SARS-CoV-2 can directly infect NHP and human pancreatic microvasculature and most types of exocrine and endocrine cells and in situ tissue

Several reports have demonstrated that SARS-CoV-2 is localized to pancreatic cells, such as pancreatic ductal epithelium, endothelial cells, acinar cells, mesenchymal cells, and β cells.8,9,10,11 Currently, there is a lack of experimental data on the infiltration and distribution of SARS-CoV-2 in exocrine and endocrine pancreatic cells on the same slice in situ. In this study, we aimed to investigate the relationship between SARS-CoV-2 and different types of islet cells, including alpha cells (α cells), β cells, delta cells (δ cells), and pancreatic polypeptide cells (PP cells), which play key roles in regulating endocrine function and glucose metabolism of pancreatic islets in NHPs and COVID-19 subjects. Human pancreatic tissue samples were used as a comparison to (1) compare the histological similarities and differences between human and NHPs pancreatic tissues and (2) compare the similarities and differences in histopathological states after SARS-CoV-2 infection and further verify the results found on NHPs pancreatic tissues.

Angiotensin-converting enzyme 2 (ACE2) and a plasma membrane-associated type II transmembrane serine protease (TMPRSS2) are two of the most important mediators for SARS-CoV-2 entry.16,17 SARS-CoV-2 entry into cells via the ACE2 receptor requires S protein priming by TMPRSS2.17 Neuropilin 1 (NRP1) is also an important host co-factor for SARS-CoV-2 infection, and recent research demonstrated that the highly expressed neuropilin 1 receptor is critical for viral entry.18,19 The expression of these viral receptors in the pancreas is uncertain, as contradictory results have been reported. Firstly, the protein expression and distribution of the main SARS-CoV-2 entry factors, ACE2, TMPRSS2, and NRP1, in combination with insulin, a β cell marker, glucagon, and α cell maker, were tested and observed by multi-label immunofluorescence in both the autopsy and NHP samples (Supplementary Fig. 1a–c). Different from previous methodologies, by multi-label immunofluorescence, we demonstrated that ACE2, NRP1, and TMPRSS2 proteins were generally expressed within β and α cells in both samples, consistent with their mRNA expression.11 Furthermore, through quantitative analysis (Supplementary Fig. 1d–h), we found significantly increased expression of NRP1 and TMPRSS1 protein after SARS-CoV-2 infection in elder NHPs compared with elder controls (Supplementary Fig. 1e). And a significantly higher co-expression proportion of NRP1 or TMPRSS1 and insulin was observed (Supplementary Fig. 1g, h), while that of ACE2 and insulin was decreased in SARS-CoV-2-infected elder NHPs (Supplementary Fig. 1f).20,21 Taken together, in response to SARS-CoV-2 invasion, β cells may inhibit ACE2 expression as a negative feedback regulation, TMPRSS1 expression is increased, and NRP1 expression, as an alternative receptor, is stimulated.

The SARS-CoV-2 receptors allow for viral infiltration. The specificity of the commercial SARS-CoV-2 antibody has been tested and validated in our recent studies.22,23,24 Firstly, we examined SARS-CoV-2 S protein immunopositivity in all pancreatic sections from autopsy and NHP samples. Further, the presence of SARS-CoV-2 RNA was verified by in situ hybridization. SARS-CoV-2 initially invaded the endothelial cells of the microvasculature in the mildly damaged pancreas and then dispersed among the exocrine and endocrine pancreatic cells. Importantly, S protein immunopositivity samples had a higher frequency and expression in SARS-CoV-2-infected elder NHPs than adult NHPs. Next, we tested the characteristics of SARS-CoV-2 distribution in the islets. Compared with the elder control group (Fig. 1a), glucagon+ α cells (reduced by 63.49%), insulin+ β cells (reduced by 58.66%), and somatostatin+ δ cells (reduced by 82.33%) were significantly decreased, while pancreatic polypeptide+ PP cells were significantly increased in SARS-CoV-2-infected elder NHPs (Fig. 1b–f). Moreover, S+insulin+ β, S+glucagon+ α, S+somatostatin+ δ, and S+polypeptide+ PP cells demonstrated that SARS-CoV-2 directly infected various types of endocrine cells (Fig. 1g–j). Furthermore, the same panel of multi-label immunofluorescence was examined in pancreatic sections of humans. Compared with the control pancreatic tissues (Fig. 2a), co-staining with pancreatic endocrine and non-endocrine markers confirmed the presence of SARS-CoV-2 antigen (S protein) and its relationship with glucagon-secreting α cells, insulin-secreting β cells, and somatostatin-secreting δ cells in the pancreas of autopsy samples (Fig. 2b–e). Quantitative analysis demonstrated that, compared with the control pancreatic tissues, somatostatin+ δ cells (reduced by 72.47%) were significantly decreased in COVID-19 patients’ samples (Fig. 2f). Consistently, S+insulin+ β, S+glucagon+ α and S+somatostatin+ δ cells demonstrated that SARS-CoV-2 directly infected various types of endocrine cells (Fig. 2g–i). Importantly, HE-stained serial sections of the same sample revealed that several pancreatic islets were undergoing degeneration and necrosis characterized by nuclear pyknosis, karyolysis, and loss of original cellular structure, leaving only homogeneously stained eosinophilic areas (Fig. 2b’). Consistent with these pathological changes, various types of islet cells were significantly impaired and reduced in number.

Fig. 1

SARS-CoV-2 directly infects the pancreatic islets of non-human primates (NHPs). a Representative multi-label IF image from the elder control NHP sample was stained for SARS-CoV-2 S1 protein (S, green), glucagon (α, red), insulin (β, cyan), somatostatin (δ, magenta), and polypeptide (P, yellow). Scale bars, 50 μm. b–d Pancreatic tissue section from one SARS-CoV-2-infected elder NHP sample was stained by the same panel of multi-label IF, showing the co-localization of S and P. Scale bars, 800 μm. c, d Representative multi-label IF image from the magnified section of (b). Inset highlights SARS-CoV-2 viral antigen co-localized with islet endocrine cells. Scale bars, 20 μm. e Pancreatic tissue section from another SARS-CoV-2-infected elder NHP sample was stained by the same panel of multi-label IF, showing the co-localization of S and markers of various islet endocrine cells. Scale bars, 800 μm. f Representative multi-label IF image from the magnified section of (e). Inset highlights co-expression of SARS-CoV-2 S protein (S, green) with glucagon (α, red), insulin (β, cyan), somatostatin (δ, magenta), and polypeptide (P, yellow). Scale bars, 100 μm. g–j Quantification of the percentage of glucagon+ α, insulin+ β, somatostatin+ δ, and polypeptide+ PP cells, as well as SARS-CoV-2 S protein+glucagon+ α, S protein+insulin+ β, and S protein+somatostatin+ δ cells, in the elder control NHPs (3 slides) and elder COVID-19 model NHPs(4 slides) (n = 10 images examined from all slides/group). Data are presented as mean ± SD. p Values were calculated by paired or unpaired two-tailed Student’s t test. *p < 0.05, **p < 0.01, and ***p < 0.001

Fig. 2

SARS-CoV-2 directly infects the human pancreatic islet cells. a Pancreatic tissue sections from the deceased control subject were stained by multi-label IF for SARS-CoV-2 S1 protein (S, green), glucagon (α, red), insulin (β, cyan), and somatostatin (δ, yellow) (n = 10 images examined in total). Scale bars, 50 μm. b Pancreatic tissue sections from the COVID-19 autopsy samples (the prototypic SARS-CoV-2 strain-infected) were stained by multi-label IF for SARS-CoV-2 S1 protein (S, green), glucagon (α, red), insulin (β, cyan), and somatostatin (δ, yellow) (n = 10 images examined in total). Scale bars, 200 μm. c Representative multi-label IF image from the magnified section of (b). Inset highlights SARS-CoV-2 viral antigen scattered in a severely damaged and necrotic islet. Scale bars, 50 μm. c The serial section from the same pancreatic tissue stained for H&E and the same necrotic islet was magnified and circled (blue) to clearly observe its pathological changes. The islet structure is disorganized, with cellular swelling and degeneration observed in the internal regions. Cytoplasmic eosinophilia is intensified, and nuclear swelling is evident. The pentagon-marked area (☆) indicates significant necrosis of islet cells, with nuclear pyknosis, karyolysis, and loss of original cellular structure, leaving only homogeneously stained eosinophilic areas. d, e Representative multi-label IF image from the magnified section of (a). Inset highlights SARS-CoV-2 viral antigen co-localized with islet endocrine cells. Scale bars, 50 μm. COVID-19, coronavirus disease 2019; H&E, hematoxylin and eosin; SARS-CoV-2, severe acute respiratory syndrome coronavirus 2. f–i Quantification of the percentage of glucagon+ α, insulin+ β, somatostatin+ δ, and SARS-CoV-2 S protein+ cells, as well as S protein+glucagon+ α, S protein+insulin+ β, and S protein+somatostatin+ δ cells, in control human pancreatic tissues and the COVID-19 patients pancreatic tissues (n = 10 images/group). Data are presented as mean ± SD. p Values were calculated by paired or unpaired two-tailed Student’s t test. *p < 0.05, **p < 0.01, and ***p < 0.001

SARS-CoV-2 infection induces circumscribed pancreatic phenotypic alterations in adult NHPs and aggravates diabetes-like pathological phenotype in elder NHPs

Firstly, we observed the HE-stained sections of all samples to examine the number and extent of major endocrine and exocrine pancreatic lesions between groups. Compared with the adult control monkeys, pathology from adult COVID-19 models was mild, displaying marginally increased degeneration and inflammatory infiltration.

Compared with the elder control monkeys, pathological phenotypes from SARS-CoV-2-infected elder monkeys (the prototypic SARS-CoV-2 strain) were more characterized by aggravated degeneration, amyloidosis, necrosis or atrophy in the islets, and increased hyperplasia, dilatation, adipocytes, and inflammatory cell infiltration in the exocrine pancreas (Table 1). By double immunohistochemical staining of insulin (AP-red) and glucagon (DAB-brown), the lesions in the islets of the elder COVID-19 models were clearly shown (Fig. 3a). Compared with the islets in the adult controls (Fig. 3a-1), several islets in the elder controls underwent degeneration, and slight amyloidosis was detected in the islet cells and nuclei (Fig. 3a-2). The islet amyloidosis appeared as homogeneous to faintly fibrillar, pale eosinophilic deposits in the pancreatic islet interstitium. Meantime, insulin+ β and glucagon+ α cells significantly decreased and were pushed aside (Fig. 3a-3). Islet cell hyperplasia was observed in one case, which was confirmed as glucagon+ α cells increased (Fig. 3a-4). Figure 3a-5 shows islet atrophy resulting in a decreased islet volume. Additionally, shedding of cells in the endocrine ducts (Fig. 3b-1), exocrine duct hyperplasia and dilatation (Fig. 3b-2, b-3), and exocrine pancreas degeneration and inflammation (Fig. 3b-4, b-5) were observed in elder COVID-19 models. As islet amyloidosis can be associated with insulin-resistant (Type 2) diabetes mellitus,25,26 we preliminarily examined these serial sections by special staining. These amyloids were positively stained using Congo red stain, and the average areas of Congo-red-positive deposition expanded by 5.86% (Fig. 3c). Surprisingly, mostly amyloids were also positively stained using Masson stain, and the average areas expanded by 4.16% (Fig. 3d).

Table 1 Summary of pathological changes and SARS-CoV-2 infection percentage in adult and elder COVID-19 NHPs and vaccinated-infected NHPsFig. 3

Characteristics of pathological changes in adult and elder COVID-19 NHPs and vaccination triggers insulin receptor activation in the COVID-19 NHP model. a The main lesions in the pancreatic islets observed in the elder prototypic SARS-CoV-2 strain-infected model (a-3 to a-5) compared with the adult (a-1) and elder control (a-2) animals. The first line shows the representative images of pancreatic tissue sections stained by H&E. The second line shows the representative IHC images of pancreatic tissue sections stained for insulin (red) and glucagon (brown). (a-1) Normal pancreatic islets and surrounding tissues in the adult control group; (a-2) a few islet cells are swelling of individual pancreatic islets in the elder control group. In the elder prototypic SARS-CoV-2 strain-infected model, (a-3) the number of islets cells is significantly decreased, and a large amount of amyloid substance is deposited in some islets; (a-4) glucagon+ alpha cells are increased in individual islets; (a-5) a few islets are atrophy. Scale bars, 50 μm or 100 μm shown in the corresponding images below. b The main lesions in the exocrine pancreas observed in the elder prototypic SARS-CoV-2 strain-infected model (b-1 to b-5). (b-1) Exfoliated cells can be observed in a few pancreatic duct lumen; (b-2) hyperplasia of local pancreatic ductal; (b-3) dilation of pancreatic duct; (b-4) pancreatic periductal inflammatory cells infiltrated around the pancreatic duct; (b-5) inflammatory cells infiltrated in the interstitium. The representative images of pancreatic tissue sections stained by H&E. Scale bars, 50 μm or 100 μm shown in the corresponding images below. c Qualitative and quantitative analysis by Congo red staining to examine prototypic SARS-CoV-2 strain-infection-elicited islet amyloidosis in the elder COVID-19 model (4 slides) compared with the elder control (3 slides) (n = 10 images examined from all slides/group). d The serial section was stained by Masson’s trichrome staining, qualitatively and quantitatively, to observe SARS-CoV-2-infection-elicited islet amyloidosis in the elder COVID-19 model infected with the prototypic SARS-CoV-2 strain compared with the elder control (n = 10 images examined in total/group). e–h Representative multi-label IF image from adult control (a), prototypic SARS-CoV-2-strain-infected COVID-19 NHP model (b), and vaccinated COVID-19 NHP model (c, d) pancreatic tissue section samples were stained for ICA512 (I5, red), insulin receptor α (IRα, yellow), insulin receptor β (IRβ, magenta), CK19 (CK, white), and trypsin (T, cyan). Scale bars, 20 μm in (a–c). c Representative multi-label IF images from section (d). Scale bars, 100 μm in (d). i Quantification of the percentage of ICA512+, insulin receptor α (Ins R α)+, insulin receptor β (Ins R β)+, CK19+, and trypsin+ cells in the control NHPs (3 slides), COVID-19 NHP model (3 slides) and vaccinated COVID-19 NHP model (3 slides) (n = 10 images examined from all slides /group). Data are presented as mean ± SD. p Values were calculated by paired or unpaired two-tailed Student’s t test. *p < 0.05, **p < 0.01, and ***p < 0.001

Previous research has demonstrated that islet amyloidosis is characterized by amyloid fibrils, and disease severity may depend more on the biochemical nature of the amyloid fibrils. In fact, the precursor peptides or intermediate oligomers, rather than the mature amyloid fibrils, are thought to be injurious and cytotoxic agents, at least in islet amyloidosis. The amyloid deposited in the pancreatic islets of human beings and NHPs is derived from islet amyloid peptide and is secreted by β cells. It can be associated with insulin-resistant (type 2) diabetes mellitus. Taken together, SARS-CoV-2 infection induces mild and limited pancreatic phenotypic alterations in adult NHPs and largely aggravates pathological phenotypes perhaps related to diabetes in elder NHPs.

To examine the clinical biochemical indicators associated with pancreatic metabolism, we collected 31 serum samples after overnight fasting from different groups (3 samples from adult control, 3 samples from elder control, 9 samples from the adult model, 9 samples from the elder model, 7 samples from vaccine + adult model) (Supplementary Fig. 2a–f). We observed that NHPs in the adult model exhibited significantly elevated levels of fasting C-peptide and C-peptide/glucose ratio, which are usually used to assess insulin resistance,27 compared with that of adult controls. For the elder model, both the expression of C-peptide and glucose had a tendency to increase compared with the elder control while without significant difference (Supplementary Fig. 2a), which may be due to the small sample size in the elder control animals. In general, there was no significant change in the fasting glucose and fasting insulin levels between the controls, corresponding models, or vaccination groups (Supplementary Fig. 2a, b). Protein phosphatase 1 and regulatory subunit 1A (PPP1R1A), which is positively correlated with insulin impairment, were relatively stable in every group, except in one elder COVID-19 model in which these indexes were significantly elevated (599.9502 pg/mL) (Supplementary Fig. 2c). Amylase and lipase levels were measured to observe whether there was detectable pancreatitis in different experimental groups; there was no significant change compared with the corresponding control group (Supplementary Fig. 2d, e). Antibodies to the 65-kD isoform of glutamic acid decarboxylase (GAD65) are demonstrated to damage the structure and function of pancreatic β cells, which is positively associated with an increased risk of both type 1 and type 2 diabetes mellitus in adults.28,29 Strikingly, the level of GAD65 antibodies in the adult model was extremely significantly elevated compared with that in the adult control; furthermore, there was no significant change in the GAD65 level between the controls, corresponding models, or vaccination groups (Supplementary Fig. 2f). All these clinical data suggest that the β cell function was affected to a certain extent post-SARS-CoV-2 infection in adult models.

Vaccination maintains homeostasis of insulin secretion by activating insulin receptors

We retrospectively collected pancreatic tissue samples from 188 rhesus macaque monkeys from different trials from 2020 to 2023. After evaluating the safety and efficacy of the vaccine, we selected 35 COVID-19 vaccines immunized and prototypic SARS-CoV-2 strain-infected adult NHPs models to compare with the uninfected adult control group and the prototypic SARS-CoV-2 strain-infected adult models group (Table 1) and further observed and studied the pancreatic tissues of animals in these groups.

For clinical data, there was no significant decrease in insulin secretion in the vaccination + COVID-19 model group (Supplementary Fig. 2b). Also, other serum parameters did not show an obvious difference between NHP models and vaccinated models (Supplementary Fig. 2a–f). To eliminate the effects of SARS-CoV-2 infection, the vaccine’s potential effect on pancreatic metabolism was directly observed by collecting and analyzing sera from 4 NHPs for 0–28 days post vaccination (dpv). Compared with fasting C-peptide and glucose levels and their ratio before vaccination, no significant changes in these levels were observed during the 28 dpv (Supplementary Fig. 2g). Importantly, glycated serum protein (GSP), which is a hemoglobin/erythrocyte-independent glycemic marker, reflects the average glucose concentration over the preceding 2–3 weeks.30 The GSP level was fairly stable during the 28 dpv (Supplementary Fig. 2h). The insulin levels showed an upward trend during the 4 weeks, and insulin expression significantly increased at 21 (P = 0.035) and 28 (P = 0.029) dpv, respectively, compared with those at 0 day (Supplementary Fig. 2i). Amylase levels also revealed an upward trend but without significant change compared with those at 0 day (Supplementary Fig. 2j). GAD65 expression increased transiently for 2–3 weeks after vaccination but without significant change compared with those at 0 day and returned to normal levels at 28 dpv (Supplementary Fig. 2k). ICAM-1 expression was slightly downward during the 4 weeks (Supplementary Fig. 2l).

The multi-label IF was examined to further explore in situ. Co-staining with trypsin; islet cell autoantibodies 512 (ICA512), which is an endocrine secretory granule marker; insulin receptor α (IRα); insulin receptor β (IRβ); and CK19, which is a cytoskeleton marker, was performed in the serial sections of pancreatic tissues in control, model, and vaccination + model groups (Fig. 3e–h). Firstly, SARS-CoV-2 infection partly elicits the expression of insulin receptor α, receptor β, and trypsin compared with the control group. Remarkably, ICA512 expression was significantly inhibited while IRα and IRβ expression significantly increased (Fig. 3i) in the vaccination + model group compared with both control and model groups, suggesting that although GAD65 might temporarily increase but back in the swing (Supplementary Fig. 2k), insulin receptors are activated to promote insulin absorption and function to maintain the homeostasis of insulin and glucose levels.

Collectively, these results indicate that vaccination maintains homeostasis of insulin secretion by activating insulin receptors and inhibiting ICA512 expression.

Pathophysiologic mechanisms of SARS-CoV-2 infection causing islet impairment and loss of β cells in elder NHPs

We found mild pathological damage to the pancreas in COVID-19 models and that SARS-CoV-2 infection largely aggravates pathological phenotypes in elder NHPs. Here, we further explored the potential pathophysiological mechanism by which SARS-CoV-2 affects the pancreas, especially on islets. Firstly, markers of islet microcirculation damage, intercellular adhesion molecules (ICAM-1) and vascular cell adhesion molecule-1 (VCAM-1), expressed on inflamed vascular endothelium;31 Ras GTPase-activating protein-binding protein 1 (G3BP1), a marker of cell stress; and cleaved caspase 3, a marker of cell apoptosis, were detected and co-stained with glucagon and insulin protein (Fig. 4a, b). Specifically, compared with the elder control group, ICAM-1 and G3BP1 expression significantly increased in the elder model group, while VCAM-1 and cleaved caspase 3 expression had no significant change (Fig. 4c). Consistent with previous research, the percentage of glucagon+ insulin+ double cells was determined in the elder model group, suggesting SARS-CoV-2-induced β cell transdifferentiation8 (Fig. 4d). Furthermore, the percentage of ICAM+ β cells significantly increased, and that of G3BP1+ β cells significantly increased in the elder model group, but without significant differences in the percentage of VCAM-1+ β cells (Fig. 4e–g). These data suggest that pancreatic microcirculation, especially in the islets, was impaired to some degree and that islet β cells were exposed to an inflammatory environment and stress after SARS-CoV-2 infection in the elder NHPs.

Fig. 4

Characteristics of the markers of microvascular damage and cellular stress in elder control and elder COVID-19 model NHPs. a, b The markers of microvascular damage and cellular stress, ICAM-1 and G3BP1, were elevated in the islets of elder COVID-19 model (a) compared with the elder control NHPs (b). Representative multi-labels IF image in the pancreas from the elder control samples were stained for glucagon (α, magenta), insulin (β, green), ICAM-1 (IC, yellow), G3BP1 (G3, cyan), VCAM-1 (VC, red) and Cleaved caspase 3 (Cas3, white). Scale bars, 20 μm. c–g Quantification of the percentage of ICAM-1+ cell, G3BP1+ cell, VCAM-1+ cell, and cleaved caspase 3+ cell, as well as glucagon+insulin+ cell, ICAM-1+insulin+ cell, G3BP1+insulin+ cell and VCAM-1+insulin+ cell in the elder control NHPs (3 slides) and elder COVID-19 model NHPs (4 slides), (n=10 images examined from all sildes/group). Data are presented as mean ± SD. p values were calculated by unpaired two-tailed Student’s t test. *p < 0.05 and ***p < 0.001

We observed focal lesions and suspected hyperplasia of collagen on HE slices in both the NHP model and autopsy samples. We further tested the expression of type 1 collagen (COL1A1); α-SMA, a marker of stellate cell activation; and CD31, a marker of vascular endothelium, and co-stained these marker proteins with Ki67, viral S protein, and insulin in NHP and human samples (Figs. 5 and 6).32 In the elder control group (Fig. 5a, b), both endocrine and exocrine regions of the pancreas showed minimal yellow-stained COL1A1+ collagen, primarily around blood vessels and connective tissues. These locations are typical for collagen fibers in normal histology. Furthermore, there is a scarce presence of activated α-SMA in the exocrine pancreas. Compared to the elder control group, two distinct pathological phenotypes were observed, depending on the location of collagen accumulation in the pancreas. In NHPs, SARS-CoV-2-infected elder NHPs (Fig. 5c–f) exhibited more clearly pathological phenotypes. In two animals from the elder model (2/6), proliferating collagen fibers distinctly divided the pancreas into different-sized lobules, and in some areas inside these lobules, the collagen fibers divided the pancreatic acini into different-sized islands, leading to loss of acinar cell mass. A few collagen fibers also accumulate in the islets. In the other type (4/6), islet amyloid tissues also robustly and extensively expressed COL1A1+ protein and remarkably replaced the islet cells, including β cells, leading to destruction of endocrine parenchyma and loss of β cells and eventually affecting endocrine metabolism, consistent with the results of Masson staining. Quantitatively, in the elder control group, the expression of COL1A1 was an average of 2.00%, the expression of α-SMA was an average of 4.59%, the expression of Ki67 was an average of 0.56%, and the expression of CD31 was an average 0.45%. Compared with the elder control group, the expression of COL1A1 (increased by 23.98%), α-SMA (increased by 14.97%), and CD31 (increased by 21.14%) significantly increased (Fig. 5h). Importantly, insulin+ COL1A1+ cells increased by 2.24% (Fig. 5i) and insulin+ α-SMA+ cells increased by 1.84% (Fig. 5j), compared with those of the elder control group, further suggesting an imbalanced expression and destruction of β cells. Consistent with our previous data, viral S protein+ CD31+ cells increased (Fig. 5k) and were extensively scattered in the severely impaired islets. Furthermore, the same panel of multi-label immunofluorescence was detected in the human samples, these pathological phenotypes were relatively slight. Compared with the control pancreatic tissues (Fig. 6a), there were also two pathological phenotypes in the COVID-19 patients’ pancreatic tissues (Fig. 6b). In the first type, collagen accumulated in and around the SARS-CoV-2-infected islet cells where stellate cells were mildly activated, and cell proliferation was slightly stimulated (Fig. 6c, d). In the second type, collagen that was extensively diffused and scattered in the interstitium of the exocrine pancreas and multi-focal tissue replaced the acinar cells. α-SMA, Ki67, and CD31 were also expressed (Fig. 6c, e). Quantitative analysis demonstrated that, compared with the control group, the expression of COL1A1 (increased by 19.52%), α-SMA (increased by 20.74%), and Ki67 (increased by 5.78%) significantly increased (Fig. 6f). Importantly, insulin+ COL1A1+ cells increased by 0.53% (Fig. 6g) and insulin+ α-SMA+ cells increased by 0.69% (Fig. 6h), compared with those of the control group. Consistently, the percentage of viral S protein+ CD31+ cells was 0.032 (Fig. 6i).

Fig. 5

SARS-CoV-2 infection aggregated activation of α-SMA and accumulation of collagen fibers from the NHP pancreatic tissues of the elder COVID-19 model. Pancreatic tissue sections from the elder control NHP samples were stained by multi-label immunofluorescence (IF) for COL1A1 (CL, yellow), α-SMA (α, red), Ki67 (K, magenta), SARS-CoV-2 S1 protein (S, cyan), insulin (β, green), and CD31 (CD, white). Scale bars, 200 μm. Representative multi-label IF image from the magnified section of (a). Scale bars, 20 μm. The pancreatic tissue section from one elder NHP model sample was stained by the same multi-label makers in (a). Scale bars, 400 μm. Representative multi-label IF image from the magnified section of (c). Inset highlights proliferating collagen fibers dividing the exocrine and endocrine pancreatic tissues into various islands. Scale bars, 20 μm. e Pancreatic tissue section from another elder prototypic SARS-CoV-2 strain-infected NHP model sample was stained by the same multi-label makers in (a). Scale bars, 800 μm. f, g Representative multi-label IF image from the magnified section of (e). Inset highlights co-localization of SARS-CoV-2 viral antigen and accumulated collagen fibers in damaged islet insulin+ β cells. Scale bars in f 40 μm. Scale bars in g 20 μm. h–k Quantification of the percentage of COL1A1+, α-SMA+, Ki67+, and CD31+ cells, as well as insulin+COL1A1+, insulin+ α-SMA+, and S protein+ CD31+ cells, in the elder control NHPs (3 slides) and elder COVID-19 NHP model (4 slides) (n = 10 images examined from all slides /group). Data are presented as mean ± SD. p Values were calculated by unpaired two-tailed Student’s t test. *p < 0.05, **p < 0.01, and ***p < 0.001

Fig. 6

SARS-CoV-2 infection aggregated activation of α-SMA and accumulation of collagen fibers in the human pancreas. a Pancreatic tissue sections from deceased control subject were stained by multi-label immunofluorescence (IF) for COL1A1 (CL, yellow), α-SMA (α, red), Ki67 (K, magenta), SARS-CoV-2 S1 protein (S, cyan), insulin (β, green), and CD31 (CD, white). Scale bars, 800 μm. b Pancreatic tissue sections from the prototypic SARS-CoV-2 strain-infected COVID-19 autopsy samples were stained by multi-label immunofluorescence (IF) for COL1A1 (CL, yellow), α-SMA (α, red), Ki67 (K, magenta), SARS-CoV-2 S1 protein (S, cyan), insulin (β, green), and CD31 (CD, white). Scale bars, 800 μm. c–e Representative multi-label IF image from the magnified section of (a). Inset highlights activation of α-SMA and proliferation of collagen fibers in both the exocrine and endocrine pancreatic tissues. Scale bars, 50 μm. f–i Quantification of the percentage of COL1A1+, α-SMA+, Ki67+, and CD31+ cells, as well as insulin+COL1A1+, insulin+ α-SMA+, and S protein+ CD31+ cells, in the control human pancreatic tissues and the COVID-19 patients pancreatic tissues (n = 10 images/group). Data are presented as mean ± SD. p values were calculated by unpaired two-tailed Student’s t test. *p < 0.05, **p < 0.01, and ***p < 0.001

SARS-CoV-2 infection of elder NHPs’ pancreas initially damages the pancreatic microcirculation, increases the stress response of islet cells, expands the area of islet amyloidosis, and increases cytotoxicity. Along with the release of inflammatory and stress factors, stellate cells are activated, inducing a sharp increase in the extracellular matrix. Islet degeneration and necrosis stimulate the loss of β cells, directly inhibit insulin expression, and impair the expression of β cell-secreting proteins. Finally, the number of β cells progressively decreased, suggesting that SARS-CoV-2 might aggravate the geriatric development of a diabetes-mellitus-like pathological phenotype in elder NHPs.

Proteomics, lipidomics, and metabolomics display differential metabolic characteristics among SARS-CoV-2-infected adult, elder, and vaccinated NHPs, consistent with their pathological phenotype

Omics-based technologies, including proteomics, lipidomics, and metabolomics, have been adopted during the COVID-19 pandemic to understand the pathophysiological processes and biochemical mechanisms behind infection and spread.33 Here, we examined, analyzed, and compared the sera before and after SARS-CoV-2 infection or vaccination from NHPs with combined multi-omics to explore the potential relationship between pancreatic and metabolic alterations.

A total of 27 serum samples were collected, including the following: 3 serum samples from control NHPs, 5 samples from aged monkeys 3 days post-infection with prototypic SARS-CoV-2 strain, 5 samples from aged monkeys 7 days post-infection with prototypic SARS-CoV-2 strain, 3 samples from adult monkeys 7 days post-infection with prototypic SARS-CoV-2 strain, 4 adult samples 7 days post-infection with delta strain, 3 samples from adult monkeys infected by the prototypic SARS-CoV-2 strain after being vaccinated twice, and 4 samples from adult monkeys infected by the delta strain after being vaccinated twice.

Firstly, through quantitative proteomics, metabolomics, and lipidomics analysis, we sorted out the total differential metabolite numbers among different groups. To be clear, metabolites upregulated or downregulated in serum were marked with different colors within the groups. Overall, compared with the control group, we evaluated different model groups; remarkably, the aged animals infected with SARS-CoV-2 for 7 days had more differentially expressed metabolites (440 in total). Compared with the adult-infected model, the aged-infected model had the highest total of differentially expressed metabolites (i.e., 580) (Supplementary Fig. 3a). This is consistent with the trend of pathological phenotypes. Furthermore, the difference between the vaccinated-prototypic SARS-CoV-2 strain-infected group and the control group was large (562 in total). Venn of proteomics, metabolomics, and lipidomics analysis between the control and different model groups, as well as among the control group, different model groups, and vaccinated-infected groups, is shown in Supplementary Fig. 4b, c. Although the amounts of differential metabolites were found between the control group and other groups, the same differential metabolites among different groups were relatively few, even without the intersection between the vaccination-related groups (Supplementary Fig. 3c). The differential metabolites screened out among the vaccination-related groups in the metabolomics analysis included 2-hydroxy-2-methylbutanenitrile, 4-pyridoxic acid, astaxanthin, dihydroactinidiolide, LPC(20:5/0:0), LPC(O-20:2), propylparaben, and δ-valerolactam. The differential metabolites screened out among the vaccination-related groups in the lipidomics analysis included lysophosphatidylcholine (LPC(20:5/0:0) and LPC(24:1/0:0)); phosphatidylcholine (PC(14:0_20:5), PC(14:0_22:6), PC(16:1_22:6), PC(18:2_20:5), PC(O-14:0_20:4)); and phosphatidylinositol (PI(16:0_20:5)), whose signal transduction pathway is related to the regulation of glucose metabolism.