Skip to main content

Quantitative 3D OPT and LSFM datasets of pancreata from mice with streptozotocin-induced diabetes: Sample data sets

Cite this dataset

Hahn, Max et al. (2022). Quantitative 3D OPT and LSFM datasets of pancreata from mice with streptozotocin-induced diabetes: Sample data sets [Dataset]. Dryad.


Mouse models for streptozotocin (STZ) induced diabetes probably represent the most widely used systems for preclinical diabetes research, owing to the compound’s toxic effect on pancreatic ß-cells. However, a comprehensive view of pancreatic β-cell mass distribution subject to STZ administration is lacking. Previous assessments have largely relied on the extrapolation of stereological sections, which provide limited 3D-spatial and quantitative information. This data descriptor presents multiple ex vivo tomographic optical image data sets of the full β-cell mass distribution in mice subject to single high and multiple low doses of STZ administration, and in glycaemia recovered mice. The data further include information about structural features, such as individual islet β-cell volumes, spatial coordinates, and shape as well as signal intensities for both insulin and GLUT2. Together, they provide the most comprehensive anatomical record of the effects of STZ administration on the islet of Langerhans in mice. As such, this data descriptor may serve as reference material to facilitate the planning, use and (re)interpretation of this widely used disease model.


Animals, STZ administration and organ isolation

The data presented in this data descriptor was acquired for Hahn et al., 2020, and all experiments were performed following the European Union guidelines for care and use of animals in research. All procedures were approved by the Animal Review Board at the Court of Appeal of Northern Norrland and of Northern Stockholm. Streptozotocin (STZ, Sigma-Aldrich) dissolved freshly in 0.1 M sodium citrate buffer (pH 4.5) was administered to 8-week-old male C57BL/6J mice by intraperitoneal (i.p.) injection, either as a single high dose (SHD, 150 mg/kg) or as multiple low doses (MLD, 50 mg/kg over 5 consecutive days). For data reliability, untreated control mice were compared with mice receiving an i.p. injection of the vehicle solution only (0.1 M sodium citrate, pH 4.5). No difference in BCM, islet number or blood glucose levels could be detected (see Hahn et al., 2020, suppl. Fig 2). Glucose measurements were regularly performed from tail vein blood with OneTouch (LifeScan, USA) or Accu-Chek (Roche, Switzerland) glucometers until death (see Table 1 and 2). Animals were killed by cervical dislocation and pancreata from diabetic groups of SHD or MLD and healthy control groups were isolated at 1-,2- and 3-weeks post administration of STZ (n=3-5, n=3-5 and n=5 respectively). Harvested pancreata were fixed in 4% paraformaldehyde (PFA, Sigma Aldrich) for 2 h, washed in 1x PBS and divided into the splenic, gastric, and duodenal lobular compartments (see also Fig. 2) before processing for whole mount immunohistochemistry and 3D imaging.

To delineate the long-term effects of hyperglycemia on GLUT2 expression, on β-cell function and islet size distribution generated data from islet transplantation experiments (see Fig. 2, data set 2) was performed as previously described. In short, pancreatic islets of Langerhans were obtained from healthy (normoglycemic) mice with the same genetic background via collagenase treatment. SHD-treated animals with the highest blood glucose levels 4 days post-STZ administration were then transplanted with 100 – 150 islets per animal into the anterior chamber of the eye under isoflurane anaesthesia to revert hyperglycemia. Once the SHD treated (n=8) and islet transplanted cohort (SHD+Tx, n=4) reached normoglycemic levels to the control (n=7), organs of all animal cohorts were harvested (28 days post-STZ administration, see above).

Pancreas processing, whole mount immunohistochemistry and tissue clearing.

Tissue processing, staining procedure, and preparation for OPT/LSFM imaging were performed as described. In brief, isolated and fixed pancreata were separated into the main lobes (SL, GL and DL respectively, see Fig. 2) permeabilized by freeze/thawing cycles, bleached to reduce autofluorescence, stained with primary and secondary antibodies, mounted in a cylinder of low melting point agarose, dehydrated with methanol and made transparent by matching the refractive index of proteins, lipids, and other cellular components with a 1:2 mixture of benzyl alcohol and benzyl benzoate (BABB), respectively. All specimens were blinded and randomized after organ harvest for all downstream processes. Primary antibody used was guinea pig anti-insulin (DAKO A0594, dilution 1:500), and secondary antibody was goat Alexa 594 anti-guinea pig (Molecular Probes, A11076, dilution 1:500). For co-expression assessments of insulin and GLUT2 (see Fig. 1 and Fig. 2, Dataset 2), pancreata were in addition to insulin labelled with primary rabbit anti-GLUT2 (Millipore, 07-1402-l, dilution 1:500) and secondary IRDye 680RD goat anti-rabbit (Licor, 926-68071, dilution 1:500).

3D imaging: Optical projection tomography (OPT) and Light sheet fluorescent microscopy (LSFM)

OPT scanning of pancreatic specimen (Dataset 1) was performed as described using a Bioptonics 3001 OPT scanner (SkyScan, Belgium) with varying exposure times (see folder “Metadata for all groups” for exposure times) of Insulin staining (filter set “insulin”: Ex:560/20nm, Em.:610nm LP) and autofluorescence (filter set “anatomy”: Ex: 425/20nm, Em.:475nm LP). The image data was generated using SkyScanner 3001 (v1.3.13, SkyScan). Samples from co-expression experiments (Dataset 2) were scanned in our custom build Near Infrared-OPT setup using LabVIEW (v20.0f1) to retrieve image data. For comparison of intensities in 3D, all images in dataset 2 were generated using equal exposure times of Insulin staining (filter set: Ex: HQ 565/30 nm, Em: HQ 620/60 nm, exp. t = 4000 ms), GLUT2 staining (filter set: Ex: HQ 665/45 nm, Em: HQ 725/50 nm, exp. t = 8000 ms) and endogenous fluorescent anatomy (filter set: Ex: 425/60 nm Em: LP 480 nm, exp. t = 500 ms).

Additional high-resolution scans (Dataset 3) of volumes of interest from representative pancreata that were OPT scanned (see above) were reimaged in a LaVision biotech 2nd generation UltraMicroscope (LaVision BioTec BmbH, Germany) with a 1x Olympus objective (Olympuse PLAPO 2XC) coupled to an Olympus MVX10 zoom body, providing between 0.36x and 6.3x magnification with a lens corrected dipping cap MVPLAPO 2x DC DBE objective. Samples mounted in low melting point SeaPlaque Agarose (39346-81-1, Lonza) were trimmed in BABB to fit the LSFM sample holder. Scans were acquired using 6.3 x magnification, which rendered a pixel size of 0.48 µm in x and y dimensions. Depending on the scan locations, the exposure time was 120-300 ms, light sheet-width was between 10-20% with 3.78 µm thickness (NA of 0.14) with a z-step size of 5 µm. Image data acquisition was performed using ImSpectorPro (version 5.0.164, LaVision BioTec GmbH, Germany). Representative islets of Langerhans with different sizes and locations in the gland were chosen based on 3D rendered OPT data sets.

Image processing, reconstruction, and 3D volume rendering

Insulin-based projection views retrieved from the Bioptonics 3001 scanner (Dataset 1) and volumetric assessments on β-cell volumes retrieved from the custom build NIR-OPT scanner (Dataset 2) were first processed with a contrast limited adaptive histogram equalization (CLAHE) algorithm, with a tile size of 64 x 64 to increase the signal-to-noise ratio for downstream islet segmentation. Secondly, a discrete Fourier transform alignment (DFTA) was performed to align opposing projection images to the same axis of rotation of a sample. However, for combined assessments of insulin and GLUT2 expression (Dataset 2) and analysis of the effect of STZ on GLUT2 staining intensity in β-cells, the CLAHE normalisation routine was not implemented. Reconstruction of OPT projection views to tomographic sections was performed using a filtered back projection algorithm in the NRecon software (V1.6.9.18, Bruker microCT, Belgium) with ring artefact correction set to 4. Resulting tomographic sections (*.bmp and *.tif Datasets 1 & 2, Record B) and raw z-sectional images (*.ome.tif, Dataset 3, Record B) generated with the Ultramicroscope II (LSFM) of each channel were converted with Imaris converter (Bitplane, UK) and the subsequential *.ims files of each channel for each sample were incorporated into one Imaris file. Individual insulin-positive islet volumes and lobular anatomies were quantified using an automated surfacing algorithm within the Imaris software (version 9.3.1, Bitplane, UK). Surface segmentation was performed using the ‘background subtraction’ function in Imaris with varying thresholding between samples. Individual threshold values for islet volume segmentation are displayed in Supplementary Table 1 and Supplementary Table 2 for Datasets 1 and 2, respectively. Surfaced islet volumes were arbitrarily categorized into small (<1 x 106 µm3), medium (1-5 x 106 µm3) and large (>5 x 106 µm3) islets of Langerhans as previously described. Volumes of 10 voxels or less were filtered out from quantification data sets (Datasets 1 and 2) to avoid inclusions of artefacts in the data analysis.

Usage notes

This is a representative sample of all data sets and data records.

Raw image projections (*.tiff) and tomographic sections (*.bmp or *ome.tiff) can be converted and imported into most 3D visualisation/quantification software such as ImageJ (, NIH, USA), Arivis (Munich, Germany), Imaris (Bitplane, UK) or 3D slicer ( When importing multiple channels generated by OPT or LSFM the numerical value of the starting image of the projection views (OPT) or z-stacks (LSFM) should always be the same for the anatomy, insulin or GLUT2 channels.

The Imaris files (*.ims) contain all iso-surfaces (anatomy, insulin or GLUT2) used for quantification and numerical data mining and can be visualised for free using the Imaris viewer (

Code availability

Custom generated scripts used for processing OPT data including COM-AR (alignment of axis of rotation during OPT scan setup), DFTA (alignment of axis of rotation post-OPT scanning) and CLAHE (improving islet segmentation) is compiled as a software package (together with video instructions on their implementation) at GitHub, Link

There are two other Dryad datasets associated with this dataset: Dataset 1 ( and Datasets 2 & 3 (


Swedish Research Council

Kempe Foundation

Umeå University

Lenanders stiftelse

Karolinska Institutet, Award: The strategic Research program in Diabetes

Novo Nordisk Foundation

Swedish Diabetes Association

Knut and Alice Wallenberg Foundation

Diabetes Research and Wellness Foundation

AF Jochnick Foundation

Familjen Erling-Perssons Stiftelse

Berth von Kantzow's Foundation

Skandia Insurance Company

European Commission, Award: 289932