Structural basis for delta cell paracrine regulation in pancreatic islets

Abstract

Little is known about the role of islet delta cells in regulating blood glucose homeostasis in vivo. Delta cells are important paracrine regulators of beta cell and alpha cell secretory activity, however the structural basis underlying this regulation has yet to be determined. Most delta cells are elongated and have a well-defined cell soma and a filopodia-like structure. Using in vivo optogenetics and high-speed Ca²⁺ imaging, we show that these filopodia are dynamic structures that contain a secretory machinery, enabling the delta cell to reach a large number of beta cells within the islet. This provides for efficient regulation of beta cell activity and is modulated by endogenous IGF-1/VEGF-A signaling. In pre-diabetes, delta cells undergo morphological changes that may be a compensation to maintain paracrine regulation of the beta cell. Our data provides an integrated picture of how delta cells can modulate beta cell activity under physiological conditions.

Fig. 1

Delta cell activity in vivo. Delta cells are activated by increasing glucose levels. Longitudinal imaging of in vivo activity in islets expressing the Ca²⁺ reporter GCaMP3 in delta cells. a Maximum projection images a representative mouse (top) and a human islet (bottom) immunostained with anti-glucagon, anti-insulin, and anti-somatostatin antibodies. Cell nuclei are shown in light blue. Graphs on the right show the cellular composition of mouse (n = 24 islets total, n = 7 mice) and human islets (n = 21 islets total, n = 3 donors). b Individual traces from four representative delta cells within a single islet in vivo and showing the profile of spiking and slow delta cells. Insulin (0.25 U/kg) was injected 2 min prior to glucose injection (0.4 g/kg i.v.). Gray-shaded area indicates a predetermined time zone used for data analysis and marks the glucose stimulation period. Basal and recovery periods are located to the left and right of the grey-shaded area, respectively. c Relative distribution (or frequency) of detected peaks in all spiking delta cells analyzed (n = 217) over the course of up to 18  min. Grey-shaded area represents the glucose-stimulated time period as described in b. d Normalized spiking delta cell activity peak amplitude for basal (n = 4574 peaks); glucose (n = 7850 peaks); and recovery (n = 10518 peaks) time periods. In e, same as in d, however the graph shows the breakdown of the normalized peak amplitude in the first or last 2 min of the glucose stimulation time period. f Hierarchical clustering analysis of the spiking delta cell activity under basal, the first or last 2 min of glucose-stimulated or recovery time periods. g Normalized delta cell peak amplitude in nonsynchronized (OFF) and synchronized (ON) peaks. In a, scale bar = 35 µm. Statistics: in c, *p = 0.0001 vs time 0 (d, e, and g) **p < 0.0001, (g) ^#p < 0.0001 vs respective OFF group. All comparisons made with OneWay Anova with a Tukey multiple comparison test. Error bars represent the 95% confidence interval (C.I.) of the mean. For a, c–e, and g, source data are provided as a Source Data file

Fig. 2

Cyto-architecture of delta cells in rodents and human islets. Delta cells send out cellular processes (filopodia) to increase their intra-islet reach. Maximum projection images of representative mouse (a) and a human delta cells (b) immunostained with anti-somatostatin antibodies. The yellow arrowheads indicate cellular extensions from delta cells. c Scanning electron microscopy of a human islet. Somatostatin-expressing delta cell is highlighted by a purple shade. Right, close-up image of the delta cell filopodia (highlighted by the cyan box in c) contacting an alpha and a beta cell. d Physical length of the delta cell filopodia in its longest axis (tip to cell body) in vivo in mouse islets (empty black circles, n = 47 cells) and in PFA-fixed pancreases from mice (black circles, n = 31 cells) and human (pink circles, n = 57 cells) samples. e Relative reach of single delta cells towards alpha and beta cells in mouse (black line) and human islets (purple line). f In vivo longitudinal follow up of a single delta cell at 10-min intervals or 19 h later. Delta cells were from SST-ChR2 islets and imaged with TPM to avoid rhodopsin activation. A yellow line highlights the delta cell morphology over time. g In vivo longitudinal follow up of a single delta cell filopodium over the course of 4 days as in f. Vessels were visualized by an i.v. injection of TRITC-labeled dextran molecules. Yellow arrowheads indicate the filopodium from the same delta cell from days 1 to 4. h Delta cell filopodia length from isolated islets treated with insulin, IGF, and VEGF-A with or without the IGF-1R receptor blocker picropodophyllin (PPP) or the VEGFR blocker Axitinib. Statistics shown for FDR (q < 0.005 for discovery): *q = 0.052, p = 0.0284; **q < 0.002, p < 0.001, ^#q < 0.002, p < 0.001 vs control (DMSO) group. Detailed statistics shown in Supplementary Data 1. Error bars represent the 95% confidence interval (C.I.) of the mean. For d, e, and h, source data are provided as a Source Data file

Fig. 3

Islet delta cell filopodia have independent Ca²⁺ spikes from the soma and contain secretory granules. Delta cell filopodia display Ca²⁺ spikes in vivo upon stimulation with glucose. a Representative Ca²⁺ dynamics of a delta cell soma (black line) and filopodia (pink) in vivo. Shaded grey area indicates the time point when glucose (0.4 g/kg) was injected i.v. b Close-up of the area highlighted in grey in a, glucose stimulation time point is indicated by a dashed line. Numbers on top of the peaks indicate the time point of the peak’s maximum amplitude. Values for the cell body are plotted on the left Y-axis while values for the filopodia are plotted on the right Y-axis. Dotted line indicates the time point when glucose was injected. c Maximum projection images of Ca²⁺ dynamics in delta cell body and filopodia, shown in a, b. Yellow squares mark the regions of interest represented in a, b. d Peak amplitude and e frequency in delta cells (n = 5) with soma and filopodia regions at basal, glucose, and recovery time points. f Peak synchronization between the delta cell soma and filopodia at basal, glucose, and recovery time points. g–i Single confocal slice of a delta cell filopodia from an SST-ChR2 islet incubated with FM-4-64 at 3 mM glucose. The delta cell filopodia plasma membrane was visualized by imaging YFP (g), FM-4-64 is shown in h and the overlay is shown in i. Graph on the right of i indicates the pixel profile of the region of interest (yellow line) drawn in i. YFP pixels (plasma membrane) are represented by a black line and plotted on the right Y-axis. FM-4-64-pixel profile is shown in pink and plotted on the left Y-axis. j Electronic tomogram of a delta cell filopodia in a mouse islet, cytoplasm is highlighted by a purple shade. Orange arrowheads indicate a microtubule and yellow arrowheads point to somatostatin granules. k 3D reconstruction of the volumetric data set shown in j. Filopodia plasma membrane is shown in white and somatostatin granules are shown in varying colors. Scale bars, c 10 µm, g–i 5 µm, and in j, k 500 nm. Error bars represent the 95% confidence interval (C.I.) of the mean. For d–f source data are provided as a Source Data file

Fig. 4

Optogenetic control of delta cell activity. Somatostatin reduces beta cell [Ca²⁺]_i oscillatory frequency and peak amplitude, and hyperglycemia is associated with impaired delta cell function and longer filopodia. a Photostimulation of delta cells expressing ChR2 evoked inward photocurrents under voltage clamp conditions (top; holding potential = −70 mV) and light-evoked, overshooting action potential-like responses under current-clamp conditions (bottom). Timing of light flashes (470 nm) is indicated by lower traces (L). Dashed line = 0 mV. b tdTomato and GCaMP6s/YFP fluorescence signal in wild-type and SST-ChR2 islets. c Top, protocol for in vivo imaging and ChR2-mediated photostimulation (450–470 nm light). Bottom, cartoon illustration representing the experimental setup and expected biological responses of GCaMP6⁺ beta cells and ChR2⁺ delta cells. d Representative beta cell [Ca²⁺]_i traces in islets from wild-type (top) or SST-ChR2 mice (bottom). e Normalized beta cell [Ca²⁺]_i peak amplitude and f peak frequency in wild-type (n = 7 islets) and SST-ChR2 islets (n = 5 islets) before and after ChR2 stimulation in vivo. e ***p < 0.001 by two-tailed student t-test and in f, ***p = 0.001 vs light-stimulated WT islets. In e, f data are displayed as fold change from nonstimulated conditions, represented by the horizontal pink dotted lines. Error bars represent the 95% confidence interval (C.I.) of the mean. For e, f source data are provided as a Source Data file

Fig. 5

Delta cell activity is impaired in prediabetes. a Representative normalized traces of GCaMP3 signals measured in vivo in delta cells from animals fed with control (CD, black lines) or HFD (orange lines). As in Fig. 1, the yellow-shaded area marks the time period used for measurements of glucose responses. b Spiking delta cell Ca²⁺ peak amplitude signal during basal, glucose, or recovery time periods normalized to basal conditions in CD (basal: n = 358; glucose: n = 588; recovery: n = 1060 peaks from n = 23 cells in four islets total and one islet per animal) or HFD (basal: n = 127; glucose: n = 224; recovery: n = 353 peaks from n = 14 cells in four islets total and one islet per animal for a total of four animals) animals. Data shown as fold change from the basal state. c Physical length of the delta cell filopodia in its longest axis (tip to cell body) from islets in situ from CD (n = 4 mice, 33 islets total, n = 33 delta cells) or HFD (n = 4 mice, 42 islets total, n = 42 delta cells) animals. Right, maximum projection images of delta cells acquired with confocal microscopy in CD or HFD animals. Scale bars, a 50  μm and (H), 10  μm. Statistics: b data shown for false discovery rate (FDR, q < 0.005 for discovery): **q < 0.002, p < 0.001 by OneWay Anova with a multi-comparison test using a two-stage linear step-up procedure of Benjamini, Krieger, and Yakutieli. In c, **p = 0.0038 by unpaired, two-tailed Student’s t-test. Error bars represent the 95% confidence interval (C.I.) of the mean. For b, c source data are provided as a Source Data file