Islet Interleukin-1Β Immunoreactivity Is an Early Feature of Cystic Fibrosis That May Contribute to Β-Cell Failure

Diabetes Care 1

Islet Interleukin-1b Rebecca L. Hull,1 Ronald L. Gibson,2 Sharon McNamara,2 Gail H. Deutsch,3 Immunoreactivity Is an Early Corinne L. Fligner,3 Charles W. Frevert,1,4 Bonnie W. Ramsey,2 and Srinath Sanda5,6 Feature of Cystic Fibrosis That May Contribute to b-Cell Failure https://doi.org/10.2337/dc17-1387

OBJECTIVE Cystic fibrosis–related diabetes (CFRD) is a common complication of cystic fibrosis (CF), increasing patient morbidity and mortality. Poor understanding of CFRD pathogenesis limits the development of targeted therapies to treat and/or prevent the disease. The aim of this study was to evaluate islet pathology, specifically, inflammation, amyloid deposition, and endocrine cell composition in subjects with CF with diabetes and with CF without diabetes.

RESEARCH DESIGN AND METHODS A retrospective analysis of archived pancreas tissue collected at autopsy was conducted using pancreas tissue from subjects with CF and diabetes (CFRD) (n = 18) and PATHOPHYSIOLOGY/COMPLICATIONS CF without diabetes (CF-no DM) (n = 17). Two cohorts of control non-CF subjects were identified, each matched to CFRD and CF-no DM subjects for age, sex, and BMI (non-CF older, n = 20, and non-CF younger, n = 20), respectively. Immunohistochem- istry was performed to assess IL-1b and islet hormone (insulin, glucagon, somatostatin, and pancreatic polypeptide) immunoreactivity; histochemistry was performed 1Department of Medicine, University of Wash- to quantify amyloid deposition. ington, Seattle, WA 2Department of Pediatrics, University of Wash- RESULTS ington, Seattle, WA 3Department of Pathology, University of Wash- Islet IL-1b immunoreactivity was substantially increased in both CFRD and CF-no DM ington, Seattle, WA subjects compared with non-CF subjects and was common in young subjects with CF 4Department of Comparative Medicine, Univer- (£10 years of age). In contrast, islet amyloid deposition was increased only in CFRD sity of Washington, Seattle, WA 5Department of Pediatrics, University of Califor- subjects. We also observe abnormal islet hormone immunoreactivity, characterized nia, San Francisco, San Francisco, CA by increased glucagon immunoreactivity, in CF-no DM and CFRD subjects compared 6Diabetes Center, University of California, San with non-CF subjects. Francisco, San Francisco, CA Corresponding author: Rebecca L. Hull, rhull@ CONCLUSIONS uw.edu, or Srinath Sanda, srinath.sanda@ucsf These findings reveal novel molecular pathways and therapeutic targets underlying .edu. islet pathology in CF subjects and may be important in developing new approaches to Received 11 July 2017 and accepted 6 January treat CFRD. 2018. This article contains Supplementary Data online at http://care.diabetesjournals.org/lookup/ Cystic fibrosis–related diabetes (CFRD) occurs in up to 50% of adults with cystic fibrosis suppl/doi:10.2337/dc17-1387/-/DC1. (CF), with the highest incidence occurring during the second to third decade of life (1), © 2018 by the American Diabetes Association. although abnormal glucose tolerance is common even in very young children with the Readers may use this article as long as the work disease (2–4). CFRD is associated with increased morbidity and mortality in both males is properly cited, the use is educational and not for profit, and the work is not altered. More infor- and females (5). Insulin therapy is the current standard of care for CFRD. Unlike type 1 mation is available at http://www.diabetesjournals diabetes, CFRD does not appear to be the result of islet autoimmunity; HLA .org/content/license. Diabetes Care Publish Ahead of Print, published online February 1, 2018 2 Islet IL-1b Immunoreactivity in Cystic Fibrosis Diabetes Care

susceptibility alleles are not observed that CFRD would be characterized not 3 from the JDRF nPOD registry) were iden- with greater frequency, although studies only by islet amyloid but also by islet ex- tified. These, designated “non-CF (youn- differ with respect to the prevalence of pression of the inflammatory cytokine ger)” (n = 20) and “non-CF (older)” (n = islet autoantibodies (6,7). Physiologically, IL-1b. We conducted a histologic study 21), were matched for age, sex, and BMI CFRD appears to have more in common to test this hypothesis. We observed to CF-no DM and CFRD subjects, respec- with type 2 diabetes. Genetic susceptibil- greater IL-1b immunoreactivity in islets tively, and to the extent possible, given ity loci for type 2 diabetes appear to con- subjects with CF with or without diabetes that autopsies were performed over fer risk for CFRD (8–10), and, similar to compared with control subjects without many years, these were also matched to type 2 diabetes, CFRD is characterized CF, while islet amyloid deposition was the time of autopsy. One non-CF control by a mixture of peripheral, particularly only increased in those individuals diag- subject [from the non-CF cohort (older)] hepatic, insulin resistance and significant nosed with CFRD. Islet endocrine compo- was excluded from further analysis, as no defects in b-cell function (4,11–13). How- sition was also disturbed, with increased pancreas specimen was available, bringing ever, the cause of b-cell dysfunction in a-cell area observed in subjects with CF the sample size from the non-CF (older) CFRD is not known. with and without diabetes. Based on group to 20. The study was approved by Work in various animal models, includ- these observations, we propose that CF institutional review boards at the Univer- ing the recently developed pig and ferret is characterized by islet inflammation, sity of Washington and SCH. models, demonstrates abnormalities in which could predispose to b-cell failure, Pancreatic tissue was obtained during glucose metabolism and insulin release, and by islet a-cell expansion, while islet autopsies performed at the University of mostly in the newborn period (14–16). amyloid formation is mainly restricted to Washington and SCH or via the nPOD pro- Human autopsy studies have been infor- CFRD. gram. Histologic sampling from the body mative in describing islet cell histological of the pancreas was routinely performed, changes with clinical manifestations. although autopsy records did not always Some small studies have reported that RESEARCH DESIGN AND METHODS indicate the specific location of the sample. subjects with CFRD have a lower density Subjects Specimens were included in the study only of endocrine (islet) tissue compared with Forty-one patients with CF were identi- if they showed no or minimal autolysis, as fi patients with CF without diabetes (17) ed by retrospective screening of autopsy assessed by CLF and RLH (University of ’ and/or loss of b-cells with no change in records at the Seattle Children s Hospital Washington samples) or GHD and RLH glucagon-producing a-cells or somato- (SCH) and the University of Washington (SCH samples). Pancreatic weight was not statin-producing d-cells compared with Medical Center for a CF diagnosis. We available; therefore, endocrine cell data are CF subjects without diabetes (18,19). screened records from SCH between presented as relative area rather than mass. However, one larger study found no dif- 1977 and 2012 and at University of Wash- ference in b-cell area among subjects ington Medical Center between 1991 and with CF with or without diabetes (20). 2013. Available clinical data (Table 1), in- Immunohistochemistry Four-micron-thick sections of formalin- As in type 2 diabetes, pancreatic islet cluding history of lung transplantation, fixed, paraffin-embedded pancreas were amyloid deposition has been identified were extracted from the medical record subjected to immunohistochemistry (IHC) as a characteristic feature of CFRD in hu- and managed using REDCap electronic mans (19,20). Islet amyloid is a well- data capture tools hosted at the Univer- (LeicaBond Max; Leica Microsystems, Buf- described pathological feature of type 2 sity of Washington’s Institute of Transla- falo Grove, IL) as follows. The following diabetes, and consistent with the etiology tional Health Sciences. Three additional primary antisera were used, for each of of this form of diabetes, islet amyloid de- cases of CF were available from the which the host species is denoted in position is usually observed in older indi- JDRF-supported Network for Pancreatic square brackets: insulin (A0564 [guinea viduals (21–23). Islet amyloid deposition Organ Donors with Diabetes (nPOD). pig], 1:4,000; Dako, Carpenteria, CA), b is toxic to b-cells, correlating with both Nine were excluded from further analysis IL-1 (49-4960 [rabbit], 1:500; ProSci In- reduced b-cell area and increased b-cell based on the following criteria: incom- corporated, Poway, CA; and ab2105 [rab- apoptosis in patients with type 2 diabetes plete medical records (n = 1), lack of avail- bit], 1:1,000; Abcam, Cambridge, MA), (21–23). Recent data have linked islet am- ability or poor quality of pancreas tissue IL-1Ra (AF-280-NA [goat], 1:500; R&D Sys- yloid deposition to islet inflammation, (n =5),absenceofpancreaticisletsonthe tems, Minneapolis, MN; and NBP1-32568 namely, showing that islet amyloid indu- available pancreas sections (n = 2), or his- [rabbit], 2 mg/mL; Novus Biologicals, Little- ces inflammatory cytokine production, tory of islet antigen 2 antibodies and zinc ton, CO), CD68 (Clone 514H12 [mouse], 0.34 chiefly interleukin (IL)-1b,bymacrophages transporter 8 autoantibodies, suggestive mg/mL; Leica Biosystems/Novacastra), and dendritic cells (24,25). In a similar of the presence of type 1 diabetes (n =1). glucagon (EP3070 [rabbit], 1:10,000; fashion, the systemic inflammatory pro- Of the remaining 35 subjects with CF, Epitomics, Burlingame, CA), somatostatin cess observed in CF subjects also seems to 18 were classified as “CFRD” based on (SC-7819 [goat], 1:200; Santa Cruz Bio- be skewed toward IL-1b–regulated cyto- presence of diabetes diagnosis by the technology, Santa Cruz, CA) and pancre- kines such as IL-8 and IL-17 (26–28). treating physician and/or insulin use in atic polypeptide (PP) (NB100-1793 [goat], Given these previous observations of the medical record; the other 17 were des- 1:250; Novus). Isotype-matched irrele- islet amyloidosis in CFRD (19,20), the in- ignated “CF-no DM.” Given the difference vant IgG-negative controls were used for flammatory properties of islet amyloid in age and BMI between the CFRD and all antisera (except those against islet hor- (24,25), and the systemic inflammatory CF-no DM cohorts, two groups of non-CF mones, which have been previously vali- milieu of CF (26–28), we hypothesized control subjects without diabetes (including dated in pancreas). care.diabetesjournals.org Hull and Associates 3

Table 1—Subject characteristics, transplant history, diabetes medications, and causes of death Non-CF (younger) Non-CF (older) CF-no DM CFRD N 20 20 17 18 Age (years) 13.1 6 6.8 28.0 6 5.5*# 13.9 6 8.3 28.6 6 9.0*# Sex (female/male) 12/8 9/11 12/5 10/8 BMI (kg/m2)19.86 7.0 22.6 6 5.5# 16.5 6 3.9 21.7 6 3.3# Diabetes duration (years) (n =17) 4.80 6 5.16

HbA1c (%) (n =11) 7.25 6 1.67 Lung transplant (yes/no/unknown) 0/19/0 0/20/0 4/13/0 10/6/2 Other transplant 1 (heart) 0 0 0 Diabetes medications Oral agent 1 Insulin 16 Unknown 2 Steroid use (yes/no/unknown) 4/13/3 0/15/5 4/13/0 11/5/2 Pancreatic enzyme supplementation (yes/no/unknown) 16/0/1 16/0/2 Cause of death Respiratory 5 5 14 14 Cardiovascular 4 4 2 d CNS 3 6 dd Infection 4 3 dd Malignancy 2 ddd Multiorgan failure 1 2 d 3 Unknown 1 dd1 Data are means 6 SD or n. CNS, central nervous system. *P , 0.05 vs. non-CF (younger). #P , 0.05 vs. CF-no DM.

Sections were deparaffinized followed and Megan Larmore, a staff member Data and Statistical Analysis by antigen retrieval (citrate buffer at from the University of Washington Cystic Comparisons between groups were done 100°C for 20 min for IL-1b [both antisera]; Fibrosis Research and Translation Center’s using one-way ANOVA for overall statisti- EDTA at 100°C for 10–20 min for IL-1Ra Host Response Core). For insulin, gluca- cal differences; pairwise comparisons [both antisera], CD68, and PP; and pro- gon, somatostatin, and PP IHC, total pan- were conducted with a Tukey test to cor- teinase K for 15 min at 37°C for somato- creas tissue and hormone-positive areas rect P values for multiple comparisons. A statin). No antigen retrieval was used for were determined automatically based on Fisher exact test was used to compare insulin or glucagon IHC. Sections then un- pixel value and density (Visiopharm soft- frequency of lung transplantation be- derwent peroxide block followed by incu- ware, Hoersholm, Denmark) and verified tween groups, and a x2 test was used to bation in 10% (v/v) normal goat serum for by manual examination of segmented im- compare frequency of islet IL-1b and 20 min at room temperature. Sections ages, as we have done previously (29). amyloid positivity among groups. In com- were then incubated in primary antisera. Endocrine cell areas were expressed as parison of differences between lung For primary antisera raised in mouse, percentage of total pancreatic tissue transplant groups, two-way ANOVA was goat, or guinea pig, linking IgGs (Leica area. Individuals performing IHC and used to test for interactions between Post Primary, rabbit anti-goat IgG, or rab- quantitative analysis thereof were blinded transplant and diabetes status on IL-1b bit anti-guinea pig IgG, respectively) were to the group assignment of specimens. and amyloid deposition. Linear regression then applied (since the Leica polymer re- Additionally, sections underwent im- was used to determine correlations be- agent recognizes rabbit IgG). Antibody munofluorescence labeling for insulin tween insulin and glucagon areas. binding was detected by poly–horseradish (I2018, clone K36AC10 [mouse] at 1:2,000; peroxidase polymerized secondary detec- Sigma-Aldrich, St. Louis, MO) followed by tion and Leica Bond Mixed Refine 3,39- Cy3-conjugated goat anti-mouse IgGs (di- RESULTS diaminobenzidene detection reagents luted 1:200; Jackson ImmunoResearch, Clinical Data (both DS9800; Leica Biosystems) followed West Grove, PA) to aid the visualization of Subjects in the CFRD group were on aver- by hematoxylin-eosin counterstaining and islets and counterstaining with thioflavin age older than in the CF-no DM group coverslipping. S (0.5% v/v) to visualize amyloid deposition. (Table 1); however, by design the non-CF Whole pancreas sections were then Islets were classified as thioflavin S positive (older) and non-CF (younger) subject digitized (Nanozoomer Digital Pathology or negative by manual counting of at least groups were matched for age and BMI system; Hamamatsu Corporation, Bridge- 50 islets per section, from which the pro- to the CFRD and CF-no DM groups, re- water, NJ). Islets (an average of 53 per portion of amyloid-positive islets was de- spectively. Pancreatic insufficiency was section) were classified as IL-1b positive termined for all subjects. Again, individuals documented in all but three CF case sub- or negative by manual counting. This was performing (immuno)histochemistry and jects (for whom data were not available), done by three independent observers, quantitative analysis thereof were blinded regardless of diabetes status. Subjects blinded to subject grouping (R.L.H., S.S., to the group assignment of specimens. with CFRD had diabetes for ;5 years, 4 Islet IL-1b Immunoreactivity in Cystic Fibrosis Diabetes Care

and all but two were insulin-treated. A his- this IL-1b positivity was observed in 29% who had been classified IL-1b positive tory of lung transplantation was more com- of islets in this group. In contrast, only with the ProSci antibody but showed no monintheCFRDcomparedwithCF-noDM 10% of non-CF (younger) subjects ex- IL-1b immunoreactivity with the Abcam group, and steroid use closely correlated hibited any IL-1b–positive islets (P , antibody.Webelievethediscrepancyin with this parameter; only one non-CF subject 0.001 vs. CF-no DM), comprising only this finding is due to lower sensitivity of the had received organ (heart) transplantation. 6% of islets per subject. No IL-1b immu- latter antibody. noreactivity was observed in the islets of b Islet IL-1b Immunoreactivity Is a non-CF (older) subjects (Fig. 1M). Valida- Islet IL-1 Immunoreactivity Is Seen Common Feature of CF tion of islet IL-1b immunoreactivity was Even in Very Young Subjects With CF We first determined whether there was performed using a second anti–IL-1b an- We next analyzed islet IL-1b immunore- an increased IL-1b immunoreactivity in tibody (Abcam) in a subset of subjects (n = activity (using the ProSci antibody) in the the islets of subjects with CF (ProSci anti- 10 [representative examples are shown in subset of pediatric CF subjects ,10 years body) (Fig. 1). Eighty-nine percent of Supplementary Fig. 1]). These subjects of age, the currently recommended age to CFRD subjects exhibited some degree were from both CF and non-CF groups and begin screening CF patients for CFRD with of islet IL-1b positivity, with 57% of the included subjects in whom IL-1b had been oral glucose tolerance tests (5). All six sub- islets on average from each subject being previously found to be positive or nega- jects with CF in this age range were in the involved (Fig. 1 M). Further, islet IL-1b tive. The presence or absence of IL-1b was CF-no DM group. Five of these showed ev- was also observed in 76% of CF-no DM confirmed in 9 out of 10 subjects. The last idence of IL-1b immunoreactivity affecting case subjects (P = 0.02 vs. CFRD), and subject was a non-CF (younger) subject 24% of islets per subject and thus were re- flective of the CF-no DM group as a whole.

Islet Amyloid Occurs Predominantly in CD-DM Subjects Islet amyloid was present in 61% of CFRD subjects, involving 20% of islets (Fig. 1N), while islet amyloid was present in 18% of CF-no DM subjects (P , 0.001 vs. CFRD), with only 3% of islets affected. Islet amyloid was not observed in either group of non-CF control subjects (Fig. 1N). Of note, none of the six young CF-no DM subjects ,10 years of age described above exhibited amyloid deposition.

Effect of Prior Lung Transplantation on Islet IL-1b and Amyloid Deposition Given the use of lung transplantation in advanced CF disease management, CF-no DM and CFRD subjects were subdivided based on lung transplant status and the presence of IL-1b and amyloid was analyzed. There was no statistically signiﬁcant interaction between transplant status and diabetes status in relation to islet IL-1b immunoreactivity or amyloid deposition (Table 2). Of note, a history of steroid use closely mirrored that of lung transplantation. Accordingly, subdivision of subjects based on steroid usage did not reveal differences in islet IL-1b immunoreactivity or amyloid deposition (data not shown).

CF Is Not Characterized by Islet IL-1Ra or CD68 Immunoreactivity Figure 1—IHC for insulin, IL-1b, and amyloid in subjects with and without CF. Representative We next sought to determine whether staining of pancreas specimens from subjects with non-CF control (A–D), CF-no diabetes (E–H), differences in the balance of IL-1b and and CFRD (I–L) groups, identifying islet b-cells using IHC for insulin (brown in A, E,andI and red in D, its antagonist, IL-1Ra, exist between sub- b H,andL), IL-1 (brown [images shown from two subjects per group in B and C, F and G,andJ and K), jects with CF with and without diabetes. and amyloid visualized by thioﬂavin S (ThioS) histochemistry (green in D, H,andL). Quantitation of the proportion of islets positive for IL-1b (M) and amyloid (N). Islet IL-1b positivity was increased in Based on data from older subjects with CF-no DM and CFRD subjects compared with age-matched non-CF control subjects, whereas islet and without type 2 diabetes (30), the ex- amyloid deposition was only increased in CFRD subjects. Scale bar = 50 mm. pectation is that high IL-1Ra levels would care.diabetesjournals.org Hull and Associates 5

Table 2—Subject characteristics and islet morphometric analyses in individuals with CF, subdivided according to diabetes and lung transplant status CF-no DM CFRD No lung transplant Lung transplant No lung transplant Lung transplant N 13 4 6 10 Age (years) 11.9 6 2.4 20.3 6 1.0 22.8 6 3.4 31.4 6 2.8 Sex (female/male) 11/2 1/3 3/3 6/4 BMI (kg/m2)15.26 0.6 20.3 6 2.8 21.9 6 1.3 22.0 6 1.2 Diabetes duration (years) dd6.6 6 3.1 4.2 6 1.0

HbA1c (%) (n =4–6) dd7.8 6 0.9 6.5 6 0.4 Subjects with islet IL-1b, n (%) 11 (91) 2 (50) 6 (100) 8 (80) IL-1b–positive islets (%) 28.8 6 8.5 31.1 6 18.8 63.5 6 9.5 61.8 6 12.3 Subjects with islet amyloid, n (%) 2 (11) 1 (25) 3 (50) 7 (70) Amyloid-positive islets (%) 3.3 6 3.2 0.3 6 0.3 5.2 6 3.2 31.0 6 12.2 Insulin-positive area (%) 1.5 6 0.3 2.8 6 0.9 1.1 6 0.4 1.3 6 0.2 Glucagon-positive area (%) 1.2 6 0.3 3.5 6 0.6 1.5 6 0.5 2.5 6 0.7 Somatostatin-positive area (%) (n = 8, 3, 4, and 5, respectively) 0.5 6 0.1 0.6 6 0.2 0.2 6 0.1 0.7 6 0.3 PP-positive area (%) (n = 8, 3, 4, and 5, respectively) 0.6 6 0.1 0.1 6 0.1* 0.2 6 0.1 0.2 6 0.1 Data are mean 6 SEM or n unless otherwise indicated. *P , 0.05 vs. CF-no DM, no transplant group. be present in CF-no DM subjects, with When normalized to the total pancreatic in more than half of subjects with CF with- lower levels being seen in CFRD subjects. tissue area, there was no significant dif- out diabetes, and in the vast majority of Contrary to this expectation, however, lit- ference in insulin-positive area between subjects with CFRD, whereas islet IL-1b tle to no IL-1Ra immunoreactivity was ob- the groups (Fig. 2A). Insulin-positive area immunoreactivity was essentially absent served in islets from any group despite was also computed as a proportion of from age-matched subjects without CF. In the use of two well-validated antibodies. islet area in a subset of subjects (n =5–10 line with previous studies (19,20), islet With the first antiserum (AF-280-NA), per group), yielding the same result amyloid was observed in subjects with IL-1Ra staining was indistinguishable (b-cell area/islet area was 39 6 0.6% in CFRD. However, islet IL-1b was far more from that seen with an isotype-matched CF-no DM, 27 6 0.2% in CFRD, and 39 6 frequently observed than islet amyloid in irrelevant IgG (data not shown). With the 0.4% in non-CF control subject; P =0.13 CF subjects, both those with and those second (NBP1-32568), only sparse islet for the comparison of CF-no DM with without diabetes, and therefore likely staining was detected, with no difference CFRD and P =0.95forCF-noDMvs. does not occur as a consequence of islet in IL-1Ra immunoreactivity being observed non-CF control). However, a significant in- amyloid deposition. This increase in islet between subjects with and subjects without crease in glucagon immunoreactivity was IL-1b in CFRD versus CF-no DM could diabetes (Supplementary Fig. 2). observed in CF-no DM and CFRD subjects have occurred as a result of the increased Pancreas sections were also stained for compared with non-CF control subjects age (and possibly age-related changes CD68 to investigate whether macrophage (Fig. 2B). A significant correlation existed such as fibrosis) in the former cohort infiltration could explain the presence of between insulin- and glucagon-positive rather than the presence/absence of di- islet IL-1b. While CD68-positive cells were areas in non-CF subjects (younger and abetes. Importantly, however, we ob- clearly present in exocrine pancreas of older cohorts) and in CF-no DM subjects served islet IL-1b immunoreactivity in subjects with CF-no DM and CFRD, no in- but not in CFRD subjects (Fig. 2C–F). In a pediatric subjects ,10yearsofage,the filtration of macrophages was observed subset of patients, pancreas sections were currently recommended age to begin within islets, regardless of IL-1b positivity also stained for somatostatin and PP. While screening CF patients for CFRD with oral in all cases but one (Supplementary Fig. 3), no significant difference between groups was glucose tolerance tests (5). The preva- suggesting that the observed islet IL-1b observed for somatostatin immunoreac- lence of islet IL-1b immunoreactivity in immunoreactivity did not derive from mac- tivity, CF-no DM subjects exhibited statis- this young cohort was similar to that in rophages. Only in one case, an infant (in the tically greater pancreatic polypeptide the CF-no DM group as a whole, sug- CF-no DM group), was intra-islet CD68 im- area compared with CFRD subjects. (Fig. gesting that islet inflammation could munoreactivity shown (Supplementary Fig. 2G and H). As with IL-1b and amyloid, we begin very early in CF patients, consis- 3E), suggesting that islet macrophages detected no significant interaction between tent with clinical studies showing de- may be present during early stages of CF. transplant status (or steroid use) and diabe- rangements in glucose tolerance and tes status in predicting islet hormone immu- b-cell function in very young subjects CF and CFRD Are Characterized by noreactivity in CF subjects (Table 2 and data with CF (2–4). Given that IL-1b is known Abnormal Islet Hormone Cell not shown). to contribute to impaired islet function Composition and viability (31), increased production We next analyzed islet endocrine cell CONCLUSIONS of IL-1b in the islets of CF subjects may composition in non-CF (younger and In this study, we identified islet inflamma- contribute to impaired b-cell function older), CF-no DM, and CFRD subjects. tion, detected by IL-1b immunoreactivity, that characterizes CFRD. 6 Islet IL-1b Immunoreactivity in Cystic Fibrosis Diabetes Care

Figure 2—Quantitation of islet endocrine cell types in subjects with and subjects without CF. Insulin area (A) and glucagon area (B) normalized to total tissue area. Linear regression analysis for the relationship between relative insulin and glucagon areas in non-CF (younger) control subjects (C), non-CF (older) control subjects (D) and CF-no DM (E) and CFRD (F) subjects. In a subset of subjects (n=10per group), somatostatin area (G) and pancreatic polypeptide area (H) normalized to total tissue area.

Unexpectedly, in contrast to some pre- without diabetes (17–19) or control sub- comparison group. This could explain the vious studies, we did not observe a de- jects without CF (20). However, the latter discrepant ﬁndings; speciﬁcally, older crease in relative b-cell area in subjects study used an older cohort of control sub- subjects are likely to have a higher BMI with CFRD relative to subjects with CF jects without CF (mean age 65 years) as a than younger subjects (especially those care.diabetesjournals.org Hull and Associates 7

with CF), and increased BMI is associated preserved in this group. In contrast, there CF subjects without diabetes) may be off- with increased islet b-cell area (32). For was no such relationship between gluca- set by increased islet IL-1Ra and suggests this reason, we were careful to age, sex, gon and insulin areas in subjects with the possibility of a targeted intervention and BMI match our comparison groups CFRD, suggesting that abnormal islet with clinically available IL-Ra antagonists. without CF and without diabetes to avoid composition characterizes diabetes in CF Our study has some limitations. First, this potential confounder. Those previous subjects. premortem oral glucose tolerance test studies that reported decreased b-cell The finding of IL-1b in the islets of CF data were not available for the vast ma- area between subjects with CF, with and CFRD subjects suggests an activation jority of subjects, precluding the deter- and without diabetes, included small of the NLRP3 inflammasome in islet cells mination of glucose tolerance status, numbers of subjects (n =6–7) (17,18), (likely b-cells). The NLRP3 inflammasome particularly in the CF-no diabetes group. whereas a larger study (n =16CFRD) is an intracellular protein complex that Further, our analysis of inflammatory cy- also failed to find a significant difference responds to a variety of intracellular and tokine immunoreactivity was limited to in b-cell area between subjects with CF extracellular signals by triggering inflam- IL-1b, and our investigation into islet hor- with and subjects with CF without diabe- matory cascades, resulting in IL-1b ex- mone immunoreactivity in some cases was tes (20). It should be noted that without pression (38,39). While we did observe limited to 10 subjects per group. In addi- pancreatic mass data (which has been immune cell infiltrates in exocrine pan- tion, given the limited number of subjects, shown to be decreased in subjects with creas of subjects with CF, these were not we could not adequately control for poten- CF [33]), it is not possible to definitively present within islets aside from one very tially confounding factors such as concom- determine whether b-cell mass is pre- young subject, suggesting that they were itant medication usage. However, despite served in CF/CFRD. However, our present not the source of the intra-islet IL-1b im- the presence of these potential confound- data, in agreement with those of Couce munoreactivity. Our current data do not ers we were still able to identify islet IL-1b et al. (20), suggest that a predominantly reveal the mechanism of activation of in- immunoreactivity as a common feature functional defect underlies the impaired flammasome activation in CF subjects. of CF and exclude islet amyloid as a likely insulin secretion seen in subjects with One hypothesis could be that mutations cause of its presence. Finally, the cross- CFRD, which would be supported by ani- in the CFTR could generate intracellular sectional nature of the analysis of postmor- mal data showing that CF pigs exhibit re- stress in b-cells, resulting in inflamma- tem specimens limits our ability to definitively duced insulin release despite preservation some activation. Another possibility is characterize a sequential process in which of islet mass (16). that extensive exocrine tissue autolysis IL-1b production occurs before islet amy- We observed a striking increase in islet in CF subjects could prime and initiate loidosis. However, the development of a-cell area in both CF-no DM and CFRD inflammasome-mediated IL-1b produc- more refined animal models and the use groups compared with non-CF control tion. Finally, systemic inflammatory pro- of IL-1–related biomarkers may be useful subjects. The source of these new a-cells cesses in subjects with CF appear skewed in answering these questions both in the is unknown but could arise because of to a program of IL-1b–regulated cyto- laboratory and at the bedside. proliferation of existing a-cells, formation kines (e.g., IL-8, IL-17 [26–28]), suggesting In conclusion, we provide a histologic of new a-cells from ductal or other pre- that circulating levels of IL-1b itself may analysis of human CF pancreas speci- cursors, or transdifferentiation of b-cells, also be increased. This could in turn result mens, in subjects with or without diabe- the latter having been proposed as a in autostimulation of IL-1b expression in tes, in which we identified islet IL-1b source of increased a-cell mass in other islets, a phenomenon that has been docu- immunoreactivity as well as increased forms of diabetes or models thereof (34). mented in cultured human islets (40). The glucagon immunoreactivity as key abnor- These data are intriguing, given that sub- islet in CF is therefore also likely exposed malities in islet morphology. These find- jects with CF exhibit impaired glucagon to increased IL-1b from both local and ings may help explain islet failure and responses to stimuli such as hypoglyce- systemic sources, making it especially vul- observed metabolic abnormalities in pa- mia (35), suggesting that a-cell expansion nerable to the effects of this proinflam- tients with CF. Future studies should may occur as an attempt to compensate matory cytokine. Additional studies are focus on a better understanding of the for decreased a-cell function. Conversely, needed to investigate the mechanisms un- physiologic effects of CFTR on b-cell bi- these morphometric data correlate with derlying islet IL-1b production in CF/CFRD. ology as well as a broader exploration of previous clinical studies that have docu- The toxic actions of IL-1b can be atten- inflammasome activation in the islet of mented impaired suppression of gluca- uated by its endogenous antagonist, subjects with CF. In addition, the develop- gon responses to oral glucose tolerance IL-1Ra. In the context of type 2 diabetes, ment of IL-1b–related peripheral bio- tests in CFRD subjects (13,36) and in vitro it has been suggested that individuals markers may be an important tool to data showing that cystic fibrosis trans- without diabetes express high levels of help in the early detection of CFRD. These membrane conductance regulator (CFTR) IL-1Ra in islets, whereas subjects with di- findings also suggest a possible role for is a negative regulator of glucagon release abetes exhibit a loss of IL-1Ra, further ex- anti-inflammatory drugs in preventing (37), situations where glucagon levels acerbating the proinflammatory effects of the development of islet pathology in CF. would be increased under conditions of IL-1b (30). In the current study, we were CF. Interestingly, in subjects with CF with- able to detect only small numbers of out diabetes, glucagon area was corre- IL-1Ra–positive cells in islets from any sub- Acknowledgments. The authors thank Sonya lated with insulin area, suggesting that ject, regardless of CF or diabetes status. Heltshe for biostatistical consultation and Brian in the face of a-cell expansion, islet endo- This excludes the possibility that any effects Johnson and Megan Larmore for expert assis- crine cell composition was still somewhat of islet IL-1b production (for example, in tance with IHC and morphometric analyses. 8 Islet IL-1b Immunoreactivity in Cystic Fibrosis Diabetes Care

Funding. This research was performed with the Verlaufsdokumentation Registry. Diabetes Care 25. Westwell-Roper C, Dai DL, Soukhatcheva G, support of nPOD, a collaborative type 1 diabetes 2016;39:1338–1344 et al. IL-1 blockade attenuates islet amyloid poly- research project sponsored by JDRF. Organ pro- 8. Blackman SM, Commander CW, Watson C, peptide-induced proinflammatory cytokine re- curement organizationspartneringwithnPODto et al. Genetic modifiers of cystic fibrosis-related lease and pancreatic islet graft dysfunction. J provide research resources are listed at http:// diabetes. Diabetes 2013;62:3627–3635 Immunol 2011;187:2755–2765 www.jdrfnpod.org/for-partners/npod-partners/. 9. Blackman SM, Hsu S, Ritter SE, et al. A suscep- 26. Mulcahy EM, Hudson JB, Beggs SA, Reid DW, This work was supported by the University of tibility gene for type 2 diabetes confers substan- Roddam LF, Cooley MA. High peripheral blood Washington Cystic Fibrosis Research and Transla- tial risk for diabetes complicating cystic fibrosis. th17 percent associated with poor lung function tion Center (National Institute of Diabetes and Diabetologia 2009;52:1858–1865 in cystic fibrosis. PLoS One 2015;10:e0120912 Digestive and Kidney Diseases grant P30-DK- 10. Furgeri DT, Marson FA, Ribeiro AF, Bertuzzo 27. Tan HL, Regamey N, Brown S, Bush A, Lloyd 089507), through a Pilot and Feasibility Award CS. Association between the IVS4G.Tmutationin CM, Davies JC. The Th17 pathway in cystic fibrosis (to R.L.H. and S.S.) and through work performed the TCF7L2 gene and susceptibility to diabetes in lung disease. Am J Respir Crit Care Med 2011;184: in the Host Response and Clinical Cores. This work cystic fibrosis patients. BMC Res Notes 2012;5: 252–258 was also supported by the University of Washing- 561 28. Tiringer K, Treis A, Fucik P, et al. A Th17- and ton’s Institute of Translational Health Sciences, 11. Austin A, Kalhan SC, Orenstein D, Nixon P, Th2-skewed cytokine profile in cystic fibrosis lungs which receives support from the National Cen- Arslanian S. Roles of insulin resistance and beta- represents a potential risk factor for Pseudomo- ter for Advancing Translational Sciences (UL1 cell dysfunction in the pathogenesis of glucose nas aeruginosa infection. Am J Respir Crit Care TR000423) and Cystic Fibrosis Foundation Re- intolerance in cystic fibrosis. J Clin Endocrinol Med 2013;187:621–629 search Development Program awards (R565CR11, Metab 1994;79:80–85 29. Rountree AM, Reed BJ, Cummings BP, et al. SINGH15R0, and VERKMA15R0). 12. Elder DA, Wooldridge JL, Dolan LM, D’Alessio Loss of coupling between calcium influx, energy Duality of Interest. No potential conflicts of in- DA. Glucose tolerance, insulin secretion, and in- consumption and insulin secretion associated terest relevant to this article were reported. sulin sensitivity in children and adolescents with with development of hyperglycaemia in the Author Contributions. R.L.H., R.L.G., C.W.F., B.W.R., cystic fibrosis and no prior history of diabetes. J UCD-T2DM rat model of type 2 diabetes. Diabe- and S.S. conceived the study and designed the Pediatr 2007;151:653–658 tologia 2013;56:803–813 experiments. R.L.H., R.L.G., S.M., G.H.D., C.L.F., 13. Lanng S, Thorsteinsson B, Røder ME, Nerup J, 30. Maedler K, Sergeev P, Ehses JA, et al. Leptin C.W.F., and S.S. collected and analyzed data. R.L.H. Koch C. Insulin sensitivity and insulin clearance in modulates beta cell expression of IL-1 receptor and S.S. performed data analysis and wrote the cystic fibrosis patients with normal and diabetic antagonist and release of IL-1beta in human islets. manuscript. R.L.G., G.H.D., C.L.F., C.W.F., and B.W.R. glucose tolerance. Clin Endocrinol (Oxf) 1994;41: Proc Natl Acad Sci U S A 2004;101:8138–8143 revised the manuscript. All authors approved the 217–223 31. Cnop M, Welsh N, Jonas JC, Jorns¨ A, Lenzen S, final version of the manuscript. R.L.H. and S.S. are 14. Olivier AK, Yi Y, Sun X, et al. Abnormal endo- Eizirik DL. Mechanisms of pancreatic beta-cell the guarantors of this work and, as such, had full crine pancreas function at birth in cystic fibrosis death in type 1 and type 2 diabetes: many differ- access to all the data in the study and take re- ferrets. J Clin Invest 2012;122:3755–3768 ences, few similarities. Diabetes 2005;54(Suppl. 2): sponsibility for the integrity of the data and the 15. Rogers CS, Stoltz DA, Meyerholz DK, et al. S97–S107 accuracy of the data analysis. Disruption of the CFTR gene produces a model 32. Butler AE, Janson J, Bonner-Weir S, Ritzel R, Prior Presentation. Partsofthisstudywerepre- of cystic fibrosis in newborn pigs. Science 2008; Rizza RA, Butler PC. Beta-cell deficit and increased sented in abstract form at the 74th Scientific Sessions 321:1837–1841 beta-cell apoptosis in humans with type 2 diabe- of the American Diabetes Association, San Francisco, 16. Uc A, Olivier AK, Griffin MA, et al. Glycaemic tes. Diabetes 2003;52:102–110 CA, 13–17 June 2014. regulation and insulin secretion are abnormal in 33. Sequeiros IM, Hester K, Callaway M, et al.; cystic fibrosis pigs despite sparing of islet cell Bristol Cystic Fibrosis Diabetes Group. MRI ap- mass. Clin Sci (Lond) 2015;128:131–142 pearance of the pancreas in patients with cystic References 17. Soejima K, Landing BH. Pancreatic islets in fibrosis: a comparison of pancreas volume in di- 1. Moran A, Dunitz J, Nathan B, Saeed A, Holme B, older patients with cystic fibrosis with and with- abetic and non-diabetic patients. Br J Radiol 2010; Thomas W. Cystic fibrosis-related diabetes: cur- out diabetes mellitus: morphometric and immu- 83:921–926 rent trends in prevalence, incidence, and mortal- nocytologic studies. Pediatr Pathol 1986;6:25–46 34. Talchai C, Xuan S, Lin HV, Sussel L, Accili D. ity. Diabetes Care 2009;32:1626–1631 18. Abdul-Karim FW, Dahms BB, Velasco ME, Pancreatic b cell dedifferentiation as a mecha- 2. Ode KL, Frohnert B, Laguna T, et al. Oral glu- Rodman HM. Islets of Langerhans in adolescents nism of diabetic b cell failure. Cell 2012;150: cose tolerance testing in children with cystic fibro- and adults with cystic fibrosis. A quantitative 1223–1234 sis. Pediatr Diabetes 2010;11:487–492 study. Arch Pathol Lab Med 1986;110:602–606 35. Moran A, Diem P, Klein DJ, Levitt MD, 3. Ode KL, Moran A. New insights into cystic 19. Iannucci A, Mukai K, Johnson D, Burke B. En- Robertson RP. Pancreatic endocrine function in fibrosis-related diabetes in children. Lancet Dia- docrine pancreas in cystic fibrosis: an immunohis- cystic fibrosis. J Pediatr 1991;118:715–723 betes Endocrinol 2013;1:52–58 tochemical study. Hum Pathol 1984;15:278–284 36. Hinds A, Sheehan AG, Machida H, Parsons 4. Yi Y, Norris AW, Wang K, et al. Abnormal glu- 20. Couce M, O’Brien TD, Moran A, Roche PC, HG. Postprandial hyperglycemia and pancreatic cose tolerance in infants and young children with Butler PC. Diabetes mellitus in cystic fibrosis is function in cystic fibrosis patients. Diabetes Res cystic fibrosis. Am J Respir Crit Care Med 2016; characterized by islet amyloidosis. J Clin Endocri- 1991;18:69–78 194:974–980 nol Metab 1996;81:1267–1272 37. Edlund A, Pedersen MG, Lindqvist A, Wierup 5. Moran A, Brunzell C, Cohen RC, et al.; CFRD 21. Clark A, Wells CA, Buley ID, et al. Islet amyloid, N, Flodstrom-Tullberg¨ M, Eliasson L. CFTR is in- Guidelines Committee. Clinical care guidelines for increased A-cells, reduced B-cells and exocrine volved in the regulation of glucagon secretion in cystic fibrosis-related diabetes: a position state- fibrosis: quantitative changes in the pancreas in human and rodent alpha cells. Sci Rep 2017;7:90 ment of the American Diabetes Association type 2 diabetes. Diabetes Res 1988;9:151–159 38. Vanaja SK, Rathinam VA, Fitzgerald KA. Mech- and a clinical practice guideline of the Cystic 22. Jurgens CA, Toukatly MN, Fligner CL, et al. anisms of inflammasome activation: recent ad- Fibrosis Foundation, endorsed by the Pediatric b-Cell loss and b-cell apoptosis in human type 2 vances and novel insights. Trends Cell Biol 2015; Endocrine Society. Diabetes Care 2010;33:2697– diabetes are related to islet amyloid deposition. 25:308–315 2708 Am J Pathol 2011;178:2632–2640 39. Guo H, Callaway JB, Ting JP. Inflammasomes: 6. Lanng S, Thorsteinsson B, Pociot F, et al. Diabe- 23. Westermark P. Quantitative studies on amy- mechanism of action, role in disease, and thera- tes mellitus in cystic fibrosis: genetic and immuno- loid in the islets of Langerhans. Ups J Med Sci peutics. Nat Med 2015;21:677–687 logical markers. Acta Paediatr 1993;82:150–154 1972;77:91–94 40. Boni-Schnetzler¨ M, Thorne J, Parnaud G, et al. 7. Konrad K, Kapellen T, Lilienthal E, et al.; DPV 24. Masters SL, Dunne A, Subramanian SL, et al. Increased interleukin (IL)-1beta messenger ribo- Initiative and the Competence Network Diabetes Activation of the NLRP3 inflammasome by islet nucleic acid expression in beta -cells of individuals Mellitus. Does b-cell autoimmunity play a role amyloid polypeptide provides a mechanism for with type 2 diabetes and regulation of IL-1beta in in cystic fibrosis-related diabetes? Analysis based enhanced IL-1b in type 2 diabetes. Nat Immunol human islets by glucose and autostimulation. J on the German/Austrian Diabetes Patienten 2010;11:897–904 Clin Endocrinol Metab 2008;93:4065–4074