Autofluorescence imaging of living pancreatic islets reveals fibroblast growth factor-21 (FGF21)-induced metabolism

Abstract

Fibroblast growth factor-21 (FGF21) has therapeutic potential for metabolic syndrome due to positive effects on fatty acid metabolism in liver and white adipose tissue. FGF21 also improves pancreatic islet survival in excess palmitate; however, much less is known about FGF21-induced metabolism in this tissue. We first confirmed FGF21-dependent activity in islets by identifying expression of the cognate coreceptor Klothoβ, and by measuring a ligand-stimulated decrease in acetyl-CoA carboxylase expression. To further reveal the effect of FGF21 on metabolism, we employed a unique combination of two-photon and confocal autofluorescence imaging of the NAD(P)H and mitochondrial NADH responses while holding living islets stationary in a microfluidic device. These responses were further correlated to mitochondrial membrane potential and insulin secretion. Glucose-stimulated responses were relatively unchanged by FGF21. In contrast, responses to glucose in the presence of palmitate were significantly reduced compared to controls showing diminished NAD(P)H, mitochondrial NADH, mitochondrial membrane potential, and insulin secretion. Consistent with the glucose-stimulated responses being smaller due to continued fatty acid oxidation, mitochondrial membrane potential was increased in FGF21-treated islets by using the fatty acid transport inhibitor etomoxir. Citrate-stimulated NADPH responses were also significantly larger in FGF21-treated islets suggesting preference for citrate cycling rather than acetyl-CoA carboxylase-dependent fatty acid synthesis. Overall, these data show a reduction in palmitate-induced potentiation of glucose-stimulated metabolism and insulin secretion in FGF21-treated islets, and establish the use of autofluorescence imaging and microfluidic devices to investigate cell metabolism in a limited amount of living tissue.

Figure 1

KLB expression and FGF21-dependent responses in pancreatic islets. (A) The cDNA from two separate mouse islet preparations (Islets 1 and 2) were amplified using oligonucleotide primers designed to recognize the N- (KLB-front) and C-terminal (KLB-end) ends of KLB. GAPDH cDNA was amplified to ensure sample integrity, and water was included as a no-DNA negative control. (B) Pancreatic islets were fixed and immunofluorescently labeled using an antibody specific for the extracellular domain of KLB (left panels). Dispersed inlet cells were also immunofluorescently labeled with both KLB (red) and insulin (green) (right panels). Immunostaining in the absence of primary antibody (No Primary) was included as a negative control. Scale bar represents 50 μm. (C) A representative Western immunoblot reveals a reduction in mouse islet ACC protein level when incubated in the presence of FGF21 (48 h; 100 ng/ml). Membranes were stripped and reprobed for β-actin as a loading control. (D) The summarized fold-ACC response normalized to β-actin for control and FGF21-treated islets. Data shown represent the mean ± S.E.M. for islets from five independently assayed and treated mice. (^∗P < 0.05).

Figure 2

Autofluorescence imaging of living pancreatic islets in a microfluidic device. Pancreatic islets were routinely brought into microfluidic devices before imaging the two-photon NAD(P)H and confocal LipDH(mNADH) responses. (A) A representative microfluidic device is shown flooded with food coloring to highlight the inlet and outlet tubing, main channel, dam wall, and reservoir. The device is composed of polydimethylsiloxane bonded to a glass coverslip, which provides an optimized optical window. Inlet- and outlet-tubing were used to bring islets into the microfluidic device and for removing effluent, respectively. Once 3 to 10 islets were loaded into the device, the inlet tubing was capped and flow was initiated from the on-chip reservoir to the outlet tubing. Islets captured in the device were held against the coverslip while media was allowed to flow past. (B) A representative islet stimulated with 2 mM glucose was imaged using two-photon 710 nm excitation and confocal 488 and 488 nm excitation (top row). This same islet was subsequently imaged after treatment with 3 mM sodium cyanide for 5 min (bottom row). Scale bar represents 20 μm. (C) The summarized NAD(P)H and 458:488 nm image ratio responses to 2 mM glucose, 3 mM cyanide, and 2 μM FCCP. The NAD(P)H data is plotted as the fold-response to 2 mM glucose. The 458:488 nm ratio plotted as arbitrary units (left axis) is a readout of the LipDH(mNADH) redox state. These pharmacological treatments induce maximal and minimal cellular redox state and were subsequently used to index the LipDH or mitochondrial NADH response (LipDH(mNADH)) (right axis). Data represent the pooled response from 13 and 30 islets harvested from 2 and 3 mice on separate days for the sodium cyanide and FCCP data, respectively (^∗P < 0.01).

Figure 3

Glucose-stimulated NAD(P)H responses. Pancreatic islets were cultured for 48 h in full RPMI media 1640 at 11 mM glucose in the absence (control) and presence (FGF21) of FGF21. Palmitate (0.4 mM) was added to a separate batch of islets after 24 h to induce palmitate toxicity. All islets were subsequently incubated in imaging media containing 2 mM glucose at 37°C (minimum 30 min), and loaded into microfluidic devices on the microscope stage to image the two-photon NAD(P)H response to varied glucose levels (as indicated). (A) Representative images of NAD(P)H autofluorescence for a device-immobilized islet at 2 and 20 mM glucose. (B) The summarized mitochondrial NAD(P)H intensities throughout the glucose dose-response from islets cultured in the absence of palmitate. Data represent the pooled response from 20 to 30 islets harvested from 3 mice on separate days. (C) The summarized mitochondrial NAD(P)H intensities throughout the glucose dose-response from islets cultured for 24 h in the presence of palmitate (0.4 mM). Data represent the pooled response from 15 to 20 islets harvested from 3 mice on separate days. (^∗P < 0.05).

Figure 4

Glucose-stimulated NAD(P)H and LipDH(mNADH) responses. Pancreatic islets were cultured for 48 h in full RPMI media 1640 at 11 mM glucose in the absence (control) and presence (FGF21) of FGF21. Islets were subsequently incubated in imaging media containing 2 mM glucose or 2 mM glucose + 0.4 mM palmitate (37°C; minimum of 30 min) followed by loading into microfluidic devices on the microscope stage to image the glucose-stimulated NAD(P)H and LipDH (mNADH) redox index responses. All responses were measured at least 20 min after subsequent addition of the indicated substrates. (A) The summarized fold NAD(P)H response of mitochondrial regions from islets exposed to glucose (2, 10, and 20 mM), high glucose (20 mM) in the presence of palmitate, and low glucose (2 mM) in the presence of citrate. (B) The summarized LipDH(mNADH) redox index for the same islets shown in (A). The LipDH response (458:488 nm intensity ratio) was indexed to pharmacological treatments that minimize and maximize mitochondrial NADH reduction. Data were collected from control islets (n = 21 (glucose), 14 (palmitate), and 21 (citrate)) and FGF21-treated islets (n = 28 (glucose), 14 (palmitate) 22 (citrate)) harvested from mice on 3 separate days. (^∗ and # indicate P < 0.05 compared to 2 mM and 2mM + FGF21, respectively).

Figure 5

Glucose-stimulated changes in mitochondrial membrane potential and insulin secretion. Pancreatic islets were cultured for 48 h in full RPMI media 1640 supplemented with 11 mM glucose in the absence (control) and presence (FGF21) of FGF21. Islets were subsequently used for measurement of mitochondrial membrane potential or measuring glucose stimulated insulin secretion. (A) Islets were incubated in imaging media containing 2 mM glucose (37°C for 1 h) followed by addition of Rh123 (10 μg/ml). Labeled islets were loaded into microfluidic devices on the microscope stage for imaging. Representative images are shown of the same Rh123-labeled control islet subsequently stimulated with 2 mM glucose (5 min), 2 mM glucose + 0.4 mM palmitate (25 min), and 20 mM glucose + 0.4 mM palmitate (5 min). (B) The summarized data from islets in glucose alone (2, 10, and 20 mM), and glucose in the presence of 0.4 mM palmitate (20 mM glucose, and 20 mM glucose + 10μM etomoxir) is reported as fold Rh123 intensity (arbitrary units, AU) relative to 2 mM glucose. Data was collected from control (n = 20–25) and FGF21-treated (n = 20–25) islets harvested from mice on 3 separate days. (^∗P < 0.05). (C) Islets were further incubated in imaging media containing 2 mM glucose (37°C for 30 min) and effluent was sequentially collected following 1 h incubations with the indicated nutrients. Sample supernatant (effluent) was carefully collected and assayed. The fractional insulin response from control and FGF21-treated islets in response to glucose alone (2, 10, and 20 mM glucose) and glucose in the presence of 0.4 mM palmitate (Control + 0.4 mM palmitate and FGF21 + 0.4 mM palmitate) are shown. Data were normalized to total insulin content collected post islet permeabilization with 1% triton X-100. Data shown are summarized from the islets harvested from 4 to 8 mice on independent days. (^∗P < 0.05).