We created an insulin-HaloTag construct (insulin-HT) in which human insulin was fused with a 33 kDa protein tag. The resulting expression plasmid was transfected into the mouse pancreatic beta cell line MIN6, and the intracellular localization of the fusion protein was first analyzed by transient incubation with the cell membrane-permeable red fluorescent dye HT-TMR. TMR fluorescence of insulin-HT nearly entirely overlapped with green signals arising from insulin-enhanced green fluorescent protein (EGFP) fusion protein (data not shown), whereas TMR signals were not detectable in untransfected cells, suggesting that the fluorescent signal on vesicles was specific. Because DNA transfection into MIN6 cells is not efficient by conventional transfection methods, we attempted to establish insulin-HT-expressing cell lines (Methods
). Consequently, we established several stable cell lines and confirmed the steady state distribution of insulin-HT. Immunostaining using anti-secretogranin III antibody revealed that insulin-HT labeled by TMR probe colocalizes with this SG-resident protein (1A). Furthermore, immunofluorescence analysis with anti-HaloTag antibody confirmed that insulin-HT primarily targets to IA-2-positive SGs but not to lysosomes (lamp I-positive) in this stable clone (). To further examine the localization of insulin-HT, we subjected cellular membranes to ultracentrifugation in a sucrose density gradient under conditions that separate the two types of secretory vesicles in beta cells. Insulin-HT primarily appeared in fractions 10 to 12, which corresponded to a peak for insulin immunoreactivity and the distribution of endogenous rab27A/B and secretogranin III but not to synaptophysin (). These observations suggest that insulin-HT fusion protein specifically targets and localizes to insulin SGs in the stable cells.
Insulin-HT targets to secretory granules in MIN6 cells.
To visualize the trafficking pathway of insulin-HT, pulse-chase experiments were performed with the stable cell line. Pre-existing insulin-HT was first masked by blocking probe and the cells were then labeled with TMR at reduced temperature (15°C), conditions where secretory proteins are known to accumulate in the ER. We observed an ER localization of TMR-labeled insulin that was newly synthesized within 1 hour (Fig. S2
and , upper panels). After the temperature was shifted to 37°C, TMR-labeled insulin appeared to transfer from the ER to the Golgi (Fig. S2
). At 15 minutes’ incubation at 37°C, insulin-HT was found at the TGN or ISGs where the EGFP-fused M6PR colocalized (, middle panels), and by 30 minutes, most TMR signals showed an identical localization with phogrin on SGs (, lower panels).
Biosynthetic pathway of insulin-HT monitored by fluorescent microscopy.
We next used two different fluorescent probes, TMR and R110, to distinguish newly synthesized insulin-HT. R110 direct probe is a cell membrane-permeable green fluorescent dye that can label target proteins upon overnight incubation. Following R110 labeling of pre-existing insulin-HT, newly (~ 1.5 hours) synthesized insulin-HT was labeled by TMR in the stable cells, whereupon fluorescent signals were observed by confocal microscopy. R110-labeled old insulin-HT signals showed a punctate pattern throughout the cytoplasm that largely did not overlap with new insulin-HT TMR signals (). This result is consistent with previous observations using atrial natriuretic factor-tagged fluorescent timer protein, in which secretory vesicles segregated into distinct populations according to age in bovine adrenal chromaffin cells 
. In contrast to chromaffin cells, in which young rather than older secretory vesicles preferentially docked to the plasma membrane, newly synthesized insulin-HT TMR signals were rarely observed in sections either from the top and bottom z-axis stages in MIN6 cells, whereas R110 signals from older insulin were present throughout the stages (3, and Fig. S1B
). A rough estimate of peripheral localization of TMR or R110 vesicles revealed that the proportion of new SGs was lower than that of old SGs (). We then employed total internal reflection fluorescence (TIRF) microscopy to confirm this observation. In contrast to R110 signals, TMR-positive signals were very sparse in the evanescent field images obtained by TIRF, but were evident in the epifluorescent images (). These results suggest that young SGs containing newly synthesized insulin are located inside the cell with few found at the plasma membrane.
Secretory granules containing newly synthesized insulin are located at the interior of the cell.
We next examined glucose-induced insulin secretion using stable cells and compared newly synthesized insulin-HT with older insulin as a function of the secretion rate. As the R110 probe is insufficient to label all pre-existing insulin-HT, we designed and performed two sets of pulse-chase experiments, using a blocking probe instead of R110 (Methods
) (, left scheme). Pulse-chase TMR-labeled insulin from the culture media and cell lysates was measured by fluorescence spectrophotometry, and the secretion rate in response to high glucose (25 mM) stimulation was compared for newly synthesized insulin-HT and older insulin. By 60 min, 36.9% of newer insulin-HT was released, whereas only 9.3% of the older protein was found in the secretion media (, right graph). Similar pulse-chase experiments using HT-biotin probe were examined, and biotin-labeled insulin was detected by immunoblotting with anti-biotin antibody (s, Fig. S4
). In this assay, 33.7% of newly synthesized insulin-HT was secreted into the extracellular spaces, with only 7.4% of older proteins being released (, right graph). Taken together, these data suggest that newly generated insulin SGs in MIN6 cells undergo exocytosis in preference to older SGs.
Newly synthesized insulin is preferentially secreted in response to high glucose.