stimulatory impact on macrophage migration (Fig. 3F). Immunodepletion of MCP-1, IL-1β and TNF decreased macrophage migration; however, immunodepletion of fetuin-A along with the above stated factors decreased macrophage migration significantly (Fig. 3G,H). To further prove the role of islet-derived fetuin-A in chemoattraction of macrophages conditioned medium from fetuin-A KD MIN6 as well as conditioned medium from palmitate (0.50 mM) treated fetuin-A KD MIN6 were also used to stimulate the macrophages. However, suppression of fetuin-A in MIN6 cells due to transfection with fetuin-A siRNA did not stimulate significant migration of macrophages (Fig. 3I,J). Optical density (OD) values from the stain extracted from the above experimental set up shown in Fig. 3K supported the data of the photomicrographs (P<0.05). These findings showed that fetuin-A secreted from MIN6 cells had a significant influence on macrophage migration in the midst of other factors like MCP-1, IL-1β and TNF.
HFD-induced inflammatory cytokine secretion in islets affects β-cell insulin secretion
Since a HFD is related to β-cell dysfunction, we tested whether this dysfunction could be attributed to increased islet inflammation. Results from a glucose tolerance test (GTT), insulin tolerance test (ITT) and homeostatic model assessment of insulin resistance (HOMA-IR) scores confirmed that regulation of blood glucose was normal in SD mice as evident from the circulatory glucose levels at subsequent time intervals. However, blood glucose levels remained high in HFD mice, particularly after 12 weeks of HFD feeding (Tables S1–S4 and Fig. 4A; P<0.05). Chronic feeding with HFD led to significant gain in weight in HFD mice (Fig. S2A) (P<0.05). Insulin secretion measured after in vivo glucose-stimulated insulin secretion (GSIS) showed no significant change in the SD groups; however, insulin secretion increased until the 16th week of HFD feeding, followed by an abrupt decrease from the 20th week (Fig. 4B; P<0.05). Furthermore, we attempted to investigate whether a HFD augmented cytokine expression in the serum and islet microenvironment. In HFD mice, high fat content in diet accelerated release of multiple cytokines (IL-1β, TNF and IL-6) from islets as well as their serum levels from the 12th week of HFD feeding, and fetuin-A release from 8th week of HFD feeding (Fig. 4C–G; Fig. S2J–M; P<0.05). The content of fetuin-A in serum in all HFD groups during the total course of treatment was also studied, which indicated a similar trend in increase of circulatory fetuin-A after 8 weeks of HFD treatment that continued to increase until 20 weeks of HFD administration (Fig. 4D; P<0.05). To understand the reason behind increased insulin secretion until 16 weeks of HFD feeding followed by the abrupt fall during the 20th week of HFD, the gene expression of insulin in islets of SD and HFD mice was compared. qPCR analysis for insulin showed that gene expression of insulin continued to increase significantly until 16 weeks of HFD feeding, with the most significant rise observed in 12- and 16-week HFD, where insulin expression was ∼2-fold compared to that of respective SD groups. However, insulin expression in 20-week HFD group was 0.45±0.05-fold that of the age-matched SD mice (Fig. 4H; P<0.05). The results for insulin gene expression corroborated with insulin ELISA analysis results from in vivo GSIS and explained the interesting pattern of serum insulin observed in HFD mice. We also checked the levels of different factors involved in TLR4-NF-κB pathway in the isolated islets, since HFD triggered fetuin-A secretion in the islets and fetuin-A is a known ligand for TLR4. No significant stimulation in TLR4-NF-κB pathway was observed in the islets of SD mice (Fig. 4I). Hence, the levels of fetuin-A and cytokines (IL-1β, TNF and IL-6) which are stimulated by NF-κB, did not show any elevation, along with unchanged levels of Pdx1 and GLUT2 (Fig. 4I). By contrast, in the HFD groups, the TLR4–NF-κB pathway was triggered due to presence of lipid and fetuin-A in the islet microenvironment (Fig. 4I; Fig. S2B,C; P<0.05). As a result, levels of fetuin-A, IL-1β, TNF and IL-6 increased (Fig. 4I; Fig. S2D–G; P<0.05). Altogether, the increase in levels of fetuin-A, IL-1β, TNF and IL-6, and activated TLR4–NF-κB pathway impaired insulin secretion through a decrease in the levels of Pdx1 and GLUT2 (Fig. 4I; Fig. S2H,I; P<0.05). Apart from ELISA analysis, the presence of fetuin-A in the islets was also confirmed by co-immunostaining the islets from SD and HFD mice groups with insulin and fetuin-A antibodies. The results depict the presence of green (insulin-positive) and red fluorescent (fetuin-A-positive) cells in HFD islets from 8 weeks onwards (Fig. 4J). These results indicate that a HFD induces secretion of fetuin-A and pro-inflammatory cytokines via activation of TLR4–NF-κB pathway in pancreatic islets, which results in decreased insulin secretion via lowering the levels of Pdx1 and GLUT2.
A HFD shifts islet macrophage polarization to a more pro-inflammatory phenotype
It is clear from the previous results that a HFD contributes to increased inflammation in the islet microenvironment owing to the elevated levels of pro-inflammatory cytokines. We assessed the accumulation of macrophages and their phenotype by immunostaining the cells isolated from islets with anti-CD11c antibodies and subjected them to flow cytometry analysis. The percentage of CD11c+ cells in the islets during the 8th week of HFD feeding was higher than in the 4th week of HFD; however, a significant increase was observed during the 16th and 20th week of HFD feeding, as evident from right shift in the population of CD11c+ cells (Fig. 5A). A similar trend was observed in gene expression of various M1 and M2 markers from the islets isolated from HFD mice groups. As general macrophage markers, we measured the gene expression patterns of Cd68 and Emr1, which continued to increase in the HFD groups starting from the 12th week of HFD (Fig. 5B; P<0.05). Paralleled with the increased content of pro-inflammatory cytokines in the islet cells, the expression of M1 marker Cd11c increased significantly from the 12th week of HFD onwards (Fig. 5C; P<0.05). In contrast, the M2 markers Cd206 (also known as Mrc1) and Arg1 showed reduced expression in the HFD groups (Fig. 5D; P<0.01). Again, we observed the same pattern of heavy accumulation of CD11c+ macrophages in the islets of HFD mice from the 12th week of HFD administration, as evident from the microscopic images of immunofluorescence staining of insulin (green fluorescence) and CD11c (red fluorescence) co-immunostained islets, while no significant change was observed throughout the duration in SD mice (Fig. 5E). Quantification of the number of infiltrated macrophages in islets depicted an almost 10-fold increase in accumulation of CD11c+ macrophages in HFD islets, particularly during the 12th, 16th and 20th week of HFD treatment, compared with respective SD groups (Fig. 5F; P<0.05). These results demonstrate that HFD influences accumulation of M1 macrophages in the pancreatic islets, which induces islet inflammation.
M1 polarized RAW 264.7 cells impair insulin secretion from MIN6 cells
To examine the effects of M1 and untreated macrophages on β-cell function, the RAW 264.7 cells were divided into two sets. One set was left untreated and the other set was stimulated with palmitate (0.50 mM) and fetuin-A (100 µg/ml) for 24 h to polarize them into the M1 type. Conditioned medium from these two sets of RAW 264.7 cells was used to treat MIN6 cells for 4 h, and the effect on insulin secretion and the signalling pathway involved therein was studied. Treatment with conditioned medium from M1 RAW 264.7 cells impaired GSIS in MIN6 cells, decreasing insulin secretion by 2.75±0.05-fold (mean±s.e.m.) compared to untreated MIN6 cells while it increased by 2.12±0.03-fold in MIN6 cells treated with conditioned medium from untreated RAW 264.7 cells in comparison to MIN6 cells treated with conditioned medium from M1 type RAW 264.7 cells (Fig. 6A; P<0.01 vs untreated MIN6 cells; P<0.05 vs M1 treated MIN6 cells). In addition, immunoblot analysis from MIN6 cells treated with conditioned medium from M1 type RAW 264.7 cells showed reduced expression of Pdx1 and GLUT2 (Fig. 6B; Fig. S5E; P<0.05). In MIN6 cells incubated with conditioned medium from M1 type RAW 264.7 cells, the expression of insulin, Pdx1, MafA and Beta2 (also known as Neurod1) was significantly reduced by ∼0.4–0.6-fold (P<0.05) compared to that of untreated cells, while their expression increased 0.33–0.38-fold (P<0.01) in MIN6 cells incubated with conditioned medium from untreated RAW 264.7 compared to MIN6 cells treated with conditioned medium from M1 type RAW 264.7 cells (Fig. 6C). Flow cytometry analysis from MIN6 cells treated with conditioned medium from M1 and untreated RAW 264.7 cells showed a significantly distinct right to left shift in insulin-positive cell population in M1 conditioned medium-treated MIN6 cells compared to that of untreated RAW 264.7 conditioned medium-treated MIN6 cells (Fig. 6D). The flow cytometry data indicated that the pool of MIN6 cells treated with conditioned medium from M1 type RAW 264.7 cells were deficient in insulin content. These results demonstrate that M1 RAW 264.7 macrophages have a severe derogatory effect on the insulin-secreting function of MIN6 cells.
The present study unravels the deleterious potential of a HFD and excess fetuin-A in increased islet inflammation, accumulation of M1 macrophages and impaired β-cell function. A hallmark of tissue inflammation is the accumulation of various types of immune cells. Detailed profiling of the immune cells along with their functional aspects may provide the essential links about the mechanism by which the immune cells disrupt the normal function of the β-cells (Lin et al., 2012). Classically activated M1 macrophages secrete excess proinflammatory cytokines, like TNF, IL-6 and IL-12, whereas M2 macrophages are linked with maintaining homeostasis and tissue repair (Rocha and Libby, 2009). To understand the detailed mechanism of increased islet inflammation during obesity and T2D, an in-depth knowledge about the factors responsible for macrophage polarization is necessary. Interestingly, the pancreatic β-cells are a hub for several factors, like MCP-1, macrophage migration inhibitory factor (MIF), IL-6 and IL-1β, that promote islet inflammation (Maedler et al., 2002; Piemonti et al., 2002; Stojanovic et al., 2012; Van Belle et al., 2014). A positive correlation between the degree of pancreatic steatosis and the number of CD68-positive macrophages or monocytes suggests that hepatic fetuin-A and palmitate stimulate fat cells to augment macrophage and monocyte infiltration via MCP-1 production (Gerst et al., 2017). In this regard the knowledge about fetuin-A as a macrophage chemoattractant in case of adipose tissue is crucial (Chatterjee et al., 2013), since β-cells are also source of fetuin-A, which is known to activate the TLR4 pathway (Mukhuty et al., 2017; Pal et al., 2012). Fetuin-A is secreted from β-cells under the stimulation of palmitate and is mediated by NF-κB, as already known to occur in hepatocytes (Dasgupta et al., 2010; Mukhuty et al., 2017). We believe that β-cell-secreted fetuin-A also does the same thing, creating an additive effect to that hepatic-secreted fetuin-A and playing a major role in islet inflammation. Eguchi et al. have shown that palmitate stimulates islet inflammation by recruiting M1 macrophages to pancreatic islets, which aids in β-cell dysfunction in a T2D mouse model (Eguchi et al., 2012). Linking the contributions of palmitate and TLR4 in islet inflammation and the fact that fetuin-A is secreted from β-cells, we hypothesized that activation of TLR4 in macrophages by fetuin-A secreted from β-cells along with excess fat might have an impact on the polarization of the macrophages. It is known that palmitate along with fetuin-A induces TLR4 signalling along with a rise in the gene expression of various proinflammatory cytokines in human adipocytes (Pal et al., 2012). To prove our hypothesis in vitro, we subjected RAW 264.7 macrophages to treatment with palmitate and fetuin-A, and also fetuin-A secreted from palmitate-treated MIN6 cells separately. Interestingly, both the treatments elicited polarization in RAW 264.7 cells, and changed their phenotype towards the M1 type, as demonstrated by increased cytokine secretion and expression of M1 markers. In this context, it must be mentioned that palmitate and fetuin-A when used alone did not trigger significant polarization after 24 h of incubation. From these results, it was evident that MIN6-secreted fetuin-A had the potential to polarize RAW 264.7 macrophages. Similar results for M1 polarization were obtained from isolated peritoneal macrophages from SD mice when treated with palmitate conditioned medium from islets. To confirm the contribution of β-cell-derived fetuin-A in macrophage polarization, our next experiment was designed to assess macrophage polarization in RAW 264.7 macrophages after suppressing fetuin-A secretion from MIN6 cells. The fetuin-A KD MIN6 cells failed to secrete a sufficient amount of fetuin-A when stimulated by palmitate. Under these circumstances, a significant reduction in secretion of pro-inflammatory cytokines and gene expression of M1 markers was observed in RAW 264.7 cells treated with palmitate conditioned medium from fetuin-A KD MIN6 cells. Data from flow cytometry analysis gave a similar indication, owing to the decrease in the population of CD11c+ RAW 264.7 cells, as indicated by a left shift, when treated with palmitate conditioned medium from fetuin-A KD MIN6 cells. The results from this experiment clearly established the role of MIN6-derived fetuin-A in inducing M1 polarization and a phenotypic change in RAW 264.7 cells. Furthermore, the association of β-cell-derived fetuin-A with increased macrophage polarization could be the possible reason behind increased islet inflammation during obesity-mediated T2D.
Another major goal of this study was to decipher the chemotactic activity of β-cell-derived fetuin-A. Fetuin-A is known to have a role in chemoattraction of macrophages in adipose tissue (Chatterjee et al., 2013) and, in this context, we must not forget that β-cells secrete fetuin-A (Mukhuty et al., 2017). β-cells are also the source of several chemotactic factors, like MCP-1, MIF and cytokines (Donath et al., 2013; Kamata et al., 2014; Piemonti et al., 2002; Stojanovic et al., 2012; Yu et al., 2006). Results from a Boyden chamber assay (macrophage migration assay) demonstrated that MIN6-produced fetuin-A was biologically active because it chemoattracted RAW 264.7 in the chemotaxis assay and its activity was well correlated with that of recombinant fetuin-A used as a standard. In this context, it must be mentioned that, recombinant fetuin-A attracted a significantly higher number of macrophages in comparison to MCP-1 and IL-1β plus TNF. Moreover, the chemotactic activity for RAW 264.7 cells was notably abrogated upon neutralizing with anti-fetuin-A antibody along with anti-MCP-1, anti-IL-1β and anti-TNF antibodies in the palmitate-treated MIN6 culture medium in comparison to what was seen with immunodepletion of MCP-1, IL-1β and TNF, indicating that other chemotactic factors present in the supernatants were not as potent as fetuin-A. Also, suppression of fetuin-A in MIN6 cells failed to induce a significant migration of macrophages even after palmitate stimulation, which showed the role of islet-derived fetuin-A in promoting macrophage migration. Altogether the results from in vitro studies proved that fetuin-A secreted from MIN6 cells under palmitate stimulation brought about polarization and chemotactic migration in RAW 264.7 macrophages. Still, a lack of knowledge about the association of β-cell derived fetuin-A with islet inflammation prompted us to examine the status of islet-associated macrophages during HFD-induced T2D. For this study, we exposed mice to aHFD for 20 weeks and monitored the various parameters after every 4 weeks. Results from HOMA-IR, GTT and ITT displayed the progression of insulin resistance after 12 weeks of HFD. While the HFD mice could compensate for increased insulin demand up to 12 weeks of HFD, insulin secretion was impaired after 16 weeks of HFD, as indicated by in vivo and in vitro GSIS. This suggested that β-cells in mice kept sufficient plasticity to maintain their insulin secretory ability for up to 16 weeks of HFD administration, after which the β-cells failed to keep up with the insulin demand. Increased fetuin-A expression in the islet cells of HFD mice supported the fact that HFD feeding correlated with elevated level of fetuin-A in the islets. The progression of insulin resistance occurred concomitantly with increased content of fetuin-A, IL-1β, TNF and IL-6 in the islet microenvironment of HFD mice. Such loss of insulin secretion and impairment of β-cell function was attributed to the reduction in levels of Pdx1 and GLUT2 in the same subset. Other factors that led to decline of insulin production were caused by an activated TLR4–NF-κB pathway in β-cells. Since HFD is identified as one of the major reasons behind islet inflammation, we specifically focused on the expression of M1 type macrophage markers in islets (Ying et al., 2019). Several studies have reported increased number of macrophages in islets of T2D models (Ehses et al., 2007; Kamata et al., 2014; Richardson et al., 2009; Ying et al., 2019). In our current study, flow cytometry results depicted an increase in accumulation of M1 macrophages in the islets of HFD mice from 12 weeks of HFD feeding, but a prominent increase was marked after 16 weeks of HFD feeding. Data from immunofluorescence staining and gene expression analysis also supported the finding that there was an increase in islet-based M1 macrophages. There was a clear phenotypic shift towards the pro-inflammatory type in islet-resident macrophages of HFD mice, especially from 12 weeks of HFD onwards. Results from this study showed that a decrease in insulin secretion occurred simultaneously with increasing islet inflammation. The total process of islet macrophage accumulation, polarization and β-cell impairment was attributed to high fat feeding since no change was observed in macrophage accumulation and polarization status in SD mice of various age groups.
To study in detail the mechanism behind inflammation-mediated reduced insulin secretion in vitro, we also focussed on the impact of macrophages in β-cell function, particularly to understand the metabolic crosstalk between β-cells and islet-associated macrophages. We must keep in mind that macrophages are present within healthy islets and have a role in expansion of β-cell mass (Banaei-Bouchareb et al., 2006, 2004; Geutskens et al., 2005; Hume et al., 1983). On the other hand, progression of T2D is associated with an increased number of islet-resident macrophages (Ehses et al., 2007). Other reports demonstrated the presence of M1-like macrophages in the islets of T2D mice models, confirming the shift in polarization of islet macrophages in diabetic mice and the role of this shift in polarization in β-cell dysfunction (Cucak et al., 2014; Eguchi et al., 2012; Jourdan et al., 2013; Richardson et al., 2009; Westwell-Roper et al., 2014). Hence, we studied the effect of M1 and untreated RAW 264.7 macrophages upon the function of MIN6 cells. The present findings suggest that secretions from M1 type RAW 264.7 cells have a detrimental effect upon insulin secretion from MIN6 cells; however, no such effect was observed in MIN6 cells treated with conditioned medium from untreated RAW 264.7 cells. Immunoblotting revealed a marked decrease in the levels of Pdx1 and GLUT2 in MIN6 cells treated with medium from M1 type RAW 264.7 cells. The reduction in insulin secretion was further supported by gene expression and flow cytometry analysis, as there was a decrease in gene expressions of Pdx1, MafA, Beta2 and insulin and left shift of the insulin-positive cell population in the same subset of MIN6 cells. Taken together, these data confirmed that M1 RAW 264.7 macrophages led to β-cell dysfunction, through a reduction in Pdx1, MafA, Beta2 and insulin gene expression.
In conclusion, this present study demonstrated the effect of HFD in increased islet cytokine content, accumulation of islet-based macrophages and their polarization, along with their impact to reduced β-cell function, which indicated an overall trend that increased fetuin-A content in the islets is associated with islet inflammation (Fig. 7). This in vivo study was supported by in vitro studies where we showed that palmitate-mediated fetuin-A secretion from MIN6 cells stimulated polarization and migration of RAW 264.7 cells, which in turn had destructive effects on MIN6 function, by impeding insulin secretion. It must be noted that elevated fetuin-A content in the serum and islet microenvironment occurs during HFD along with collateral damage to β-cell function and increased islet inflammation. Under these circumstances, our present study is pertinent in revealing the pernicious involvement of hyperlipidemic condition and fetuin-A in macrophage polarization and infiltration in the islets and the resultant loss of β-cell function therein.
MATERIALS AND METHODS
Reagents and antibodies
Cell culture materials were purchased from Gibco-BRL/Life Technologies. Palmitate was bought from SRL, India. All other chemicals were bought from Sigma. The list of antibodies used are: anti-NF-κBp65 (51-0500), anti-pNF-κBp65(Ser536) (MA5-15160), anti-fetuin-A (PA5-47157), anti-TLR4 (48-2300), anti-TNF (P350), anti-IL-1β (BB-AB0155), anti-IL-6 (BB-AB0160), anti-Pdx1 (sc-25403), anti-GLUT2 (sc-9117), anti-insulin (53-9769-82), anti-CD11c (PA5-35326 and 553801) and anti-β actin (BB-AB0024) antibodies, which were purchased from Santa Cruz Biotechnology, BD PharMingen, BioBharati LifeScience Pvt. Ltd and Thermo Fisher Scientific. Alkaline phosphatase-conjugated anti-rabbit-IgG (A-3687) and anti-goat-IgG (A-4062) secondary antibodies were purchased from Sigma. Cy3-tagged anti-goat-IgG (AS015) and TRITC (Rhodamine)-tagged anti-rabbit-IgG (AS040) secondary antibodies were purchased from ABclonal Technology, US. The list of antibodies, including dilutions used, and siRNAs are given in Table S5. Insulin, IL-1β, IL-6, TNF and fetuin-A ELISA kits were purchased from RayBiotech, Norcross, USA. Gene-specific qPCR primers used in the study are stated in Table S6 and were procured from IDT, Coralville, Iowa, USA.
Cell lines and cell culture
The mice insulinoma MIN6 cell line and mouse macrophage cell line RAW 264.7 were purchased from the National Centre for Cell Science, Pune, India. MIN6 cells were cultured in Dulbecco's modified Eagle's medium (DMEM)/F12 with 15% (v/v) fetal bovine serum (FBS), penicillin (100 units/ml) and streptomycin (100 μg/ml) and RAW 264.7 cells were cultured in DMEM containing 10% (v/v) FBS (fetal bovine serum), penicillin (100 units/ml) and streptomycin (100 μg/ml) in a humidified atmosphere containing 5% CO2, with temperature maintained at 37°C. Palmitate solution was prepared by conjugating palmitate and bovine serum albumin in the ratio 1.35:2.0 (Dasgupta et al., 2010). The preparation was checked to be free of endotoxins (Pierce High-Capacity Endotoxin Removal Spin Columns). Untreated or control cells were cultured in absence of palmitate and fetuin-A for 4 h in serum-free medium. In a separate incubation, the RAW 264.7 cells were incubated with palmitate in doses of 0.25, 0.50, 0.75 or 1.0 mM and fetuin-A in doses of 10, 20, 40, 80, 100, 200, 400 or 800 µg/ml for 24 h. Untreated cells were used as negative control and, for a positive control, of M1 type RAW 264.7, the cells were treated with 100 ng/ml LPS for 24 h (Montana and Lampiasi, 2016). The RAW 264.7 cells were also treated with palmitate (0.50 mM), fetuin-A (100 µg/ml) and palmitate (0.50 mM) plus fetuin-A (100 µg/ml) to study the changes in markers of M1 and M2 type macrophages. LPS (100 ng/ml) was used on a separate set of cells as a positive control. The cells were stored for western blotting and gene expression studies and the medium was stored for western blot and ELISA analysis. MIN6 cells were treated with various doses of palmitate (0.25, 0.50, 0.75 or 1.0 mM) for 3 h and the medium was used for ELISA analysis for IL-1β, TNF and MCP1.
Silencing fetuin-A by siRNA interference
For siRNA transfection studies, fetuin-A siRNA was transfected to MIN6 cells by using Lipofectamine™ 2000 (Invitrogen) and cells cultured in opti-MEM, according to the protocol provided by the manufacturer. The cells were transfected with fetuin-A siRNA for 72 h followed by incubation without or with palmitate for 4 h for performing further experiments. The medium from fetuin-A KD MIN6 cells treated with palmitate (0.50 mM) was added to the RAW 264.7 cells for 24 h. The RAW 264.7 cells and the medium were preserved for ELISA analysis, western blot analysis, flow cytometry and qPCR.
Conditioned medium experiment with MIN6 and RAW 264.7 cells
The MIN6 cells were either left untreated or treated with palmitate (0.25, 0.50, 0.75 and 1.0 mM) for 4 h to stimulate fetuin-A secretion. The medium from MIN6 cells was then added to the RAW 264.7 cells and were incubated for 24 h. The medium was subjected to ELISA analysis, and cells were subjected to western blot, gene expression and flow cytometry analysis.
In another set of experiments, the RAW 264.7 cells were kept untreated or with palmitate (0.50 mM) plus fetuin-A (100 µg/ml) for 24 h to induce M1 polarization. At the end of the treatment the medium was removed and the RAW 264.7 cells were incubated with fresh incomplete medium without serum for 4 h. The medium was further treated to MIN6 cells for 4 h and the cells and medium were preserved for western blot, gene expression, flow cytometry and ELISA analysis.
Glucose-stimulated insulin secretion
For the GSIS assay, the above treated MIN6 cells were incubated in glucose-free Krebs–Ringer bicarbonate buffer containing NaCl (115 mmol/l), MgSO4 (1.2 mmol/l), KCl (4.7 mmol/l), KH2PO4 (1.2 mmol/l), NaHCO3 (20 mmol/l), HEPES (16 mmol/l), CaCl2 (2.56 mmol/l), and bovine serum albumin (0.2%) at 37°C for 1 h. Next, cells were incubated in Krebs–Ringer bicarbonate buffer supplemented with 2.8 mM or 22.4 mM glucose at 37°C for 1 h. The medium was subjected to ELISA analysis after completion of the treatment.
Immunoprecipitation and immunoblotting
For studying phosphorylation status of proteins briefly, 200 μg of total protein of RAW 264.7 or MIN6 or isolated islet cells were incubated with 3 μg anti-NF-κB antibody overnight at 4°C. Next the lysate was incubated with protein-A–agarose for 4 h. The immunocomplex was precipitated by centrifugation, then washed thoroughly, boiled in 4× sample buffer, vortexed and centrifuged at 4000 g for 10 min. The supernatant was collected and resolved on 10% SDS-PAGE followed by immunoblotting with antibodies specific to pNF-κBp65(Ser536) and NF-κB to study the change in phosphorylation normalized to a constant amount of total NF-κB (Pal et al., 2012).
Western blot analysis was performed as described by Pal et al. (2012). 80 µg of total protein obtained from cell lysates or medium was separated by using 10% SDS-PAGE gels. Then the proteins were transferred to PVDF membranes (Millipore, Bedford, USA) by using the Wet/Tank Blotting System (Bio-Rad Laboratories Inc, Hercules, USA). The membranes were incubated in blocking buffer for 1 h followed by incubation using various primary antibodies at the dilution ratio of 1:500 for overnight at 4°C (Table S5). The membranes were then properly washed and incubated with alkaline phosphatase-tagged secondary antibody at a 1:2000 dilution for 2 h. The membranes were then again properly washed and the protein bands were developed using 5-bromo 4-chloro 3-indolyl phosphate/nitrobluetetrazolium (BCIP/NBT) as substrate. The intensities of protein bands were assessed by Image Lab Software (Bio-Rad Gel DocTMXR+, USA). Densitometry analysis was conducted using ImageJ software.
The cells were lysed using TRI Reagent (Sigma) according to the manufacturer's protocol, to extract total RNA. cDNA was synthesized from 3 µg RNA by reverse transcriptase PCR using Revert AidTM first-strand cDNA synthesis kit (Fermentas Life Sciences). The changes in the gene expression were confirmed by real-time PCR (Applied Biosystems Quant Studio 5). Real-time PCR was performed using gene specific primers stated in Table S6, in the following reaction conditions: initial activation step at 95°C for 15 min, then 40 cycles of denaturation at 95°C for 30 s, annealing at 55°C for 30 s and final extension at 72°C for 30 s. Every cell set had a corresponding glyceraldehyde-3-phosphate dehydrogenase (Gapdh) amplification in separate reactions to be used as endogenous control. The Ct values were calculated using these corresponding gapdh values (Pal et al., 2012).
Cell migration assay
The RAW 264.7 cell migration assay was performed in a Boyden chamber (Millipore QCM 24-well colorimetric cell migration assay kit) with upper and lower chamber system (Fig. 3A). In the upper chamber, RAW 264.7 cells were incubated with culture medium in the lower chamber with (1) fetuin-A (100 µg/ml), MCP-1 (100 ng/ml), cytokine (TNF plus IL-1β; 10 ng/ml each) or palmitate (0.50 mM) conditioned medium from MIN6 cells, (2) palmitate (0.50 mM) conditioned medium from MIN6 cells containing antibodies for IL-1β, TNF and MCP1, (3) palmitate (0.50 mM) conditioned medium from MIN6 cells containing antibodies for IL-1β, TNF, MCP1 and fetuin-A, (4) conditioned medium from fetuin-A KD MIN6 cells, or (5) conditioned medium from 3 h palmitate (0.50 mM) treated fetuin-A KD MIN6 cells. The RAW 264.7 cells in the upper chamber were allowed to migrate for 4 h. At the end of the incubation the upper compartments (inserts) were removed from the system and stained in the staining solution provided in the kit for 20 min following the manufacturer's protocol. After a brief washing the inserts were air dried and photographed using a brightfield microscope (magnification 200×) (Leica DM 3000, Leica Microsystems GmbH). After viewing in the microscope, the inserts were dipped in de-staining solution following manufacturer's protocol to extract the dye. The OD of the extracted solution was measured at 560 nm and plotted.
Animals and treatments
Male Swiss albino mice aged 1–2 months and weighing 20–25 g were conditioned at a 25±2°C temperature and 12 h light–12 h dark cycle. Mice were divided into 10 groups: five standard diet (SD) fed groups and five high fat diet (HFD) fed groups; each group had n=5 animals. The control mice used for age match were kept on standard diet ad libitum having 57.3% carbohydrate, 18.1% protein and 4.5% fat. The five SD fed groups were maintained on a standard diet for up to 20 weeks. HFD groups were maintained on a HFD containing 15% carbohydrate, 20% protein, 65% fat for 20 weeks. At the end of 4, 8, 12, 16 and 20 weeks, one group at a time was killed, and the serum and tissues were collected for further analysis. All the experiments were performed following the guidelines prescribed by CPCSEA with the approval of the Internal Animal Ethics Committee, Visva-Bharati (IAEC/VB/2017/01).
The mice were fasted for 12 h and a glucose tolerance test (GTT) was performed by determining the concentration of blood glucose before and after gavaging 1 g glucose per kg body weight orally. Blood glucose level was measured after 0, 20, 40, 60 and 120 min using an Accuchek glucometer. An Insulin Tolerance Test (ITT) was performed similarly by injecting 0.7 U insulin per kg body weight following Pal et al. (2012), and then measuring the percentage change of blood glucose after 0, 20, 40, 60 and 120 min using an Accuchek glucometer. The insulin resistance index [i.e. homeostatic model assessment of insulin resistance (HOMA-IR)] was calculated according to the formula: fasting insulin (μU per litre)×fasting glucose (mg dl−1)/405 (Pal et al., 2012). For in vivo GSIS measurements, blood was collected from the tail vein at baseline insulin levels (0 min), and at 0, 5 and 35 min after glucose loading (1 g glucose per kg body weight). Insulin secretion levels were assessed by ELISA. Insulin levels at 5 min and 35 min of glucose loading was considered for in vivo GSIS (Krishnamurthy et al., 2007; Riopel et al., 2011).
Isolation and culture of islets
Isolation of pancreatic islets were undertaken by following the method of Meng et al. (2009). These isolated islets were purified by hand picking and plated in six-well culture plates and kept in humidified 5% CO2 atmosphere at 37°C by dissolving in serum-free medium. These freshly isolated islets were either left untreated or treated with palmitate (0.25, 0.50, 0.75 and 1.0 mM) for 4 h. The cell culture medium of these islets was used to treat peritoneal macrophages isolated from standard diet-fed mice.
Isolation and culture of peritoneal macrophages
The peritoneal macrophages were isolated from SD-fed mice and seeded in six-well tissue culture plates at 106 cells/well (Zhang et al., 2008). These macrophages were treated with medium from islets incubated with various doses of palmitate for 24 h. At the end of the treatment the macrophages and the cell culture medium were stored separately for qPCR and ELISA analysis, respectively.
Histology and immunofluorescence
The mice were euthanized and then their pancreas was removed. The tissues were fixed in 4% formaldehyde for 24 h and then stored in 70% ethanol. The tissues were dehydrated using alcohol gradient before paraffin embedding and block preparation. 54–56°C paraffin was used for embedding and block preparation. The blocks were cut at 5 µm thickness. The sections were heat fixed on Mayer's albumen-coated slides. The sections were subjected to hematoxylin and eosin staining (Fischer et al., 2008). For immunofluorescence, the sections of pancreatic tissue were deparaffinized at first followed by permeabilization, blocking with 10% (v/v) goat serum and co-immunostaining with insulin and fetuin-A primary antibodies or insulin and CD11c primary antibody as a marker of M1 macrophages. In case of insulin-fetuin-A co-immunostaining, the sections were simultaneously incubated with Alexa Fluor 488-tagged insulin and fetuin-A primary antibody for overnight at 4°C, then thoroughly washed and again incubated with Cy3-tagged secondary antibody for 2 h. In case of insulin and CD11c co-immunostaining, the sections were simultaneously incubated with Alexa Fluor 488-tagged insulin and CD11c primary antibody for overnight at 4°C then thoroughly washed and again incubated with TRITC-tagged secondary antibody for 2 h. DAPI was used as a nuclear stain. The images were taken at 200× magnification and 1.0× optical zoom in confocal laser scanning microscope (Leica TCS SP8).
Insulin, IL-1β, TNF, IL6, MCP1 and fetuin-A levels in the serum, islets and cell culture medium were measured with the help of ELISA kits of mouse insulin, IL-1β, TNF, IL-6 and fetuin-A (Raybiotech, Norcross, USA). ELISA was performed according to manufacturer's instructions.
Islet cells from SD and HFD mice or RAW 264.7 cells treated with palmitate (0.50 mM), fetuin-A (100 µg/ml), palmitate (0.50 mM) plus fetuin-A (100 µg/ml) or palmitate conditioned medium from MIN6 cells were double stained using FITC-tagged CD11c antibody and propidium iodide (PI). MIN6 cells treated with conditioned medium from untreated RAW 264.7 or M1 type RAW 264.7 cells were double immunostained with Alexa Fluor 488 tagged-insulin antibody and PI. The immunostained cells were analyzed using a flow cytometer (BD Accuri).
GRAPHPAD PRISM 8.0 (GraphPad Software, Inc., La Jolla, CA, USA) was used for statistical analysis. Data was analyzed using one-way analysis of variance (ANOVA) for multiple groups and two-tailed, unpaired Student's t-test for two groups, where the P value indicated significance. Means were compared by a post hoc multiple range test. All values are expressed as means±s.e.m. P<0.05 was considered as statistically significant.
The authors appreciate the use of facilities as extended by Prof. Samir Bhattacharya, Prof. Shelley Bhattacharya, Dr Surjya Kumar Saikia and the Head, Department of Zoology, Visva-Bharati University, India. We thank Mr Dipanjan Chattopadhyay and Mr Snehasis Das, Department of Zoology, Visva-Bharati University, for their assistance in Flow cytometry analysis.
Conceptualization: A.M.; Methodology: A.M., C.F.; Validation: A.M., C.F.; Formal analysis: C.F.; Investigation: R.K.; Resources: R.K.; Data curation: A.M., C.F.; Writing - original draft: A.M., R.K.; Writing - review & editing: A.M.; Supervision: R.K.; Project administration: R.K.; Funding acquisition: R.K.
This study was supported by grants from the Science and Engineering Research Board (SERB), Department of Science & Technology, Govt of India (Grant No. ECR/2017/001028/LS).
Peer review history
The peer review history is available online at https://journals.biologists.com/jcs/article-lookup/doi/10.1242/jcs.258507.
The authors declare no competing or financial interests.