Kit^MerCreMer /R26^YFP mice were generated as previously described (9). C57BL/6J, CX3CR1^GFP and CCR2^-/- mice (B6.129S4-Ccr2tm1Ifc/J) were obtained from The Jackson Laboratory (Bar Harbor, ME, USA). Only male mice were used for all experiments performed. All mice were bred and maintained in a specific pathogen-free animal facility of the Nanyang Technological University (Singapore).

To induce obesity and T2DM, 6-8 week-old mice were fed with a western diet (WD) composed of 40 kcal % of fat and 43 kcal % of carbohydrate (Research Diet, D12079Bi, New Brunswick, NJ, USA). The body weight was monitored on a weekly basis and after 24 weeks, the mice reached on average a bodyweight between 45-50 g and were sacrificed for collection of pancreas.

Kit^MerCreMer /R26^YFP fate-mapping mice were used to trace the ontogeny and turnover rates of distinct myeloid cell subsets. A total of 4 mg tamoxifen (TAM) (T5648; Sigma-Aldrich, St. Louis, MO, USA) per mouse was administered for five consecutive days by oral gavage for adult labelling as previously described (9). After TAM administration, c-kit⁺ hematopoietic stem cells (HSCs)/progenitors localized in the bone marrow (BM) of Kit^MerCreMer /R26^YFP mice were labelled with yellow fluorescent protein (YFP⁺) and all cells deriving from YFP⁺ BM-HSC/progenitors would maintain YFP expression when seeding the peripheral tissues/organs. This effect makes it possible to delineate macrophages that are derived from adult BM-definitive haematopoiesis.

The pancreas was perfused through the bile duct with 250 μg/ml collagenase P (Sigma Aldrich, St. Louis, MO, USA) diluted in HBSS (Sigma Aldrich). After perfusion the pancreata were removed, cut into small pieces digested with collagenase P at 37°C for 15 minutes. After vigorous shaking for 1 min, the homogenous mixture was passed through a metal sieve and larger chunks were further broken down through grinding. Subsequently, the mixture was poured through a 70 μm cell filter allowing exocrine cells passing through, whereas islets remained retained in the strainer. The islets were transferred into Petri dishes, handpicked under a microscope and incubated 10 min at 37°C with Accutase^® cell detachment solution (Biolegend, San Diego, CA, USA) to obtain a single cell solution. The exocrine pancreas was separately incubated with DNase I (20 U/ml) (Life Technologies, Carlsbad, CA, USA) at 37°C for 1 hour. The obtained cells were passed through a 100 μm filter and resuspended in NH₄CL red cell lysis buffer for 5 min to eliminate contaminating erythrocytes.

Endocrine and exocrine pancreatic single-cells were pre-incubated with 10 μg/ml anti-Fc receptor antibody for 15 min on ice at 4°C in the dark, further incubated with fluorochrome-labelled antibodies at 4°C for another 20 min, and then washed and re-suspended in PBS 2% for analysis on five-laser flow cytometers (FACSymphony™; BD Bioscience, San Jose, CA, USA). The following antibodies were used: APC-Cy7-labelled CD45 (30F11), BV605-labelled Ly6C (HK1.4), BV510-labelled CD11c (N418), Alexa Fluor700-labelled CD206 (C068C2), Percp-Cy5.5-labelled CD301 (LOM-14), APC-Cy7-labelled CD11b (M1/70), Pe-Cy7 and Alexa Fluor647-labelled Tim-4 (RMT4-54) all purchased from Biolegend. PE-labelled F4/80 (BM8), eFluor450-labelled MHC class II (M5/114.15.2) and PE-Cy7-labelled CD11c (N418) all purchased from ThermoFisher Scientific (Waltham, MA, USA). Percp-Cy5.5-labelled CD11b (M1/70) was purchased from BD Biosciences and APC-labelled anti-CCR2 was purchased from R&D Systems (Minneapolis, NM, USA).

The pancreas was fixed in 10% formalin at autopsy followed by dehydration in 30% sucrose and then embedded in Optimal Cutting Temperature (O.C.T.) compound (Tissue Tek, Sakura Finetek, Torrance, CA, USA) for subsequent analysis. ten μm sections were cut and stained with anti-mouse insulin antibody (Cell Signaling Technology Inc., Danvers, MA, USA) followed by Donkey anti-rabbit Alexa Fluor647 (Biolegend) and analysed by a Zeiss Live Cell Observer microscope. The islet size was examined by compiling measurements of 30 insulin-stained pancreas islets using ImageJ and Zeiss Zen Blue software. Islet macrophages were stained with rabbit anti-Iba1 antibody (dilution 1:400, FUJIFILM Wako Chemicals, Richmond, VA, USA) followed by donkey anti-rabbit Alexa Fluor647 (dilution 1:300, Biolegend). Their numbers were manually counted from 18-21 islet images taken with the microscope. Three animals for each group were analysed.

Kit^MerCreMer /R26^YFP fate-mapping mouse was transcardially perfused with 15 ml 4% paraformaldehyde. The pancreas was excised, immersed in 4% paraformaldehyde overnight, followed by dehydration in 30% sucrose before being embedded in OCT. Eight μm sections were cut and stained with anti-Iba1 antibody followed by donkey anti-rabbit Alexa Fluor647. Images were taken by a Zeiss Live Cell Observer microscope.

Flow cytometry data were analysed using FlowJo 10 (TreeStar, Ashland, OR, USA). Microscopy data were processed with Zeiss Zen Blue and ImageJ. For comparison between two groups, Student’s t-test was used to determine the level of significant differences. One-way ANOVA analysis was performed for multiple comparisons of more than two groups. Data were plotted using Prism 9.1.2 (GraphPad Software, La Jolla, CA, USA).

Pancreatic resident macrophages are present in both endocrine and exocrine areas. The exocrine pancreatic tissue comprises a heterogeneous macrophage family. The F4/80^hi tissue-resident macrophages are subdivided into three clear subpopulations based on the expression of embryonic-marker Tim-4 and MHC II (Tim-4⁺MHCII^low, Tim-4⁺MHCII⁺ and Tim-4^-MHCII⁺) (Figure 1A). Only Tim-4 expressing cells co-express mannose receptor (CD206) and the lectin CD301, whereas the majority of Tim-4^- cells do not (Figure 1B). In addition, other myeloid cells such as Ly6G⁺ neutrophils, Siglec F⁺ eosinophils, CD11c⁺MHCII⁺ dendritic cells (DCs), Ly6C⁺ monocytes and F4/80^intLy6C^-MHCII⁺ monocyte-derived macrophages are present in the exocrine pancreas, although all in lower frequency (Figure 1D). In the islets, according to previously published work (12), the macrophages are the main CD45⁺ cell fraction (> 90%). They represent a small homogenous population expressing F4/80, CD11b, MHC II and CD11c (Figure 1C), but negative for Tim-4 and CD206.

In summary, resident macrophages represent the major myeloid population in the pancreas with four distinct subpopulations; a homogenous islet-resident F4/80^hi population and a more heterogeneous exocrine-resident F4/80^hi cell fraction, consisting of three distinct subpopulations based on the expression of Tim-4 and MHC II.

We used the Kit^MerCreMer /R26^YFP fate-mapping mouse (9) - where YFP expression can be induced in early c-kit⁺ BM progenitors to investigate the cell replenishment kinetics driven by the BM input in distinct pancreatic myeloid cells (Figure 2A). After treating adult Kit^MerCreMer /R26^YFP mice with tamoxifen (TAM), the YFP labelling index of exocrine tissue-resident macrophages (Tim-4⁺MHCII^low, Tim-4⁺MHCII⁺ and Tim-4^-MHCII⁺), monocytes (Ly6C^hiMHCII^-), monocyte-derived macrophages (Ly6C^-MHCII⁺) and neutrophils (Ly6G⁺) was measured by flow cytometry analysis 1 month and 10 months post-TAM gavage. As observed already in other organs (9), neutrophils showed the maximal level of YFP labelling (Figure 2B), consistent with their known short half-life and therefore fast turnover rates. Hence for standardization purposes, all other obtained YFP values were normalized to the level of neutrophil labelling, which is considered 100%.

In the exocrine pancreas, as expected, due to their fast-turnover rates, Ly6C^hi monocytes and Ly6G⁺ neutrophils showed the highest YFP labelling (Figures 2B, C). Resident Tim-4^-MHCII⁺ and monocyte-derived Ly6C^-MHCII⁺ macrophages were also gradually replenished by BM-derived cells but at a slower rate. After 10 months of chase, more than 70% of Tim-4^-MHCII⁺ macrophages and 80% of Ly6C^-MHCII⁺ macrophages were substituted by monocyte-derived counterparts (Figure 2C). Conversely, the remaining two F4/80^hi resident exocrine macrophages exhibit minimal YFP labelling (around 20-25%) with most cells remaining unlabeled, hence maintaining their embryonic origin. The Tim-4⁺MHCII^low fraction stayed stable at around 20% YFP labelling whereas Tim-4⁺MHCII⁺ macrophages increased by 5% reaching around 30% of YFP⁺ cells (Figure 2C). Correlating with the increased refilling kinetics, the Tim-4^-MHCII⁺ cell fraction within the F4/80^hi macrophages progressively increases with age and represents the largest F4/80^hi fraction of the pancreatic stroma in older mice, whereas Tim-4⁺MHCII^low frequency drastically declines with age (Figure 2D).

In summary, the three exocrine subpopulations dynamically change in proportion during aging.

Two stromal Tim-4⁺ resident macrophage subpopulations are long-lived, and most of their cells retain their fetal origin with aging. On the other hand, BM-derived exocrine Tim-4^-MHCII⁺ resident macrophages are progressively increasing and replacing their embryonic derived counterparts over time.

During aging, islet size and islet macrophage numbers (Figures 3A–C) significantly increase over time (Figures 3A–C). We performed an extensive long-term fate-mapping analysis, to investigate whether BM-derived cells contribute to replenishing the embryonic-derived islet macrophages. The YFP labelling index of endocrine tissue-resident macrophages was measured in Kit^MerCreMer /R26^YFP mice at several time points (1, 4, 6, 8 and 10 months) post-TAM gavage (Figure 4A). Each sample of islets was pooled from two mice, and the normalization was done using the mean neutrophil YFP percentage of the same two mice. In the pancreatic islets, resident F4/80^hi macrophages were gradually replaced over time by BM-derived cells (Figures 4B, C), reaching after 10-month TAM treatment around 70% of normalized YFP labelling (Figure 4D). To validate our flow cytometry results and to visualize the YFP⁺ macrophages directly within the islet, we performed immunofluorescent staining using Iba-1 as a surrogate marker for resident islet macrophages (22). Our analysis confirms the presence of YFP⁺Iba1⁺ cells at the capsular and in the intra-islet region (Figure 4E) which confirms that embryonic-derived islet macrophages are replenished by BM-derived cells over time.

Chemokine/chemokine receptor interactions control the migration of immune cells. Chemokine receptors such as CCR2, CCR5 and CX3CR1 regulate the recruitment of the monocytes into the tissues (23–25). Therefore, we analyzed the expression of these three main chemokine receptors on monocytes, monocyte-derived macrophages, resident macrophages isolated from the exocrine region and macrophages obtained from the islets. In the exocrine pancreas, only Ly6C^hi monocytes, monocyte-dependent F4/80^int macrophages and F4/80^hiTim-4^-MHCII⁺ macrophages expressed CCR2 on the surface of their cells (Figure 5A) whereas Tim-4⁺MHCII^low and Tim-4⁺MHCII⁺ macrophages showed minimal CCR2 expression. CCR5 was expressed similarly on all tested myeloid cells, whereas high expression of CX3CR1 was restricted to F4/80^hiTim-4^-MHCII⁺ macrophages and monocyte-derived F4/80^int macrophages. Interestingly, the expression of CCR2 on islet F4/80^hi macrophages was weak or almost undetectable with the antibody used in this study, whereas CCR5 and CX3CR1 were clearly expressed on their cell surface (Figure 5A). Immunofluorescence staining also confirmed that no CCR2⁺ cells were detected in the islet, whereas clearly, CX3CR1⁺ cells were visible within the endocrine area (Figure 5B).

Since the CCL2/CCR2 axis refills the resident macrophage pool in tissues such as intestine, adipose tissue, heart, and skin, we investigated the contribution of this axis to the replenishment of tissue-resident macrophages in the pancreas. We analysed their frequency and absolute numbers in exocrine and endocrine pancreas of WT and CCR2^-/- mice, the latter mouse strain known for its impaired egress of Ly6C^hi monocytes from the BM and lack of peripheral monocytes (26). Both mouse strains were fed with WD for 24 weeks and were analyzed at an age of 8 months. Correlating with the CCR2 expression, we found that intra acinar Ly6C^hi monocytes, monocyte-derived F4/80^intMHCII⁺ macrophages and F4/80^hiTim-4^-MHCII⁺ macrophage numbers were significantly reduced in CCR2^-/- mice when compared to WT counterparts. Conversely, the numbers of Tim-4⁺MHCII^low and Tim-4⁺MHCII⁺ macrophage subsets were similar in WT and CCR2^-/- mice (Figures 5C, D). Interestingly, islet F4/80^hi macrophages were clearly detectable in both mouse strains and their numbers were comparable (Figures 5E, F). Their presence within the islets was further validated by immunohistochemistry staining on pancreas sections obtained from CCR2^-/- mice (Figure 5G).

In summary, the maintenance of long-lived exocrine TIM-4⁺ macrophages is CCR2 independent. Surprisingly, islet resident F4/80^hi macrophage numbers are not reduced in CCR2^-/- mice, hence their replenishment is independent of the CCL2/CCR2 axis and modulated possibly by other chemokine receptors, such as CCR5 and CX3CR1.

We have recently shown that obesity accelerates the turnover kinetics of adipose tissue-resident macrophages (27). To assess whether obesity affects the replenishment of distinct pancreatic macrophage subpopulations, we established another adult fate-mapping experiment in Kit^MerCreMer /R26^YFP mice maintained either on a normal chow diet (NCD) or on a western diet (WD) for 24 consecutive weeks (Figure 6A). When compared to 3 months old mice, both experimental groups increased their body weights during aging, while the WD-fed mice were overweight and diabetic (Table 1). In fact, mean body weight and fasting blood glucose were significantly higher in WD-fed mice compared with aged-matched NCD-fed controls. The islet area also increased with aging but was equal between the two aged-matched NCD- and WD-ed groups (Table 1). Exocrine and endocrine resident macrophage subpopulations were subsequently analyzed by flow cytometry and YFP labelling was monitored as a parameter for the monocyte-dependent replenishment. As already observed in aging mice (Figure 2D), the exocrine Tim-4^-MHCII⁺ cells represent the largest F4/80^hi fraction of the pancreatic stroma in obese mice, whereas both Tim-4⁺ populations decrease their frequency and numbers (Figures 6B, C, left panels). In WD-fed mice, islet macrophage numbers were increased when compared to young lean mice but were comparable with aged lean (Table 1, Figures 6B, C, right panels). Although it was reported that obesity affects the replenishment kinetics of adipose tissue-resident macrophages (23), in the pancreas, there was no significant perturbation. Both Tim-4⁺ subpopulations remained minimally labeled and maintained their long-lived phenotype also in the tissue microenvironment of obese mice (Figure 6D). Tim-4^-MHCII⁺ and islet macrophages were also comparably YFP labeled between lean and obese aged mice (Figure 6D), which indicates that obesity did not accelerate the monocyte driven replacement and resident macrophages maintained the same turnover kinetics.