Development and characterization of lipid nanosystems. In this study, VitE:SM nanosystems were proposed as a strategy to target TAMs in the liver to treat pancreatic cancer metastasis. As mentioned in the introduction, these types of nanocarriers have demonstrated significant potential for delivering therapeutic agents to treat cancer [34], and have proven capable of penetrating dense cancer spheroids, specifically in pancreatic cancer [36, 57]. Likewise, VitE:SM nanoemulsions have shown preferential biodistribution to the liver, and the role of VitE and SM in modulating inflammation has been previously described [46, 47]. The formulated compositions were prepared by adapting an ethanol injection method, enabling a single-step preparation and the straightforward formation of the nanosystems [34, 35, 37] (Fig. 1a). Sphingomyelin (SM) is composed of two long carbon chains (hydrophobic tail), a secondary amine group acting as a linker and a phosphate group (hydrophilic head). Considering its lipophilicity, it is expected that it can form emulsions when combined with oils such as Vitamin E (VitE), a fact that has been validated experimentally and through computational simulation studies [34]. Alternative oils can also form nanoemulsions, in combination with SM, following the same rational and experimental approach. The nanoemulsion compositions could also be complemented by additional phospholipids, including Phosphatidylcholine (PC), Phosphatidylinositol (PI), Phosphatidylserine (PS), Phosphatidylglycerol (PG), or Palmitic acid (PA), since there is growing evidence suggesting that certain phospholipids can influence TAM reprogramming [58, 59]. Consequently, the hydrophobic carbon chain can be incorporated into the oily core of the structure, while the hydrophilic phosphate head is exposed on its surface.

Thus, VitE:SM nanoemulsions and alternative compositions were formulated for comparative purposes, as detailed in Fig. 1b and Table 1. Following formation, all formulations showed appropriate physicochemical properties, with minor differences in size and surface properties, rendering homogeneous populations. In terms of colloidal stability, the formulations were stored at 4°C, and their physicochemical properties, such as particle size, population homogeneity (PDI) and zeta potential were monitored over time. Formulations containing VitE, either with SM or PC (1:0.1 w/w), remained stable for at least 30 days (Figure S1a-b), in agreement with previous results showing high stabilities for VitE:SM nanoemulsions [34].

Subsequently, the lipid nanosystems were incubated with complete RPMI cell culture medium supplemented with 1% FBS at 1:6 (v/v) and measured for up to 24 h at 37°C under orbital shaking. Their size and PDI were determined at different time points by further diluting the sample in ultrapure water at 1:10 (v/v). In line with the results, all formulations showed a good stability with minimal increase in particle size and polydispersity over time (Figure S1c-k).

The lipid nanosystems comprising VitE:SM demonstrate the most favorable M2 macrophage reprogramming capacity. Since our primary objective was to reprogram macrophages with nanotechnology, we first assessed the behavior of VitE:SM nanoemulsions in relation to nanosystems with different lipid compositions (Fig. 1b, Table 1) in immortalized murine bone-marrow-derived macrophages (muIBMDM) [60] to determine their potential toxicity and impact on M2 macrophage reprogramming.

IBMDM macrophages were polarized for 24 h with LPS + IFN-γ for M1 polarization and IL-4 to induce M2 polarization. IL-4 is the most common cytokine used to polarize macrophages to an M2 state [61,62,63]; however, it is important to note that M2 polarization of macrophages is co-regulated by a variety of cytokines and chemokines released by PDAC cells and CAFs in the TME [15]. Following polarization, M2 macrophages were treated with nanoemulsions for 4 h to assess toxicity and reprogramming by qRT-PCR analysis after a 24-h post-treatment period, and by Western blotting to evaluate the expression of the principal M2 marker Arg1 at 48 h post-treatment (Fig. 2a). To assess associated toxicity induced by the different formulations, a bioluminescent assay that measures the release of adenylate kinase from damaged cells was employed, revealing that VitE:SM was one of the least cytotoxic formulations (Fig. 2b), which was confirmed by light microscopy (Figure S2a). VitE:OA:SM was discarded due the high cytotoxicity observed (Figure S2a).

The specific lipid composition of the nanosystems has an influence on M2 macrophage reprogramming. a Experimental design for macrophage polarization and treatment with the lipid nanosystems: murine immortalized bone marrow-derived macrophages (IBMDM) were induced to an M1 phenotype using LPS (10 ng/mL) + IFN-γ (10 ng/mL) or to an M2 phenotype with IL4 (10 ng/mL) for 24 h. Following polarization, lipid nanosystems (1 mg/mL) were incubated with macrophages for a 4-h duration: VitE:SM, M:SM, VitE:PC, VitE:SM:PI, VitE:SM:PS, VitE:SM:PG, and VitE:SM:PA at different concentrations of PA. Subsequently, IBMDM cells were washed and cultured in RPMI medium for qRT-PCR and toxicity assays for 24 h, or Western blot analyses for 48 h. b Fold change in relative luciferase activity (i.e., toxicity) ± SD determined in IBMDM cultures treated with the indicated different nanoemulsion compositions compared to untreated M2 polarized macrophages, set as 1.0 (n = 3). c Analysis of the levels of the principal M2 marker Arg1 by qRT-PCR. Bars represent the mean fold change ± SD (n = 3), with untreated M2 set as 1.0. d Top: Representative Western immunoblots of ARG1 protein expression levels. Bottom: Densitometric analysis of the immunoblots is represented in the bar diagram. Bars represent the mean fold change ± SD (n = 4), with untreated M2 set as 1.0. e Summary table of three parameters chosen for evaluating the best nanoemulsion composition. For toxicity, low < 4 AU, medium = 4–7 AU, high > 7 AU. ∗  = p < 0.05; ∗  ∗  = p < 0.01; ∗  ∗  ∗  = p < 0.001; ∗  ∗  ∗  ∗  = p < 0.0001; ns = not significant. One-way ANOVA test for multiple comparisons with Dunnett’s post hoc test, compared to the untreated M2 sample

Regarding reprogramming towards a more M0/M1 phenotype, qRT-PCR and Western Blot analysis of Arginase 1 (Arg1/ARG1) expression, a well-known marker for M2 macrophages in PDAC [64] was performed (Fig. 2a). By qRT-PCR analysis we observed a general decrease in Arg1 expression with all the tested formulations, with a more pronounced decrease observed in the case of the VitE:SM formulation compared to untreated M2 macrophages (Fig. 2c). This trend was similarly observed at the protein level. While a decrease in ARG1 expression was achieved with nearly all the formulations, it was only statistically significant in the case of nanoemulsions containing VitE:SM and VitE:SM:PI (Fig. 2d).

Based on these data, we determined that the most effective formulation for macrophage reprogramming and the most suitable low toxic composition for further development was the nanosystem containing VitE:SM (Fig. 2e). To investigate the potential synergies offered by VitE and SM in TAM reprogramming, we conducted an additional experiment in which IBMDM macrophages were first polarized with IL-4 for 24 h to induce M2 polarization. The M2 macrophages were then treated for 4 h with the VitE:SM nanoemulsions, SM, VitE, or a free combination of VitE and SM to assess toxicity (% live cells) and M2 reprogramming via flow cytometry (CD45 + <  CD11b +  < F4/80 +  < CD206 +) 24 h post-treatment (Figure S2b). Our results confirm that the VitE:SM nanoparticle composition exhibits a synergistic effect to reprogram M2 TAMs and demonstrates improved cell viability compared to the other compositions. To further analyze the structure of this composition, we employed field emission scanning electron microscopy (Figure S3).

We then investigated whether primary murine macrophages derived from primary bone marrow monocytes (BMDM) treated with VitE:SM maintained their reprogramming over time and whether these effects were reversible upon subsequent exposure to the M2-polarizing cytokine IL-4 (Figure S4a). Indeed, Western blot analysis of ARG1 revealed a continued reduction of this M2 marker over time in BMDMs treated with the VitE:SM nanosystem. Additionally, VitE:SM nanoemulsion-treated macrophages that were repolarized with IL-4 did not increase ARG1 to levels observed in control groups (Figure S4b). Considering the implications for in vivo use, these findings suggest that despite the continuous influence from tumor secreted pro-TAM factors, treated macrophages are likely to retain the polarization state induced by the VitE:SM nanoemulsions over time; however, as stated above, M2 polarization of macrophages in vivo is co-regulated by a variety of cytokines and chemokines released by PDAC cells and CAFs in the TME [15].

We hypothesized that the potent reprogramming capacity of the VitE:SM formulation arises from two key properties. On the one hand, it is well-established that the phagocytosis of lipid membranes by macrophages affects their programming and metabolism [32, 65, 66]. Some studies utilize cell membrane-derived nanoparticles (nanoghosts) for macrophage reprogramming, attributing this effect to components like cytokines and chemokines present in the cell membranes of nanoghosts. However, it is also plausible that membrane sphingolipids such as SM play a crucial role in macrophage reprogramming [44,45,46,47, 66]. On the other hand, the immunostimulant properties of VitE may also contribute to macrophage reprogramming [40,41,42]. Thus, the dual actions of VitE and SM suggests a potential synergy in enhancing the reprogramming capacity of this composition, making this combination particularly promising for further experimentation.

RNAseq analysis of M2 and M2 VitE:SM nanoemulsion-treated macrophages. To further investigate the effects of the VitE:SM formulation on M2 macrophage polarization at the transcriptomic level, we performed RNAseq on IL-4-stimulated (M2) and IL-4-stimulated VitE:SM nanoemulsion-treated muBMDMs. Among the genes significantly upregulated in VitE:SM-treated macrophages, we found genes related to inflammation (i.e., Tnf, Ptgs2, Il-11, Il-1a, Cxcl1) and with antigen presentation (i.e., H2-M2). Additionally, among the downregulated genes, we identified Cd51 and Cd300e, which are related to M2 polarization, and Angpt1 which is a pro-angiogenic factor (Fig. 3a).

RNAseq analysis of M2 and M2 VitE:SM nanoemulsion-treated macrophages. a Volcano plot showing differentially expressed genes in VitE:SM nanoemulsions-treated (1 mg/mL) M2 macrophages versus naïve M2 macrophages. Red dots, genes upregulated; blue dots, genes downregulated. Genes of interested have been labelled in black. b Gene set enrichment analysis (GSEA) plots showing a significant enrichment in the indicated gene signatures for VitE:SM nanoemulsions-treated M2 macrophages. Normalized enrichment score (NES) and p- and q- values are indicated for each plot. c Histogram representation of the NES of significantly altered gene signatures down- (blue) and upregulated (red) in VitE:SM-treated M2 macrophages

We next performed a Gene Set Enrichment Analysis (GSEA) employing Molecular Signature Database (MsigDB) signatures (including Gene Ontology, Reactome and Hallmarks) or custom gene signatures. Interestingly, VitE:SM nanoemulsion-treated macrophages recover a pattern related to resting differentiated macrophages or monocytes, rather than a gene set associated with M2 or IL-4 stimulated macrophages. We also found mainly an enrichment in gene signatures related to TNF-TNFR1 signaling, but also macrophage activation via Toll receptors, interferon-related signaling, reactive oxygen species and signatures related to innate immune system activation, indicating that VitE:SM treatment induces a reprogramming of M2 macrophages towards a more inflammatory and less immunosuppressive phenotype, which is in line with our previous results and hypothesis (Fig. 3b-c).

The intraperitoneal injection of VitE:SM nanoemulsions reduces tumor burden in an orthotopic KPC tumor model. To initially validate the in vivo efficacy of the nanosystems, we conducted a preliminary study employing murine PDAC cells obtained from spontaneous tumors from the KPC (LSL-KrasG12D/+; LSL-Trp53R172H/+; Pdx-1-Cre) mouse model [67]. These cells were orthotopically injected into the pancreas of syngeneic and immunocompetent C57BL/6 wild-type (WT) mice (Fig. 4a), with the nanosystem treatment commencing on the seventh day post-implantation in two distinct non-parallel experiments. In each experiment, randomized groups of 4 or 5 mice were intraperitoneally injected with a vehicle solution (0.9% NaCl) for the control mice or 69 mg/kg of VitE:SM nanoemulsions for treated mice. This regimen was applied for five consecutive days per week. Mice were euthanized on day 27 post-implantation, after confirming tumor growth in a sentinel mouse by Magnetic Resonance Imaging (MRI) (Figure S5a). Total body weight was recorded at the start and end of the treatment to assess nanoemulsion toxicity, revealing no differences between groups throughout the study period (Fig. 4b). In this model, tumor presence typically impairs weight gain and often leads to reduced food consumption or tumor-associated cachexia, so the lack of weight gain is expected. Full-body necropsy was then performed and no liver toxicity was observed at the organotypic level (Fig. 4c), nor were differences observed in liver weight (Fig. 4d). At the histopathological level, examination of H&E-stained liver slides showed that the structure and architecture of the liver remained unaffected by the treatment. There was no evidence of inflammation, necrosis, or loss of architecture in the central vein or sinusoids. (Fig. 4e). Finally, the assessment of liver toxicity following treatment with VitE:SM nanoemulsions, as determined by liver function tests, did not reveal elevated values of ALT, AST and GGT when compared to standard reference ranges (Table S3). Liver toxicity was also absent in non-implanted mice treated with three doses of VitE:SM nanoparticles administered every two days over a one-week period. Serum analysis after multiples treatments showed normal ALT enzyme levels, with no significant difference between control and VitE:SM-treated mice (Figure S5b). Additionally, the mice did not experience weight loss during the treatment process (Figure S5c). This underscores the biocompatibility of the chosen nanosystem, as evidenced by the absence of toxicity associated with the nanoemulsion. Moreover, unlike other treatments that deplete TAMs, such as clodronate or toxin-conjugated monoclonal antibodies [31, 68], VitE:SM nanosystems could offer a safer therapeutic option.

Intraperitoneal injection of VitE:SM nanoemulsions reduces tumor burden in an orthotopic KPC tumor model. a Experimental set-up for the in vivo syngeneic model in C57BL/6 mice. Mice were orthotopically implanted with 2,500 KPC cells/pancreas and treatment was initiated 7 days post-implantation, receiving VitE:SM nanoemulsions treatments (69 mg/kg) 5 days per week for three weeks via intraperitoneal (IP) injection. b Fold change ± SD of the total weight of the animals at the start and completion of the treatment. The initial weights for the control (n = 10) and VitE:SM (n = 9) groups from two independent experiments were pooled and each set as 1.0 to observe the weight evolution throughout the experiment. Days post-implantion (DPI). c Representative images of the liver at the experimental endpoint from control (n = 10) and VitE:SM-treated (n = 9) groups from two independent experiments. Scale = 1cm. d Liver weight from c. Data were normalized for each independent experiment (n = 2) with the bars representing the mean fold change ± SD, with the control set as 1.0. Unpaired t test. ns = not significant. e Representative images (40X) of H&E-stained liver slices at the experimental euthanize time point from control or VitE:SM-treated mice. Scale = 400µm. Zoom areas are depicted within squares. Scale = 100µm. CV Central vein, S Sinusoids. f Representative images of the tumor (T) and adjacent peritoneum tumor-derived masses (PM) at the experimental endpoint for control (n = 10) and VitE:SM-treated (n = 9) groups from two independent experiments. An additional control group treated with non-effective nanoparticles M:SM (n = 4) was included. All tumor images are presented in Figure S6. Scale = 1cm. S = Spleen served as an anatomical reference for the pancreas. g Fold change in T + adjacent PM weight ± SD from f. Data were normalized for each duplicate experiment with the control set as 1.0. One-way ANOVA with Dunnett’s post hoc test, compared to control. ∗  ∗  = p < 0.01. ns = not significant

Both at the level of the tumor alone (Figure S5d) or including the adjacent peritoneal tumor-derived masses (Fig. 4f-g), a reduction in total tumor weight was evident in mice treated with the VitE:SM nanosystem compared to the control group (Fig. 4f-g and S6). Importantly, and to specifically highlight the activity of the VitE:SM formulation, an additional nanoemulsion control group was included. Specifically, mice were also treated with Mygliol (M):SM nanoemulsion, which were shown to be less effective at the level of M2 TAM reprogramming (Fig. 2). Tumor weights confirmed a significant decrease in tumor burden with intraperitoneal administration of VitE:SM nanoemulsions but not with the M:SM nanoparticles (Figure S5d and 4g), indicating that the VitE:SM formulation has specific anti-tumoral properties.

Taken together, these findings align with other studies that have also utilized nanoparticles to modulate the immune system; however, the composition of the nanoemulsion appears to be critical to achieve an anti-tumor effect. Thus, the integration of nanotechnology to enhance immune responses and/or affect tumor growth represents a growing and consistent trend and highlights the efficacy of nanoparticle-based approaches in immuno- and tumor modulation [69, 70].

Targeting capacity of the TopFluor® (TF)-labelled VitE:SM nanoemulsions in vitro and in vivo. To further evaluate the in vivo application of the nanosystems, nanoemulsions were formulated using SM labeled with the green fluorophore TopFluor® (TF) for tracking purposes in mice (Fig. 5a and Table 2). First, confocal microscopy images of IBMDM macrophages revealed green fluorescence in those cultures treated with VitE:SM:TF nanoemulsions (Fig. 5b). Detection of the fluorophore by flow cytometry, using Ex488nm Em530/30nm, showed nearly complete staining of macrophages (Fig. 5c). Since the goal of this study was to target liver TAMs, we assessed which route of administration would favor liver biodistribution, as measured by flow cytometric analysis of tumors, liver and lungs. VitE:SM:TF nanoemulsions were administered intraperitoneally or retro-orbitally to KPC orthotopically implanted C57BL/6 mice seven days post-surgery, at a dose of 25 mg/kg of VitE:SM:TF nanoemulsions. Mice were euthanized 48 h later to assess TF biodistribution in the tumor and the two principal organs susceptible to develop metastasis: liver and lungs. Following organ digestion, we confirmed by flow cytometry that while intraperitoneal administration is optimal for reaching the tumor, retro-orbital injection is more effective in reaching the liver and the lungs (Fig. 5d). To further compare retro-orbital injection with another intravenous route, such as tail vein injection, we conducted an additional biodistribution experiment, which confirmed that retro-orbital injection was more efficient in delivering the treatment to the liver, the primary target organ of this study (Fig. 5e). To determine the kinetics of VitE:SM:TF nanoemulsion liver targeting, and specifically liver macrophage (i.e., CD45 + , CD11b + , and F4/80 +) targeting (Fig. 5f), VitE:SM:TF fluorescence was assessed 1 h, 4 h, 24 h, and 72 h post retro-orbital injection. Flow cytometry analysis of digested livers indicated that peak accumulation of VitE:SM:TF in liver macrophages was reached at 24 h, with a decrease at 72 h (Fig. 5g). Thus, we established that the optimal treatment regimen for subsequent in vivo intervention studies to be retro-orbital injection every 48 h.

Targeting capacity of the TopFluor® (TF) labelled VitE:SM nanoemulsions in vitro and in vivo. a TopFluor® was incorporated in the surfactant (TopFluor®-labelled-sphingomyelin) and injected in the aqueous phase as previously explained (Fig. 1a). VitE:SM:TF = Vitamin E + TopFluor®-labelled-sphingomyelin nanosystem. b Confocal images of muBMDM cells treated or not with TF-labelled nanoemulsions (VitE:SM:TF, 0.5 mg/mL). Scale bar = 20 µM. DAPI: blue, TF: green. c Cytometry dot plots of muIBMDM cells in the presence or absence of VitE:SM:TF (0.5 mg/mL) nanoemulsions. Murine IBMDMs containing VitE:SM:TF nanoemulsions were detected at a wavelength of Ex488 nm Em530/30 nm. SSC-A: Side Scatter (Area). d Flow cytometry analysis of the percentage of TopFluor®-positive live cells ± SD in three different organs (liver, tumor, and lung) under two treatment modalities: intraperitoneal or retro-orbital (n = 2 mice per treatment). e Flow cytometry analysis of the percentage of TopFluor®-positive live cells ± SD in the liver under two intravenous treatment modalities (tail vein injection or retro-orbital) compared to control (n = 5 control, n = 5 mice per each treatment). ∗  = p < 0.05; ∗  ∗  ∗  = p < 0.001. One-way ANOVA with Dunnett’s post hoc test, compared to the retro-orbital condition. f Schematic representation of the markers used to identify macrophages: CD45 + , CD11b + and F4/80 + . g Flow cytometry analysis of the percentage of macrophages (CD45 + , CD11b + and F480 +) ± SD targeted by the VitE:SM:TF nanoemulsions over time, with data collected at 1 h, 4 h, 24 h and 72 h. Unpaired t test comparing 24 and 72 h. ns  not significant

Loading of VitE:SM nanoemulsions with the TGF-βR1 inhibitor (LY2157299) reduces tumor growth and diminishes TAM liver infiltration. Next, we decided to further explore the potential of VitE:SM nanoemulsions to encapsulate and deliver the TGF-βR1 inhibitor Galunisertib LY2157299, referred to as LY (VitE:SM:LY) (Fig. 6a and Table 2). We hypothesized that LY would inhibit the activity of fibrotic cytokines in vivo and prevent the M2 polarization of macrophages via TGF-β [71,72,73], as previously discussed, thereby further obstructing the formation of the pre-metastatic niche in the liver [48,49,50]. Moreover, the encapsulation of LY into the VitE:SM nanosystem should in theory enhance its delivery and mitigate non-specific toxicity often associated with the administration of free LY [74, 75]. We first tested the VitE:SM and VitE:SM:LY compositions at different concentrations (1 and 0.5 mg/mL) in muBMDM cells polarized to an M2 state with IL-4. Western blot analysis of ARG1 protein levels revealed a decrease in this M2 marker across all treated samples, with no significant differences observed between different concentrations or with LY encapsulation (Fig. 6b). Additionally, qRT-PCR analysis of various M1 or M2 markers consistently demonstrated a decrease in classical M2 markers such as Arg1 and Egr2 (Fig. 6c) and an increase in some M1 markers such as Cd86, Il1b, and Il12b (Fig. 6d). Of note, the addition of LY to the nanoemulsions did not lead to an improvement in macrophage reprogramming, which is in slight contrast to what we have previously seen with another TGF-βR1 inhibitor in human M2-polarized macrophages [56]. We speculated that the potent effect of the VitE:SM vehicle was masking the influence of LY in vitro, but we also reasoned that the additive effect offered by encapsulating LY in the nanosystems would be more evident in vivo.

Testing VitE:SM nanoemulsions loaded with the TGF-βR1 inhibitor (LY2157299). a For LY loading, the TGF-βR1 inhibitor was dissolved in the organic phase and injected in the aqueous phase as previously explained (Fig. 1a). VitE:SM = Vitamin E + sphingomyelin nanosystem; VitE:SM:LY = Vitamin E + sphingomyelin + encapsulated TGF-βR1 inhibitor. b Left: Western immunoblot analysis of ARG1 protein levels in muBMDMs polarized as in Fig. 2a to M0, M1 or M2 and treated with empty or LY encapsulated VitE:SM nanoemulsions at two different concentrations: 1 and 0.5 mg/mL. Right: Densitometric analysis of the immunoblots, with the bars representing the mean fold change ± SD (n = 3), with untreated M2 set as 1.0. c Analysis by qRT-PCR of two M2 markers, Arg1 and Egr2, in the same samples as described in b. Bars represent the mean fold change ± SD (n = 3), with untreated M2 set as 1.0. d Analysis by qRT-PCR of M1 markers, Cd86, Il12b and Il1b, in the same samples as in b and c. Bars represent the mean fold-change ± SD (n = 3), with untreated M1 set as 1.0. ∗  = p < 0.05; ∗  ∗  = p < 0.01; ∗  ∗  ∗  = p < 0.001; ∗  ∗  ∗  ∗  = p < 0.0001; ns  not significant. One-way ANOVA with Dunnett’s post hoc test, compared to M2 in b and c and to M1 in d

Towards this end, C57BL/6 mice were orthotopically implanted with KPC cells, and VitE:SM:TF or VitE:SM:LY:TF nanoemulsion treatment was initiated on day 4 post-implantation at a dose of 25 mg/kg, administered retro-orbitally every 48 h (3 days per week). Three euthanize time points to assess tumor progression were established: days 21, 28, and 35 post-implantation (Fig. 7a). At the indicated time points, tumors were extracted, and a reduction in tumor size in mice treated with empty nanoemulsions (VitE:SM:TF) and with nanoemulsions loaded with LY (VitE:SM:LY:TF) was observed (Fig. 7b). Monitoring tumor weight over time revealed that VitE:SM:TF and, more significantly, VitE:SM:LY:TF exerted a cytostatic effect compared to control mice (Fig. 7c). Likewise, the sum of the data across all time points collectively exhibited a significant reduction in tumor weight in mice treated with both nanoemulsion formulations (Fig. 7c). Histological analysis of tumor samples revealed a higher proportion of tumoral area in mice from the control group, whereas areas of healthy pancreas were still prominent in many mouse pancreata from the treatment groups (Fig. 7d-e).