All mouse experiments were approved by the Animal Research and Care Committee at the Children’s Hospital of Pittsburgh and the University of Pittsburgh IACUC. C57/BL6, INS1^Cre knock-in (18) and ROSA-mTmG mice (19) were all purchased from the Jackson Lab (Bar Harbor, MA, USA). Since male mice are aggressive, only female mice were used in the current study for parabiosis. PDL was done as previously described (16). Briefly, after a midline incision was made in the upper abdominal quadrant, the mouse pancreatic main duct at pancreatic neck region was located under a stereoscopic microscope with the help of a sterile swab. Here the pancreatic duct was ligated with 6-G suture at the left side of the portal vein between the gastro-duodenal and splenic lobes. The inferior pancreaticoduodenal artery, superior pancreaticoduodenal artery and pancreatic part of the splenic artery should be minimally affected by the ligation. Parabiosis was done as previously described (17). Briefly, two mice were surgically joined first through their joints at elbow and knee with 4-G suture and then attached of their skin to allowing a firm connection without skin strain. Fasting blood sugar was measured as previously described (20).

The digestion of the pancreas was performed as described previously (3). Briefly, pancreatic duct perfusion was performed, after which the pancreas was subsequently digested with 0.25 mg/ml collagenase (Sigma-Aldrich, Pittsburgh, PA, USA) for 30 minutes and with 10 μg/ml trypsin and 10 μg/ml DNase for approximately 20 minutes to obtain a single pancreatic cell population. Flow cytometry was done based on direct fluorescence for mT and CD45 or F4/80, using a pre-incubation with PE-cy7-conjugated anti-CD45 or anti-F4/80 (Becton-Dickinson Biosciences, San Jose, CA, USA), respectively.

Immunohistochemistry for insulin (21) and CD45 (10) was done as previously described. Briefly, mouse pancreas underwent fixation in 4% formalin for 4 hours, followed by cryo-protection in 30% sucrose for 36 hours. The cryo-sectioning was performed at 6μm thickness of the tissue. Primary antibodies for immunostaining were rat polyclonal anti-CD45 (Becton-Dickinson Biosciences) and guinea pig anti-insulin (Dako, Carpinteria, CA, USA). Indirect fluorescent staining was obtained with either Cy2- or Cy3- conjugated secondary antibodies (Jackson ImmunoResearch Labs, West Grove, PA, USA). Nuclear staining was done with Hoechst 33342 solution (Invitrogen, Carlsbad, CA, USA).

For in vivo experiments, 5 mouse pairs were used for each condition. Additional controls that were used for parabiosis ROSA-mTmG: INS1^Cre mice were INS1^Cre: INS1^Cre mice and INS1^Cre: INS1^CremTmG mice. Quantifications were done in at least 6 slides with a distance of 100µm from each other per mouse pancreas or at least 5000 cells per sample. All data were statistically analyzed by one-way ANOVA with a Bonferroni correction. All error bars represent S.D. (standard deviation). Significance was presented as * when p<0.05. No significance was presented as NS. N values are either shown by individual data or by indicated in the figure legends.

A ROSA-mTmG red fluorescent mouse was joined to an INS1^Cre knock-in mouse by parabiosis to establish a chimeric circulation. In this model, the red mT+ circulating cells from ROSA-mTmG mice gradually formed chimeric blood with the paired non-fluorescent INS1^Cre knock-in mice. After the chimeric circulation was established, PDL was performed on the INS1^Cre knock-in mice. Afterwards, the contribution of red cells derived from ROSA-mTmG mice to new beta-cells in the PDL-INS1^Cre knock-in mice was assessed ( Figure 1A ). This approach not only allowed for detection of direct conversion of circulation-derived cells into beta-cells, but also allowed for detection of cell fusion. In the event of cell fusion, the Cre recombinase from the beta-cells would activate the mG from the ROSA-mTmG locus. Thus, overall, insulin immunostaining and analysis of direct fluorescence (mT or mG) would determine potential contribution of BMCs to beta-cells in the PDL-treated inflamed pancreas ( Figure 1B ). The pre-existing beta-cells will stain positive for insulin, but negative for green mG or red mT fluorescence. If red cells from ROSA-mTmG mice do not stain positive for insulin, they did not directly contribute to beta-cells, but were merely present in the pancreas. If cells from the ROSA-mTmG mice stain positive for insulin, but remain red and do not become green fluorescent, they directly contributed to beta-cells and are not green because they do not have cre recombinase. If cells from ROSA-mTmG mice stain positive for insulin, but lose their red fluorescent and become green, they must have fused with pre-existing beta-cells ( Figure 1B ).

We analyzed the length of time required to generate a stable chimeric circulation in mice after parabiosis surgery ( Figure 2A ). For this analysis, blood was taken from each mouse of the pair at serial time points to determine the degree of chimerism. We did not detect red cells in INS1^Cre knock-in mice before parabiosis. Moreover, as early as day 3 after parabiosis, significant blood exchange was already detected in both animals ( Figures 2B, C ). At day 7 after parabiosis, the chimerism rate reached about 50% and became stable ( Figures 2B, C ). Thus, stable blood chimera is established within 7 days after parabiosis surgery. To ensure that the PDL was performed at a time when the circulation was fully chimeric, we performed the PDL 2 weeks after parabiosis.

We and others have previously shown that PDL does not alter blood glucose levels since half of the pancreas (head part of the pancreas) remains unaffected (3, 12–14). Here, we also followed the blood glucose levels up to 4 weeks after PDL. We did not detect alterations in blood glucose levels, suggesting that parabiosis itself does not have effects on the glycemia of PDL-mice ( Figure 3A ). CD45 immunostaining was performed on the INS1^Cre knock-in mice at different time points after PDL, showing similar inflammatory infiltration and histological alterations in these mice ( Figure 3B ), compared to PDL-pancreas without parabiosis (10, 16).

No mT red fluorescent cells were detected in the control INS1^Cre knock-in mice that were joined with another INS1^Cre knock-in mice in a parabiosis either before or after PDL (image 1 week after PDL is shown in Figure 4A ). Only rare mT red fluorescent cells were detected before PDL in INS1^Cre knock-in mice that were joined to the ROSA-mTmG mice by parabiosis. However, many mT red cells were detected in the PDL-pancreas starting from 3 days after PDL ( Figure 4A ).

We looked for mT red ( Figure 4B ) or mG ( Figure 4C ) beta-cells in the inflamed pancreas at different time points after PDL in INS1^Cre knock-in mice that were joined to the ROSA-mTmG mice by parabiosis, but we did not detect any. These data suggest that there is no direct contribution of BMCs to beta-cells in the PDL-pancreas.

Pancreatic macrophages have two origins, one is residential macrophages, and the other is circulation-derived macrophages. This parabiosis model allows examination of the contribution of these two populations to pancreatic inflammation. We assessed the pancreatic inflammatory cells by flow cytometry. First, the digested PDL-pancreas was analyzed at different time points for a pan-leukocyte marker CD45 and red fluorescence. Only rare CD45+ inflammatory cells were detected before PDL, but then CD45+ inflammatory cell numbers dramatically increased as early as day 3 after PDL, peaked at 1 week after PDL, and went down at 4 weeks after PDL ( Figures 5A, B ). Moreover, red CD45+ cells were just slightly less than half of total CD45+ cells ( Figures 5A, B ). Since the circulation is 50% derived from each mouse, these data suggest that most of the CD45+ cells in the inflamed pancreas were recruited from circulation, consistent with a previous study (22). Similar results were obtained using a macrophage-specific marker F4/80 instead of CD45 ( Figures 5C, D ), while the total percentage of F4/80+ cells was slightly lower than CD45+ cells ( Figures 5B, C ). These data suggest that the majority of inflammatory cells in the PDL-pancreas were recruited macrophages from the circulation.