Journal of Autoimmunity xxx (2017) 1e10

Contents lists available at ScienceDirect

Journal of Autoimmunity
journal homepage: www.elsevier.com/locate/jautimm

Anti-IL-21 monoclonal antibody combined with liraglutide effectively

reverses established hyperglycemia in mouse models of type 1
diabetes
Anna K. Ryden a, b, Nikole R. Perdue a, Philippe P. Pagni a, Claire B. Gibson a,
Sowbarnika S. Ratliff c, Rikke K. Kirk d, Travis J. Friesen a, Claus Haase e, Ken Coppieters a,
Matthias G. von Herrath a, Tamar E. Boursalian a, *
a
Type 1 Diabetes R&D Center, Novo Nordisk Inc., Seattle, WA, USA
b
Pacific Northwest Diabetes Research Institute, Seattle, WA, USA
c
Type 1 Diabetes Center, La Jolla Institute for Allergy and Immunology, La Jolla, CA, USA
d
Department of Histology and Imaging, Novo Nordisk A/S, Måløv, Denmark
e
Department of Immunopharmacology, Novo Nordisk A/S, Måløv, Denmark

a r t i c l e i n f o a b s t r a c t

Article history: Immunotherapy for type 1 diabetes (T1D) has previously focused on suppressing the autoimmune
Received 11 April 2017 response against pancreatic beta cells to preserve endogenous insulin production and regulate glucose
Received in revised form levels. With increased attention toward combination therapy strategies, studies indicate the multi-
28 June 2017
functional cytokine interleukin-21 (IL-21) may be a suitable target as an immuno-modulatory arm, while
Accepted 5 July 2017
Available online xxx
glucagon-like peptide-1 receptor (GLP-1R) agonists may be appropriate as a beta cell protective arm in
combination therapy for T1D. We report here that treatment with anti-IL-21 monoclonal antibody delays
diabetes onset in the spontaneous non-obese diabetic (NOD) and NOD.scid adoptive transfer models,
Keywords:
Type 1 diabetes
while its effect in reversing recent-onset hyperglycemia is limited. However, the combination of anti-IL-
Immunotherapy 21 plus the GLP-1R agonist liraglutide is effective in reversing established disease compared to either
Combination therapy monotherapy in both the NOD and rat insulin promotor-lymphocytic choriomeningitis virus glycoprotein
IL-21 (RIP-LCMV-GP) models of autoimmune diabetes. Enhanced efficacy is particularly evident in severely
GLP-1 hyperglycemic mice, with return to normoglycemia remaining stable for the majority of mice even after
Mouse models therapy is withdrawn. Importantly, increased beta cell proliferation does not appear to be the pre-
dominant mechanism. In conclusion, combination therapy with anti-IL-21 and liraglutide is able to
consistently reverse disease in mouse models of T1D. The observed effects rival the most effective
experimental disease-modifying treatments tested in preclinical studies.
© 2017 Published by Elsevier Ltd.

1. Introduction exogenous insulin. Previous clinical trials aimed at the suppression

of autoimmunity in T1D include targeting of T cells with anti-CD3
Type 1 diabetes (T1D) is an autoimmune disease characterized [1,2] or CTLA-4-Ig [3], or B cells with anti-CD20 [4], typically
by infiltration of immune cells into the pancreas, resulting in resulting in modest and transient preservation of stimulated C-
destruction of insulin-producing beta cells. The search for an peptide as a measure of functional beta cell mass. Such limited
effective therapy has focused on both suppression of autoimmunity effects may not translate to sufficient patient benefit to justify the
and restoration and/or preservation of beta cell function, so that safety profile of some of these agents [5,6].
patients are no longer dependent upon receiving lifelong Immune suppression can also be achieved by proinflammatory
cytokine blockade, aimed at dampening inflammation while pro-
moting immune deviation and/or regulatory pathways. In T1D, ef-
forts in targeting the cytokines interleukin (IL)-1, tumor necrosis
* Corresponding author. Type 1 Diabetes R&D Center, Novo Nordisk Inc., 530 factor alpha (TNFa), or IL-12/23 have shown limited clinical success
Fairview Ave N, Seattle, WA, USA.
[7], while tocilizumab, anti-IL-6 receptor, is currently in a Phase 2
E-mail address: [email protected] (T.E. Boursalian).

http://dx.doi.org/10.1016/j.jaut.2017.07.006
0896-8411/© 2017 Published by Elsevier Ltd.

n, et al., Anti-IL-21 monoclonal antibody combined with liraglutide effectively reverses established
Please cite this article in press as: A.K. Ryde
hyperglycemia in mouse models of type 1 diabetes, Journal of Autoimmunity (2017), http://dx.doi.org/10.1016/j.jaut.2017.07.006
2 A.K. Ryden et al. / Journal of Autoimmunity xxx (2017) 1e10

trial (NCT02293837). reversing established hyperglycemia in NOD mice, thus we com-

Given the limited success of immunotherapies to date, an bined anti-IL-21 with the GLP-1R agonist liraglutide in this setting
alternative approach toward achieving effective treatment for T1D and found significant enhancement of efficacy, reversing disease in
is needed. Development of combination therapies that target both nearly all mice treated.
autoimmunity and beta cell destruction, thus allowing long-term
protection of beta cell mass, may represent a more effective treat- 2. Materials and methods
ment modality for T1D. Ample evidence suggests that IL-21, a
member of the common g-chain family of cytokines, may be a 2.1. Mice
pivotal component of the pathogenic cascade in autoimmune dis-
eases, including T1D [8,9]. Thus, IL-21 blockade may be a suitable Female NOD mice in prevention and reversal studies were
candidate for the immuno-modulatory arm of a combination purchased from Jackson Laboratory, (Sacramento, CA, USA). Female
therapy. IL-21 has pleiotropic actions affecting the differentiation NOD and NOD.scid mice in NOD.scid transfer studies were pur-
and function of several immune cell types. It is produced mainly by chased from Taconic (Ry, Denmark). Male and female H-2b rat in-
T follicular helper (Tfh) cells, T helper (Th) 17 cells, and natural sulin promotor-lymphocytic choriomeningitis virus glycoprotein
killer T (NKT) cells, with lower levels produced by additional im- (RIP-LCMV-GP) mice were bred at the La Jolla Institute (LJI). All
mune cell types [10,11]. The IL-21 receptor (IL-21R) is broadly procedures were approved by the Institutional Animal Care and Use
expressed on lymphohematopoietic cells, resulting in IL-21 Committees of Novo Nordisk Research Center, Seattle, WA, USA,
responsiveness by a wide variety of cell types [9]. In humans, and LJI, La Jolla, CA, USA and The Danish Animal Inspectorate.
genome-wide association studies indicate that the IL-2/IL-21 region
on chromosome 4q27 is associated with T1D [12]. In a recent 2.2. Treatments
single-cell transcriptomic study of at-risk children, increased IL-21
expression by autoreactive CD4 T cells was among the most sig- Anti-mouse-IL-21 antibody, clone 397.18.2.1 (mouse IgG1,
nificant associations leading up to islet autoantibody development kappa), and isotype control anti-2,4,6-trinitrophenol (TNP), clone
[13]. In addition, patients with established T1D exhibit increased IL- DE8 (mouse IgG1, kappa) [26], were produced recombinantly at
21 production by CD4 effector T cells, increased circulating Tfh cell Novo Nordisk A/S. Binding to murine IL-21 and in vitro biological
numbers, and upregulation of Tfh associated genes, including IL-21 activity were confirmed for clone 397.18.2.1 (data not shown).
[14,15]. A Tfh signature was also found in islet-specific T cells in a Antibody stock solutions were diluted in PBS for intraperitoneal
transgenic mouse model of autoimmune diabetes [15]. Additional injection. Liraglutide stock solution was from Victoza® Flexpens®
preclinical studies demonstrate that the IL-21 pathway is required (Novo Nordisk, Måløv, DK) and diluted in PBS for subcutaneous
for disease development in the non-obese diabetic (NOD) model, injection.
where mice are predisposed to spontaneous development of
autoimmune diabetes, and in a viral variant of this model, either 2.3. Diabetes prevention and reversal in NOD mice
through ablation of IL-21R [16e18] or through preventive treat-
ment with soluble IL-21R, blocking IL-21 action [19]. Moreover, 2.3.1. Diabetes prevention in NOD mice
transgenic overexpression of IL-21 in pancreatic beta cells results in Experiments were performed at Novo Nordisk, Seattle. Treat-
leukocytic infiltration of islets and destruction of beta cells in ments began at 13 weeks of age. Treatment groups: Anti-IL-21
C57Bl/6 mice that are normally diabetes resistant [17]. Collectively, 25 mg/kg (3
/week for 2 weeks), 2.5 mg/kg (3
/week for 2
these studies indicate that IL-21 may govern diabetes development weeks), 25 mg/kg given once, and isotype control anti-TNP 25 mg/
by subtly affecting a range of leukocyte subsets, rather than kg (3
/week for 2 weeks). Mice were monitored once weekly for
depleting specific cell types (as in the case of anti-CD3 and anti- diabetes onset through 30 weeks of age (diabetes ¼ blood glucose
CD20 treatment) or blocking a single pathway (like CTLA-4-Ig). 250 mg/dL, on two consecutive days).
The hormone glucagon-like peptide-1 (GLP-1) is released from L
cells in the gut in response to food intake [20]. It functions as an 2.3.2. Diabetes reversal in NOD mice
incretin hormone, stimulating insulin release and inhibiting Experiments were performed at Novo Nordisk, Seattle. NOD
pancreatic glucagon secretion in a glucose-dependent manner. mice were screened twice weekly for diabetes onset
Furthermore, GLP-1 acts as a regulator of gastric emptying, and as a (diabetes ¼ blood glucose 250 mg/dL, on two consecutive days)
satiety signal in the brain, leading to reduced food intake [21]. GLP- and assigned to treatment groups as they became diabetic. Mice
1R agonists are currently approved for treatment of type 2 diabetes were screened for study inclusion from 10 to 26 weeks of age.
and obesity. In rodent models, GLP-1R signaling was shown to in- Treatment groups: untreated; anti-IL-21, 25 mg/kg (2
/week, 5
crease beta cell replication, decrease beta cell apoptosis, and induce total administrations); liraglutide 1 mg/kg (daily, 35 days, with
expansion of beta cell mass [22]. In preclinical studies where GLP- ramp-up of 0.3 mg/kg Day 1, 0.6 mg/kg Day 2, and 1 mg/kg Days
1R agonists were administered in combination with immune 3e35); and anti-IL-21 plus liraglutide, dosed as above for each
modulators such as anti-CD3 [23], anti-lymphocyte serum [24], or monotherapy. Blood glucose was monitored twice per week
lisofylline [25], remission was induced in a higher proportion of through 70 days post-onset, or until mice became terminally hy-
NOD mice compared to monotherapy treatment. Thus, GLP-1R perglycemic (blood glucose >600 mg/dL) at which time they were
agonists may be viable candidates to combine with an immune humanely euthanized.
modulator such as anti-IL-21 to treat T1D, with the hypothesis that
GLP-1R agonist therapy will protect/enhance remaining functional 2.4. NOD.scid adoptive transfer
beta cell mass, while anti-IL-21 immunotherapy will halt further
beta cell destruction. Experiments were performed at Novo Nordisk, Denmark.
Here we assessed the ability of anti-IL-21 monoclonal antibody Spleens were harvested from 14 to 16 week old NOD mice, red cells
therapy to prevent, delay, or reverse diabetes in mouse models of lysed, and a cell suspension prepared. Cells were injected intrave-
T1D. We found that anti-IL-21 monotherapy significantly delayed nously into 5 week old NOD.scid recipients (15
106 splenocytes
onset of diabetes in a dose-dependent manner in two distinct per animal). Twice weekly treatment with anti-IL-21 (fixed doses of
mouse models of T1D. Anti-IL-21 antibody was less effective in 625, 125, 31.3 or 6.25 mg per dose; roughly corresponding to 25, 5,

n, et al., Anti-IL-21 monoclonal antibody combined with liraglutide effectively reverses established
Please cite this article in press as: A.K. Ryde
hyperglycemia in mouse models of type 1 diabetes, Journal of Autoimmunity (2017), http://dx.doi.org/10.1016/j.jaut.2017.07.006
 A.K. Ryden et al. / Journal of Autoimmunity xxx (2017) 1e10 3

1.25, and 0.25 mg/kg, respectively) or isotype control antibody reversal studies, with the exception that liraglutide treatment
(625 mg per dose) began at cell transfer, with a 5
loading dose duration was four weeks.
given on the day of transfer. Treatment continued through the end
of observation (16 weeks of age) or diabetes diagnosis. Recipients 2.6. Histology
were monitored weekly for diabetes (diabetes ¼ blood glucose
250 mg/dL, two of 3 consecutive days). Diabetic recipients were 2.6.1. Histology for NOD.scid adoptive transfer experiments
euthanized at diagnosis. Pancreata were harvested at diabetes diagnosis or experimental
end. Three-mm sections were cut from formalin-fixed paraffin-
2.5. Diabetes reversal in RIP-LCMV-GP mice embedded (FFPE) tissues, hematoxylin/eosin (H&E) stained to
assess insulitis, and analyzed blindly on an Olympus light micro-
Experiments were performed at LJI. Three independent cohorts scope; insulitis scoring scale described in Fig. 1C legend, according
of 6e11 week-old RIP-LCMV-GP mice were infected with 104 to standards set at Novo Nordisk, Denmark. Adjacent sections were
plaque-forming units of LCMV strain Armstrong 53b, injected stained for insulin (polyclonal, Abcam, UK) to detect beta cells, and
intraperitoneally in a volume of 100 ml. Infected mice were moni- glucagon (polyclonal, Dako, Denmark), pancreatic polypeptide
tored twice daily for diabetes from day 7e13 post infection (polyclonal, Millipore, USA), and somatostatin (polyclonal, Dako,
(diabetes ¼ blood glucose 250 mg/dL for both daily measure- Denmark) for non-beta cells. Slides were scanned with a Hama-
ments). At diagnosis, mice were assigned to treatment groups and matsu NanoZoomer Digital Pathology System at 40X magnification,
blood glucose was monitored for four weeks post disease onset. and beta/non-beta stained sections were quantified for insulin
Treatment groups/regimens were as in Section 2.3.2 for NOD positive area using Visiopharm® software (Hoersholm, Denmark).

Fig. 1. IL-21 blockade delays diabetes onset and improves insulitis in mouse models of T1D. (A) Diabetes onset in NOD mice. Bracket beneath the x-axis indicates dosing period.
For the single 25 mg/kg Anti-IL-21 dose, antibody was administered at week 13. Circles ¼ untreated (n ¼ 19); diamonds ¼ isotype control (n ¼ 18); triangles ¼ anti-IL-21, 2.5 mg/kg
x 6 (n ¼ 16); open squares ¼ anti-IL-21, 25 mg/kg x 1 (n ¼ 20); closed squares ¼ anti-IL-21, 25 mg/kg x 6 (n ¼ 17). (B) Diabetes onset in adoptively transferred NOD.scid mice. Mice
received treatment twice a week starting at adoptive transfer (AdTr) of cells (indicated by arrow at x-axis) through endpoint. Circles ¼ untreated (n ¼ 10); open diamonds ¼ isotype
control (n ¼ 10); triangles ¼ anti-IL-21, 6.25 mg (n ¼ 8); inverted triangles ¼ anti-IL-21, 31.3 mg (n ¼ 9); closed diamonds ¼ anti-IL-21, 125 mg (n ¼ 9); squares ¼ anti-IL-21, 625 mg
(n ¼ 10). For both (A) and (B): BGV ¼ blood glucose value; Rx ¼ treatment. **p < 0.01, ***p < 0.001, ****p < 0.0001, ns ¼ not significant (Log-rank Mantel-Cox test). Statistical
significance beside group names in figure legends indicates comparison to isotype control group. Data are representative of three experiments. (C) Left: Insulitis scores from
treatment groups represented in (B). H&E stained pancreatic sections were scored using the following scale: White ¼ Score 0, no insulitis; diagonal stripes ¼ score 1, perivascular/
periductal infiltrates with leukocytes touching islet perimeters, not penetrating; dotted ¼ score 2, leukocytic penetration up to 25% of islet; vertical stripes ¼ score 3, leukocytic
penetration up to 75% of islet; black ¼ score 4, end-stage insulitis, <20% islet mass remaining. Fractions were calculated based on observations of 3e29 islets in 9e10 mice in
diabetic groups and 6e43 islets in 8e10 mice in anti-IL-21-treated groups. Statistically significant differences between groups for Score 0 (no insulitis) were determined by Fisher's
exact test: ****p < 0.0001 for the 625 mg treatment group compared to any other group, and for the 125 mg treatment group compared to any other group. No significant differences
were found when comparing any other treatment groups. Right: Insulinþ area in pancreata of mice from treatment groups represented in (B). Sections were stained for beta cells
(insulinþ) and non-beta cells (glucagonþ, pancreatic polypeptideþ, somatostatinþ). Circles ¼ untreated; open diamonds ¼ isotype control; triangles ¼ anti-IL-21, 6.25 mg; inverted
triangles ¼ anti-IL-21, 31.3 mg; closed diamonds ¼ anti-IL-21, 125 mg; squares ¼ anti-IL-21, 625 mg. Horizontal bars represent the mean for each group; error bars represent standard
deviation. Insulinþ area was quantified using Visiopharm® software. Significance was determined by non-parametric one-way ANOVA with Dunn's post-test for multiple com-
parisons. *p < 0.05, **p < 0.01, ***p < 0.001. All other comparisons were not significant.

n, et al., Anti-IL-21 monoclonal antibody combined with liraglutide effectively reverses established
Please cite this article in press as: A.K. Ryde
hyperglycemia in mouse models of type 1 diabetes, Journal of Autoimmunity (2017), http://dx.doi.org/10.1016/j.jaut.2017.07.006
4 A.K. Ryden et al. / Journal of Autoimmunity xxx (2017) 1e10

Fig. 2. Anti-IL-21 þ liraglutide is more effective in reversing disease and enhances survival compared to monotherapy, particularly in severely hyperglycemic mice. (A)
Blood glucose values during and after treatment with anti-IL-21 (top right), liraglutide (bottom left), anti-IL-21 þ liraglutide (bottom right), or untreated (top left). Each line
represents blood glucose values over time of an individual animal. Arrowheads on x-axis indicate anti-IL-21 administrations (right panels). Liraglutide was administered daily for 35
days (bottom panels). Grey shaded areas indicate withdrawal of liraglutide. N per group indicated in graph titles. (B) Kaplan-Meier plots for diabetes reversal (left) and survival
(right). Circles ¼ untreated; triangles ¼ liraglutide alone; squares ¼ anti-IL-21 alone; diamonds ¼ anti-IL-21 þ liraglutide. Left: Proportion of mice remaining diabetic through

n, et al., Anti-IL-21 monoclonal antibody combined with liraglutide effectively reverses established
Please cite this article in press as: A.K. Ryde
hyperglycemia in mouse models of type 1 diabetes, Journal of Autoimmunity (2017), http://dx.doi.org/10.1016/j.jaut.2017.07.006
 A.K. Ryden et al. / Journal of Autoimmunity xxx (2017) 1e10 5

2.6.2. Histology for NOD reversal experiments Mice were treated with anti-IL-21 at fixed doses of 625, 125, 31.3, or
Six-mm sections from frozen pancreata were H&E stained at the 6.25 mg/dose (roughly corresponding to 25, 5, 1.25, and 0.25 mg/kg,
Histology and Imaging Core, University of Washington (UW) respectively), twice a week from the time of cell transfer through 16
(Seattle, WA). Slides were scanned with a Hamamatsu NanoZoomer weeks of age (the end of observation) and monitored for diabetes
Digital Pathology System at 20X magnification. Insulitis scoring development. Similar to the NOD model, a dose-dependent pro-
from H& E staining was performed blindly using the scale described tection from disease development was observed in the NOD.scid
in Fig. 3B legend, according to standards set at Novo Nordisk, model (Fig. 1B). Anti-IL-21 administration at the highest dose
Seattle. Adjacent sections were stained for CD8 (biotin, clone (625 mg), completely prevented diabetes development, while
53e6.7, BD Biosciences, San Jose, CA) and insulin (polyclonal, Dako), 125 mg per dose provided full protection in nearly all mice. A sig-
and detected with streptavidin-AF647 (Thermo Fisher, Waltham, nificant delay was observed with 31.3 mg anti-IL-21, while no effect
MA) and goat anti-guinea pig IgG-AF594 (Invitrogen, Carlsbad, CA). was detected at the lowest dose or with isotype control treatment
Images were captured using a Zeiss AxioVert 200M inverted mi- when compared to untreated mice. The protective effect was not
croscope at 20X magnification. due to depletion of lymphocytes or a lymphocyte subset, as flow
cytometric analysis of peripheral blood showed no differences in
2.6.3. Insulin and Ki67/insulin area lymphocyte subsets between treated and untreated mice (data not
FFPE pancreata were collected 18 days post-diabetes onset. shown).
Four-mm sections were immunostained for insulin and Ki67 at UW The dose-dependent effect in NOD.scid transfer recipients was
on a Leica BOND-MAX automated immunohistochemistry staining further supported by histological analyses. While the two highest
platform using optimized staining protocols. Digital images were doses were equivalent in preventing hyperglycemia (Fig. 1B), the
scanned with a Hamamatsu NanoZoomer Digital Pathology System highest dose of anti-IL-21 (625 mg) resulted in a lower insulitis score
and Visiopharm® software was used for morphometric analyses. (Fig. 1C, left) and higher insulin-positive area at study termination
Areas were considered Ki67/insulin double-positive when Ki67- (Fig. 1C, right) compared to the next highest dose (125 mg), with
positive nuclei came in contact with 75% of insulin-positive area. reduced effects observed at each subsequent lower dosage. Thus,
when assessing insulitis and remaining insulin-containing areas,
2.7. Statistics the highest dose had the greatest effect on diabetes development,
although by measure of glycemic control, the two highest doses
Statistical analyses were performed using GraphPad Prism 6 were equally effective. Overall, the data demonstrate that neutral-
software (La Jolla, CA) and R version 3.3.1 (Vienna, Austria). Sta- izing anti-IL-21 can prevent or delay diabetes development in a
tistical tests applied to each data set are specified in each figure highly significant and dose-dependent manner in two distinct
legend. Statistically significant differences: *p < 0.05, **p < 0.01, models of T1D.
***p < 0.001, ****p < 0.0001.
3.2. Anti-IL-21 combined with liraglutide reverses hyperglycemia in
3. Results NOD mice

3.1. Anti-IL-21 delays diabetes onset in mouse models of T1D The promising results in preventing or delaying diabetes onset
with anti-IL-21 treatment led us to explore whether recent-onset
The effect of IL-21 blockade on diabetes onset was assessed in hyperglycemia could be reversed in NOD mice with anti-IL-21
the spontaneous NOD mouse model. At 13 weeks of age (late pre- treatment. A pilot study revealed that anti-IL-21 monotherapy
diabetic stage), NOD mice were treated with anti-IL-21 using administered at 25 mg/kg five times over two weeks had limited
dosing regimens that varied in the dose level or number of doses effect in reversing disease in NOD mice (data not shown). In an
administered (25 or 2.5 mg/kg, 3
/week for two weeks, or 25 mg/ effort to enhance the effect of anti-IL-21, the GLP-1R agonist lir-
kg given once) (Fig. 1A). Untreated and isotype control antibody aglutide was added to the treatment regimen, with the hypothesis
groups were included. All mice were euglycemic at study start, and that liraglutide may enhance beta cell viability/function while anti-
blood glucose was monitored through 30 weeks of age. Compared IL-21 combats autoimmunity. Mice were assigned to treatment
to controls, diabetes was significantly delayed in all treatment groups as they became diabetic, and a combination of anti-IL-21
groups in a dose-dependent manner. Notably, a statistically sig- given five times over two weeks plus daily liraglutide for 35 days
nificant delay was seen even in animals receiving a single injection was compared to each monotherapy and to untreated animals.
of anti-IL-21 at 25 mg/kg, and protection was achieved at this late- Mice were monitored through day 70 post-onset. Mean age and
stage pre-disease. mean blood glucose level at onset were comparable across groups
Anti-IL-21 treatment was also assessed for disease prevention or (data not shown). Once diabetic, all but two untreated mice did not
delay in the NOD.scid adoptive transfer model, where diabetes recover from hyperglycemia (Fig. 2A, top left). In preliminary
development relies on the transfer of spleen cells from NOD mice studies, liraglutide monotherapy did not prevent or delay diabetes
into immunodeficient NOD.scid mice. The transferred cells expand onset in NOD mice (data not shown). Similarly, liraglutide mono-
and ultimately cause autoimmunity in a more rapid, synchronized therapy did not affect progression to terminal hyperglycemia in
disease progression compared to the spontaneous NOD model. newly diabetic NOD mice, although in the first days of treatment a

treatment period. Right: Survival through day 70 post diabetes onset, survival defined as animals that did not become terminally hyperglycemic (blood glucose 600 mg/dL for 2
consecutive days) during the observation period. Arrowheads on x-axes indicate anti-IL-21 administrations. Liraglutide was administered daily for 35 days beginning at diagnosis.
Grey shaded area ¼ withdrawal of liraglutide. Statistically significant differences between groups are shown: *p < 0.05, **p < 0.01, ****p < 0.0001, ns ¼ not significant (Log-rank
Mantel-Cox test) Data are representative of two experiments. BGV ¼ blood glucose value. (C) Disease status of individual animals at 35 days post disease onset was plotted for anti-
IL-21 monotherapy (left) and anti-IL-21 þ liraglutide (right) treatment groups subdivided by age and blood glucose value at onset. Squares ¼ mice that were cured at day 35;
triangles ¼ mice that transiently reverted to normoglycemia; circles ¼ mice that remained diabetic. Shaded areas highlight mice with blood glucose 350 mg/dL at onset (the cut-
off for severe hyperglycemia). (D) Data only include mice with blood glucose 350 mg/dL at onset. White bars ¼ diabetic; black bars ¼ cured. Significance determined by Fisher's
exact test.

n, et al., Anti-IL-21 monoclonal antibody combined with liraglutide effectively reverses established
Please cite this article in press as: A.K. Ryde
hyperglycemia in mouse models of type 1 diabetes, Journal of Autoimmunity (2017), http://dx.doi.org/10.1016/j.jaut.2017.07.006
6 A.K. Ryden et al. / Journal of Autoimmunity xxx (2017) 1e10

temporary drop in blood glucose to pre-diabetic levels was noted in in insulitis scores determined from the H&E staining, showing that
10 of 18 mice (Fig. 2A, bottom left). Two liraglutide-treated mice in mice cured by either anti-IL-21 alone or combined with lir-
returned to normoglycemia through the dosing period of 35 days, aglutide, the extent of insulitis was decreased (Fig. 3B). Non-
but became hyperglycemic again upon liraglutide withdrawal. responders in these groups had insulitis similar to non-
Anti-IL-21 monotherapy reversed hyperglycemia in half (9/18) of responders in the other treatment groups. These data comple-
the mice (Fig. 2A, top right). Importantly, it was only when anti-IL- ment the efficacy data and illustrate the protective effect of anti-IL-
21 and liraglutide were combined that nearly all animals (16/18) 21 treatment in diabetic NOD mice, both as monotherapy and in
experienced lasting normalization of blood glucose throughout the combination with liraglutide.
35-day liraglutide treatment period (Fig. 2A, bottom right). Notably, Previous studies in mice demonstrate that liraglutide can induce
the majority of mice remained normoglycemic upon withdrawal of beta cell proliferation in vivo [27]. Thus, the proliferation marker
liraglutide through day 70. Kaplan-Meier survival plots highlight Ki67 was used to examine whether the observed enhanced efficacy
the enhanced efficacy of anti-IL-21 plus liraglutide combination of combination therapy could be attributed to beta cell prolifera-
therapy and confirm that the difference between monotherapy and tion. Mice were treated with either anti-IL-21 or liraglutide alone,
combination therapy at the end of treatment was statistically sig- anti-IL-21 plus liraglutide, or were left untreated. Samples were
nificant (Fig. 2B, left). Additionally, combined treatment of anti-IL- collected 18 days post-onset, corresponding to four days after the
21 with liraglutide resulted in improved survival compared to final anti-IL-21 dose, so as not to wash out effects from this treat-
either monotherapy or untreated mice throughout 70 days of ment arm. This resulted in decreased duration of liraglutide treat-
observation (Fig. 2B, right). ment as compared to the efficacy studies. Pancreata were stained
To explore whether the enhanced efficacy of combination for Ki67 and insulin, and the frequency of Ki67þinsulinþ area per
therapy compared to anti-IL-21 monotherapy was related to either total insulinþ area was determined. In eight separate sections
age at onset or level of glycemia at onset, a post-hoc comparison of sampled from various areas of each pancreas, beta cell proliferation
disease status was carried out between these two groups at 35 days was rarely detected (Fig. 4). Moreover, sections from individual
post-onset (the end of daily liraglutide treatment), subdivided by animals showed considerable variability in the extent of beta cell
age and blood glucose value at onset (Fig. 2C). Based on the proliferation. Overall, the data indicate no difference in the extent
observation that the anti-IL-21 monotherapy group appeared to of beta cell proliferation between treatment groups or between
show a trend toward correlation between glycemia at disease onset mice that were cured versus those that were not cured. Thus, beta
and efficacy of treatment, we built a single parameter logistic cell proliferation does not appear to be the predominant underlying
regression model on this treatment group, which yielded a blood mechanism behind the increased efficacy observed when liraglu-
glucose at onset of approximately 350 mg/dL as the discriminator tide is added to an anti-IL-21 treatment regimen.
between treatment efficacy and failure (p ¼ 0.00075, Chi-square
ANOVA). We then built Cox Proportional Hazard Models for the 3.4. Anti-IL-21 combined with liraglutide reverses hyperglycemia in
anti-IL-21 monotherapy and anti-IL-21 plus liraglutide combina- the RIP-LCMV-GP model
tion groups to investigate the effect of age at onset and blood
glucose at onset on treatment outcome. A univariate model Monotherapy and combination treatment regimens analogous
confirmed that while blood glucose at onset was a significant factor to those employed in the NOD studies were tested in the acute RIP-
for disease outcome following treatment (p ¼ 0.0028), age at onset LCMV-GP diabetes model on the C57BL/6J genetic background. In
was not (p ¼ 0.70). For mice with blood glucose <350 mg/dL at these mice, beta cells in the pancreas transgenically express a
onset, the proportion of diabetes reversal was similar between the glycoprotein from LCMV. Following viral infection, the immune
two groups (Fig. 2C, bottom quadrants), with 87.5% (7/8) mice cured response redirects its activity toward beta cells, resulting in their
with anti-IL-21 monotherapy and 90.1% (10/11) mice cured with rapid destruction and subsequent hyperglycemia in the mice. This
combination therapy. However, when comparing mice with blood model is considered more stringent than the recent-onset NOD
glucose 350 mg/dL at entry, the data reveal that treatment with model. Anti-IL-21, liraglutide, or combination therapy was admin-
anti-IL-21 plus liraglutide was more effective in reversing disease in istered to LCMV-infected mice from disease onset and mice were
severely hyperglycemic animals compared to anti-IL-21 alone, monitored for four weeks thereafter. Similar to the NOD findings,
regardless of age at onset (Fig. 2C, upper quadrants, and Fig. 2D). enhanced recovery from hyperglycemia was observed in the com-
Together these data indicate that addition of liraglutide to an anti- bination treatment group compared to either monotherapy (Fig. 5).
IL-21 treatment regimen may represent a viable option for T1D Taken together, these data indicate that whereas anti-IL-21 alone
treatment. has limited capacity to reverse hyperglycemia in diabetic mice,
anti-IL-21 combined with liraglutide is highly effective in disease
3.3. Anti-IL-21 plus liraglutide treatment results in reduced insulitis reversal in two separate mouse models of T1D.
but has no effect on beta cell proliferation
4. Discussion
To investigate possible mechanisms by which treatment effects
were achieved, histological analyses were performed on pancreatic The present study provides preclinical support for therapeutic
tissue from NOD mice receiving anti-IL-21 or liraglutide mono- use of combined anti-IL-21 and liraglutide treatment to preserve
therapy, combination therapy, or no treatment. Fig. 3A shows functional beta cell mass in newly diagnosed patients with T1D. A
representative images of pancreata stained with H&E to assess islet clinical proof-of-principle trial is currently ongoing in adults with
morphology and insulitis (top), or stained for CD8 and insulin newly diagnosed T1D to investigate applicability of this concept in
(bottom) to assess the presence of CD8 T cells within islet infiltrates. humans (NCT02443155). The benefits of sustained endogenous
Untreated and liraglutide-treated mice were terminally hypergly- insulin secretion post-diagnosis are well established and include
cemic at sample collection and thus had few visible islets remain- reduction of both short- and long-term complications from dia-
ing. These pancreata exhibited heavy cellular infiltration, whereas betes [28]. Thus, a treatment reducing or halting the advanced
mice cured with anti-IL-21 alone or in combination with liraglutide immunopathological process of beta cell destruction around clin-
exhibited decreased cellular infiltration. Infiltrates contained both ical diagnosis is highly anticipated. Several clinical trials, mainly
CD8 (bottom panels) and CD4 T cells (not shown). This is reflected using immune modulators approved for other indications,

n, et al., Anti-IL-21 monoclonal antibody combined with liraglutide effectively reverses established
Please cite this article in press as: A.K. Ryde
hyperglycemia in mouse models of type 1 diabetes, Journal of Autoimmunity (2017), http://dx.doi.org/10.1016/j.jaut.2017.07.006
 A.K. Ryden et al. / Journal of Autoimmunity xxx (2017) 1e10 7

Fig. 3. Decreased insulitis and islet destruction in mice treated with anti-IL-21 with or without liraglutide. (A) Representative histology images from all treatment groups. Top
row: H&E staining; bottom row: CD8 (green) and insulin (red) staining. Blood glucose at sample collection for images shown: Untreated and liraglutide, >600 mg/dL; anti-IL-21,
128 mg/dL; anti-IL-21 þ liraglutide, 146 mg/dL. (B) Insulitis scores from all treatment groups. Treatment groups and n per group are indicated on the x-axis. Semi-quantitative
insulitis scoring from H&E staining: White ¼ Score 0; no insulitis; dotted ¼ score 1; perivascular/periductal infiltrates with leukocytes touching islet perimeters, not pene-
trating; diagonal stripes ¼ score 2; <50% of islet area infiltrated; vertical stripes ¼ score 3; >50% of islet area infiltrated with remaining beta cell mass; checkered ¼ score 4 > 50% of
islet area infiltrated with no remaining beta cell mass; black ¼ score 5; islet atrophy with no insulin staining and no/few infiltrates. Scoring scale here (recent onset setting) differs
from that in Fig. 1C (prevention setting). In recent onset, mice are treated after diagnosis and thus a sizeable portion of islets are either atrophied or have infiltration with no insulin,
less frequently seen in the prevention setting. Fractions were calculated based on observations of 8e20 islets per mouse in diabetic groups and 21e24 islets per mouse in nor-
moglycemic groups. In each of the normoglycemic groups, 3 mice were analyzed at day 35 and 3 at 70 days post onset. Statistically significant differences between groups for Score
0 (no insulitis) are shown: *p < 0.05, **p < 0.01, ***p < 0.001 (Fisher's exact test). No significant differences were found between cured mice treated with anti-IL-21 compared to
mice cured with anti-IL-21 þ liraglutide for each score. Note a trend for reduced frequency of islets of score 4 (heavy infiltration with no beta cell mass left) in diabetic mice treated
with anti-IL-21 þ liraglutide compared to mice treated with anti-IL-21. However, with only one diabetic mouse presenting no more than 8 islets in the anti-IL-21 þ liraglutide group,
the trend did not reach statistical significance.

demonstrate that blockade of immunological pathways can result in distinct induced and spontaneous disease models on different
in significant, albeit temporary, preservation of beta cell function genetic backgrounds is recommended to increase translational
[29]. However, in order to safely achieve the degree of beta cell potential to human disease [32], which we have effectively
maintenance associated with clinically meaningful benefit, com- accomplished in these studies. When assessing the present data in
bination therapy targeting multiple disease pathways may be RIP-LCMV-GP mice in particular, it should be noted that diabetes
necessary [30]. reversal has rarely been achieved in this model, and only partially
The animal models employed herein were selected for their seen with anti-CD3 or select combination therapies [33e35]. The
documented stringency, to more faithfully mimic the advanced combined dataset presented here, therefore, suggests that anti-IL-
disease state found at the time of T1D diagnosis in humans. In NOD, 21 plus liraglutide combination therapy displays efficacy compa-
there is a general consensus that early disease stages are relatively rable to some of the most powerful immuno-modulatory agents
easy to treat, while successful prevention in older animals and tested in animal models for T1D.
particularly treatment of overtly hyperglycemic mice has been A substantial body of data indicates a pivotal role for IL-21 in the
accomplished by only a handful of agents [31]. Likewise, more acute T1D disease process. How this cytokine ultimately drives beta cell
disease models such as the NOD.scid transfer and RIP-LCMV-GP decay is incompletely understood, but both pathogenic and pro-
models are notoriously difficult to treat. Finally, preclinical testing tective immune cell types are thought to be affected by IL-21R

n, et al., Anti-IL-21 monoclonal antibody combined with liraglutide effectively reverses established
Please cite this article in press as: A.K. Ryde
hyperglycemia in mouse models of type 1 diabetes, Journal of Autoimmunity (2017), http://dx.doi.org/10.1016/j.jaut.2017.07.006
8 A.K. Ryden et al. / Journal of Autoimmunity xxx (2017) 1e10

Fig. 4. There is no significant increase in beta cell proliferation in mice treated with anti-IL-21, liraglutide, or both. FFPE pancreatic tissues were collected from mice 18 days
post-diabetes onset. Eight 4-mm sections cut 40 mm apart were stained for Ki67 and insulin. Ki67þ and/or insulinþ areas were quantified using Visiopharm® software.
Circles ¼ untreated; squares ¼ liraglutide alone; triangles ¼ anti-IL-21 alone; diamonds ¼ anti-IL-21 þ liraglutide. C ¼ animals that were cured at endpoint. Each vertical set of
symbols represents the eight sections analyzed from a single animal. Note that for all animals, there are multiple data points from the eight independent measurements at 0%. Data
points were averaged for each animal in each treatment group, and the mean of those averages was determined for a given treatment group. These means were analyzed by non-
parametric one-way ANOVA with Dunn's post-test for multiple comparisons and showed no statistical differences among treatment groups (data not shown).

data from healthy subjects and patients with rheumatoid arthritis

treated with recombinant anti-IL-21 monoclonal antibody indeed
suggest that disease modification can be achieved without
concomitant increases in adverse events [40].
Despite its ability to prevent diabetes in our models, anti-IL-21
monotherapy was less effective in restoring glycemic control in
diabetic animals, confirming previously published data using an IL-
21R-Fc fusion protein [19]. We therefore reasoned that a second
agent, acting on the metabolic component of the disease process,
was required to induce remission in established diabetes. Previous
studies suggest that GLP-1R agonists may help improve beta cell
survival and function [41e44]. Studies have demonstrated the po-
tential of liraglutide to protect beta cells against various forms of
endoplasmic reticulum (ER) stress, a state associated with inade-
quate insulin release and apoptosis [27,45,46]. Liraglutide was
recently tested in two phase 3 trials as add-on to insulin treatment
in patients with long-standing T1D [47,48]. Liraglutide reduced
Fig. 5. Anti-IL-21 plus liraglutide reverses disease in the RIP-LCMV-GP model of HbA1c levels, insulin requirements, and body weight but was
T1D. Mice infected with LCMV were monitored twice daily for diabetes from day 7e13 accompanied by increased rates of symptomatic hypoglycemia and
post infection. At diagnosis, mice were assigned to treatment groups and assessed for hyperglycemia with ketosis, thereby limiting clinical use in this
four weeks post disease onset for diabetes reversal. Circles ¼ isotype control; group. However, subgroup analyses found that C-peptide-positive
triangles ¼ liraglutide alone; squares ¼ anti-IL-21 alone; diamonds ¼ anti-IL-
21 þ liraglutide. Arrowheads on x-axis indicate anti-IL-21 administrations. Liraglutide
subjects showed a tendency toward improved HbA1c and lower
was administered daily throughout the observation period. *p < 0.05, **p < 0.01 (Log- rates of symptomatic hypoglycemia and hyperglycemia with
rank, Mantel-Cox test). Cumulative incidence from 3 independent cohorts of mice is ketosis, compared to C-peptide-negative participants [48]. These
included. data suggest that liraglutide could act on beta cell functionality in
C-peptide-positive subjects with T1D and could, therefore, be
signaling. Importantly, many IL-21 responsive cell types, including useful when combined with an immune modulator in this patient
CD8 T cells [36], CD4 regulatory T cells [37], and dendritic cells [18], population. Our present study supports this hypothesis in two
have been shown to be involved in the pathology of human T1D distinct animal models for recently diagnosed, C-peptide-positive
[38], making IL-21 an attractive pharmacological target [9]. Due to T1D.
the multifunctional nature of IL-21, pharmacological blockade is The precise mechanism underlying the observed enhanced ef-
likely to affect several leukocyte lineages. However, in contrast to fect, including the notable maintenance of glycemic control
other immuno-modulatory agents [1,2,4,39] we found no evidence following treatment withdrawal, remains to be fully characterized.
of global quantitative changes in leukocyte populations among It is tantalizing to speculate that the combination of a short course
treatment groups (data not shown). Nonetheless, the potent effects of anti-IL-21 with concomitant daily liraglutide treatment re-
of anti-IL-21 on islet infiltration and prevention of hyperglycemia instates immunological tolerance, allowing the remaining beta cell
shown here indicate that IL-21 is an essential component of the mass to survive, regain functionality and eventually control blood
diabetogenic process. While neutralizing IL-21 function results in glucose. In a flow cytometric phenotypic analysis comparing
subtle, non-depleting effects on leukocyte populations required for lymphocyte subsets between treated and untreated mice, we were
host defense against pathogens and tumor surveillance, its effect on unable to detect significant differences in populations of T cells
the immunopathology of autoimmune diabetes is profound. Safety (including naïve, effector and central memory, Treg and Tfh), B cells

n, et al., Anti-IL-21 monoclonal antibody combined with liraglutide effectively reverses established
Please cite this article in press as: A.K. Ryde
hyperglycemia in mouse models of type 1 diabetes, Journal of Autoimmunity (2017), http://dx.doi.org/10.1016/j.jaut.2017.07.006
 A.K. Ryden et al. / Journal of Autoimmunity xxx (2017) 1e10 9

(including mature B, plasma cells, plasmablasts, germinal center, Conflict of interest

and memory B), NK cells, dendritic cells, or monocytes/macro-
phages (data not shown). Likewise, longitudinal analysis in NOD Novo Nordisk markets liraglutide for the treatment of type 2
prevention and reversal studies of peripheral blood autoreactive diabetes and obesity, but not for type 1 diabetes. A.K.R., N.R.P.,
CD8 T cells specific for islet-specific glucose-6-phosphatase cata- C.B.G., R.K.K., T.J.F., P.P.P, C.H., K.C., M.G.v.H., and T.E.B. are employees
lytic subunit-related protein (IGRP) was not informative, as their of Novo Nordisk. No other potential conflicts of interest relevant to
presence in the peripheral blood of NOD mice decreases signifi- this article were reported.
cantly after 16 weeks of age whether treated or not ([49] and data
not shown). This limits analysis in the NOD prevention studies Author contributions
since the treatment window was from 13 to 15 weeks of age, and in
NOD reversal studies would not yield interpretable results with A.K.R., N.R.P., C.B.G., S.S.R., R.K.K., and T.J.F. performed experi-
mice becoming diabetic at variable ages. Furthermore, in an ments. P.P.P. designed the RIP-LCMV studies, interpreted the results,
extensive panel of serum cytokines/chemokines assayed in our and contributed to the manuscript. C.H. designed and interpreted
studies, the majority of analytes tested where either below detec- the NOD prevention and NOD.scid studies. T.E.B. designed and
tion limits or did not show significant differences between treat- interpreted the NOD prevention and reversal studies, performed
ment groups. However, there were a few subtle yet intriguing experiments and wrote and edited the manuscript. K.C. wrote the
trends in treated mice (Supplementary Table 1), most of which discussion and edited the manuscript. M.G.v.H conceived of the
were not statistically significant. For example, in mice that received studies.
combination treatment, we noted an increase in IL-22 which is
thought to alleviate oxidative and ER stress and restore mucosal Acknowledgments
immunity in diabetes [50e52]. Interestingly, IL-22 is linked to
upregulation of the pancreatic beta cell regenerative genes, Reg1 The authors are grateful for the excellent technical assistance of
and Reg2 [53], and mice lacking IL-21R also show increased mes- Justen Cracraft and Jaimie Granger (Novo Nordisk Research Center,
sage for Reg family members Reg2 and PAP [16], suggesting a Seattle), Stine Bisgaard, Rose B. Kildetoft, Mie Berndorff, Julie Jensen
possible additive or synergistic effect of the combination therapy. and Camilla F. Sorensen (Novo Nordisk A/S, Denmark), and Malina
There were also increases in the pro-inflammatory cytokines McClure (LJI) in animal studies, Brian Johnson and Megan Larmore
interferon (IFN)-g and TNFa in all treated mice with trends toward (Histology and Imaging Core, University of Washington) in histo-
greater increase in the combination group. While this seems on first logical assays, Jon Rue (Novo Nordisk Research Center, Seattle) in
thought counterintuitive, there are reports of these cytokines statistical analyses, and Erinn Lanxon-Cookson (Novo Nordisk
having a protective role in disease development in NOD mice Research Center, Seattle) in cytokine assays. The authors
[54,55]. Increased granulocyte macrophage colony-stimulating acknowledge the editorial assistance of Ellie Ling.
factor (GM-CSF) in liraglutide-treated groups was seen, both as T.E.B. is the guarantor of this work and, as such, had full access to
monotherapy and in combination with anti-IL-21 which is also all the data in the study and takes responsibility for the integrity of
intriguing given a report that GM-CSF can prevent diabetes devel- the data and the accuracy of the data analysis.
opment in NOD mice [56]. Additionally, we noted an increase in This research did not receive any specific grant from funding
three chemokines, most notably macrophage inflammatory protein agencies in the public, commercial, or not-for-profit sectors.
(MIP)-3a, which suggests the possibility that this systemic increase
in chemokines could be drawing harmful cells away from the local Appendix A. Supplementary data
disease site. Collectively, these data may not provide conclusive
information on exact mechanism of action, but do point toward a Supplementary data related to this article can be found at http://
subtle treatment-mediated immune modulation that appears to be dx.doi.org/10.1016/j.jaut.2017.07.006.
enhanced with the addition of liraglutide to anti-IL-21 treatment.
In mice, GLP-1R agonists can induce beta cell proliferation,
References
leading to increased beta cell mass [27,57]; however, beta cell
proliferation in human adults is a rare phenomenon and its un- [1] K.C. Herold, S.E. Gitelman, U. Masharani, W. Hagopian, B. Bisikirska,
controlled occurrence should obviously be avoided [58]. Our pre- D. Donaldson, et al., A single course of anti-CD3 monoclonal antibody
sent data do not point to increased beta cell proliferation as the hOKT3gamma1(Ala-Ala) results in improvement in C-peptide responses and
clinical parameters for at least 2 years after onset of type 1 diabetes, Diabetes
underlying mechanism behind the potent efficacy seen in mice 54 (2005) 1763e1769.
treated with liraglutide plus anti-IL-21. Likewise, unchanged beta [2] B. Keymeulen, M. Walter, C. Mathieu, L. Kaufman, F. Gorus, R. Hilbrands, et al.,
cell proliferation was reported in mice effectively treated for hy- Four-year metabolic outcome of a randomised controlled CD3-antibody trial
in recent-onset type 1 diabetic patients depends on their age and baseline
perglycemia with combined anti-CD3 and the GLP-1R agonist residual beta cell mass, Diabetologia 53 (2010) 614e623.
exendin-4 [23]. Ongoing mechanistic studies for the enhanced [3] T. Orban, B. Bundy, D.J. Becker, L.A. DiMeglio, S.E. Gitelman, R. Goland, et al.,
treatment effect will focus on exploration of liraglutide-mediated Co-stimulation modulation with abatacept in patients with recent-onset type
1 diabetes: a randomised, double-blind, placebo-controlled trial, Lancet 378
reduction of apoptosis and/or ER stress in beta cells. Finally, anti- (2011) 412e419.
IL-21 is unique in its role as a cytokine blocker in combination [4] M.D. Pescovitz, C.J. Greenbaum, H. Krause-Steinrauf, D.J. Becker, S.E. Gitelman,
therapy, as we found no such enhanced effect with anti-TNFa R. Goland, et al., Rituximab, B-lymphocyte depletion, and preservation of beta-
cell function, N. Engl. J. Med. 361 (2009) 2143e2152.
blockade combined with liraglutide in the RIP-LCMV-GP model [5] B. Keymeulen, S. Candon, S. Fafi-Kremer, A. Ziegler, M. Leruez-Ville,
(data not shown). C. Mathieu, et al., Transient Epstein-Barr virus reactivation in CD3 monoclonal
In conclusion, combination therapy with anti-IL-21 and lir- antibody-treated patients, Blood 115 (2010) 1145e1155.
[6] J.L. Kroll, C. Beam, S. Li, R. Viscidi, B. Dighero, A. Cho, et al., Reactivation of
aglutide is able to consistently reverse disease in mouse models of
latent viruses in individuals receiving rituximab for new onset type 1 diabetes,
T1D. The observed effects rival the most effective experimental J. Clin. Virol. 57 (2013) 115e119.
disease-modifying treatments tested in preclinical studies to date. [7] G.T. Nepom, M. Ehlers, T. Mandrup-Poulsen, Anti-cytokine therapies in T1D:
An ongoing clinical proof-of-principle trial in human adults with concepts and strategies, Clin. Immunol. 149 (2013) 279e285.
[8] T. Gharibi, J. Majidi, T. Kazemi, R. Dehghanzadeh, M. Motallebnezhad,
newly diagnosed T1D will evaluate this combination therapy versus Z. Babaloo, Biological effects of IL-21 on different immune cells and its role in
monotherapy and placebo (NCT02443155). autoimmune diseases, Immunobiology 221 (2016) 357e367.

n, et al., Anti-IL-21 monoclonal antibody combined with liraglutide effectively reverses established
Please cite this article in press as: A.K. Ryde
hyperglycemia in mouse models of type 1 diabetes, Journal of Autoimmunity (2017), http://dx.doi.org/10.1016/j.jaut.2017.07.006
10 A.K. Ryden et al. / Journal of Autoimmunity xxx (2017) 1e10

[9] R. Spolski, W.J. Leonard, Interleukin-21: a double-edged sword with thera- 21 promotes CD8þ CTL activity via the transcription factor T-bet, J. Immunol.
peutic potential, Nat. Rev. Drug Discov. 13 (2014) 379e395. 190 (2013) 3977e3984.
[10] J. Parrish-Novak, S.R. Dillon, A. Nelson, A. Hammond, C. Sprecher, J.A. Gross, et [37] L.E. Clough, C.J. Wang, E.M. Schmidt, G. Booth, T.Z. Hou, G.A. Ryan, et al.,
al., Interleukin 21 and its receptor are involved in NK cell expansion and Release from regulatory T cell-mediated suppression during the onset of
regulation of lymphocyte function, Nature 408 (2000) 57e63. tissue-specific autoimmunity is associated with elevated IL-21, J. Immunol.
[11] R. Spolski, W.J. Leonard, Interleukin-21: basic biology and implications for 180 (2008) 5393e5401.
cancer and autoimmunity, Annu. Rev. Immunol. 26 (2008) 57e79. [38] T.L. van Belle, K.T. Coppieters, M.G. von Herrath, Type 1 diabetes: etiology,
[12] K. Asano, H. Ikegami, T. Fujisawa, M. Nishino, K. Nojima, Y. Kawabata, et al., immunology, and therapeutic strategies, Physiol. Rev. 91 (2011) 79e118.
Molecular scanning of interleukin-21 gene and genetic susceptibility to type 1 [39] M.R. Rigby, K.M. Harris, A. Pinckney, L.A. DiMeglio, M.S. Rendell, E.I. Felner, et
diabetes, Hum. Immunol. 68 (2007) 384e391. al., Alefacept provides sustained clinical and immunological effects in new-
[13] A.K. Heninger, A. Eugster, D. Kuehn, F. Buettner, M. Kuhn, A. Lindner, et al., onset type 1 diabetes patients, J. Clin. Investig. 125 (2015) 3285e3296.
A divergent population of autoantigen-responsive CD4þ T cells in infants [40] S. Ignatenko, B.K. Skrumsager, U. Mouritzen, Safety, PK, and PD of recombi-
prior to beta cell autoimmunity, Sci. Transl. Med. 9 (2017). nant anti-interleukin-21 monoclonal antibody in a first-in-human trial, Int. J.
[14] R.C. Ferreira, H.Z. Simons, W.S. Thompson, A.J. Cutler, X.C. Dopico, D.J. Smyth, Clin. Pharmacol. Ther. 54 (2016) 243e252.
et al., IL-21 production by CD4þ effector T cells and frequency of circulating [41] D.A. Cunha, L. Ladriere, F. Ortis, M. Igoillo-Esteve, E.N. Gurzov, R. Lupi, et al.,
follicular helper T cells are increased in type 1 diabetes patients, Diabetologia Glucagon-like peptide-1 agonists protect pancreatic beta-cells from lipotoxic
58 (2015) 781e790. endoplasmic reticulum stress through upregulation of BiP and JunB, Diabetes
[15] R. Kenefeck, C.J. Wang, T. Kapadi, L. Wardzinski, K. Attridge, L.E. Clough, et al., 58 (2009) 2851e2862.
Follicular helper T cell signature in type 1 diabetes, J. Clin. Investig. 125 (2015) [42] M. Ferdaoussi, S. Abdelli, J.Y. Yang, M. Cornu, G. Niederhauser, D. Favre, et al.,
292e303. Exendin-4 protects beta-cells from interleukin-1 beta-induced apoptosis by
[16] R. Spolski, M. Kashyap, C. Robinson, Z. Yu, W.J. Leonard, IL-21 signaling is interfering with the c-Jun NH2-terminal kinase pathway, Diabetes 57 (2008)
critical for the development of type I diabetes in the NOD mouse, Proc. Natl. 1205e1215.
Acad. Sci. U.S.A. 105 (2008) 14028e14033. [43] S. Tsunekawa, N. Yamamoto, K. Tsukamoto, Y. Itoh, Y. Kaneko, T. Kimura, et al.,
[17] A.P. Sutherland, T. Van Belle, A.L. Wurster, A. Suto, M. Michaud, D. Zhang, et Protection of pancreatic beta-cells by exendin-4 may involve the reduction of
al., Interleukin-21 is required for the development of type 1 diabetes in NOD endoplasmic reticulum stress; in vivo and in vitro studies, J. Endocrinol. 193
mice, Diabetes 58 (2009) 1144e1155. (2007) 65e74.
[18] T.L. Van Belle, S. Nierkens, R. Arens, M.G. von Herrath, Interleukin-21 [44] B. Yusta, L.L. Baggio, J.L. Estall, J.A. Koehler, D.P. Holland, H. Li, et al., GLP-1
receptor-mediated signals control autoreactive T cell infiltration in pancreatic receptor activation improves beta cell function and survival following in-
islets, Immunity 36 (2012) 1060e1072. duction of endoplasmic reticulum stress, Cell Metab. 4 (2006) 391e406.
[19] H.M. McGuire, S. Walters, A. Vogelzang, C.M. Lee, K.E. Webster, J. Sprent, et al., [45] M. Shimoda, Y. Kanda, S. Hamamoto, K. Tawaramoto, M. Hashiramoto,
Interleukin-21 is critically required in autoimmune and allogeneic responses M. Matsuki, et al., The human glucagon-like peptide-1 analogue liraglutide
to islet tissue in murine models, Diabetes 60 (2011) 867e875. preserves pancreatic beta cells via regulation of cell kinetics and suppression
[20] D.J. Drucker, Deciphering metabolic messages from the gut drives therapeutic of oxidative and endoplasmic reticulum stress in a mouse model of diabetes,
innovation: the 2014 Banting Lecture, Diabetes 64 (2015) 317e326. Diabetologia 54 (2011) 1098e1108.
[21] C.S. Suen, P. Burn, The potential of incretin-based therapies in type 1 diabetes, [46] L. Zhao, H. Guo, H. Chen, R.B. Petersen, L. Zheng, A. Peng, et al., Effect of Lir-
Drug Discov. Today 17 (2012) 89e95. aglutide on endoplasmic reticulum stress in diabetes, Biochem. Biophys. Res.
[22] J. Pettus, I. Hirsch, S. Edelman, GLP-1 agonists in type 1 diabetes, Clin. Commun. 441 (2013) 133e138.
Immunol. 149 (2013) 317e323. [47] B. Ahren, I.B. Hirsch, T.R. Pieber, C. Mathieu, F. Gomez-Peralta, T.K. Hansen, et
[23] N.A. Sherry, W. Chen, J.A. Kushner, M. Glandt, Q. Tang, S. Tsai, et al., Exendin-4 al., Efficacy and safety of liraglutide added to capped insulin treatment in
improves reversal of diabetes in NOD mice treated with anti-CD3 monoclonal subjects with type 1 diabetes: the adjunct two randomized trial, Diabetes Care
antibody by enhancing recovery of beta-cells, Endocrinology 148 (2007) 39 (2016) 1693e1701.
5136e5144. [48] C. Mathieu, B. Zinman, J.U. Hemmingsson, V. Woo, P. Colman, E. Christiansen,
[24] N. Ogawa, J.F. List, J.F. Habener, T. Maki, Cure of overt diabetes in NOD mice by et al., Efficacy and safety of liraglutide added to insulin treatment in type 1
transient treatment with anti-lymphocyte serum and exendin-4, Diabetes 53 diabetes: the adjunct one treat-to-target randomized trial, Diabetes Care 39
(2004) 1700e1705. (2016) 1702e1710.
[25] Z. Yang, M. Chen, J.D. Carter, C.S. Nunemaker, J.C. Garmey, S.D. Kimble, et al., [49] J.D. Trudeau, C. Kelly-Smith, C.B. Verchere, J.F. Elliott, J.P. Dutz, D.T. Finegood,
Combined treatment with lisofylline and exendin-4 reverses autoimmune et al., Prediction of spontaneous autoimmune diabetes in NOD mice by
diabetes, Biochem. Biophys. Res. Commun. 344 (2006) 1017e1022. quantification of autoreactive T cells in peripheral blood, J. Clin. Investig. 111
[26] G. Kohler, C. Milstein, Derivation of specific antibody-producing tissue culture (2003) 217e223.
and tumor lines by cell fusion, Eur. J. Immunol. 6 (1976) 511e519. [50] S.Z. Hasnain, D.J. Borg, B.E. Harcourt, H. Tong, Y.H. Sheng, C.P. Ng, et al., Gly-
[27] K. Tamura, K. Minami, M. Kudo, K. Iemoto, H. Takahashi, S. Seino, Liraglutide cemic control in diabetes is restored by therapeutic manipulation of cytokines
improves pancreatic Beta cell mass and function in alloxan-induced diabetic that regulate beta cell stress, Nat. Med. 20 (2014) 1417e1426.
mice, PloS One 10 (2015) e0126003. [51] M. Hu, S. Yang, L. Yang, Y. Cheng, H. Zhang, Interleukin-22 alleviated
[28] M.W. Steffes, S. Sibley, M. Jackson, W. Thomas, Beta-cell function and the palmitate-induced endoplasmic reticulum stress in INS-1 cells through acti-
development of diabetes-related complications in the diabetes control and vation of autophagy, PloS One 11 (2016) e0146818.
complications trial, Diabetes Care 26 (2003) 832e836. [52] X. Wang, N. Ota, P. Manzanillo, L. Kates, J. Zavala-Solorio, C. Eidenschenk, et al.,
[29] A.K. Ryden, J.D. Wesley, K.T. Coppieters, M.G. von Herrath, Non-antigenic and Interleukin-22 alleviates metabolic disorders and restores mucosal immunity
antigenic interventions in type 1 diabetes, Hum. Vaccin. Immunother. 10 in diabetes, Nature 514 (2014) 237e241.
(2014) 838e846. [53] T. Hill, O. Krougly, E. Nikoopour, S. Bellemore, E. Lee-Chan, L.A. Fouser, et al.,
[30] M. von Herrath, Combination therapies for type 1 diabetes: why not now? The involvement of interleukin-22 in the expression of pancreatic beta cell
Immunotherapy 2 (2010) 289e291. regenerative Reg genes, Cell Regen. (Lond) 2 (2013) 2.
[31] L.K. Shoda, D.L. Young, S. Ramanujan, C.C. Whiting, M.A. Atkinson, [54] D.O. Sobel, J. Han, J. Williams, J.W. Yoon, H.S. Jun, B. Ahvazi, Gamma interferon
J.A. Bluestone, et al., A comprehensive review of interventions in the NOD paradoxically inhibits the development of diabetes in the NOD mouse,
mouse and implications for translation, Immunity 23 (2005) 115e126. J. Autoimmun. 19 (2002) 129e137.
[32] B.O. Roep, M. Atkinson, M. von Herrath, Satisfaction (not) guaranteed: re- [55] X.D. Yang, R. Tisch, S.M. Singer, Z.A. Cao, R.S. Liblau, R.D. Schreiber, et al., Effect
evaluating the use of animal models of type 1 diabetes, Nat. Rev. Immunol. of tumor necrosis factor alpha on insulin-dependent diabetes mellitus in NOD
4 (2004) 989e997. mice. I. The early development of autoimmunity and the diabetogenic process,
[33] D. Bresson, L. Togher, E. Rodrigo, Y. Chen, J.A. Bluestone, K.C. Herold, et al., J. Exp. Med. 180 (1994) 995e1004.
Anti-CD3 and nasal proinsulin combination therapy enhances remission from [56] S. Gaudreau, C. Guindi, M. Menard, G. Besin, G. Dupuis, A. Amrani, Gran-
recent-onset autoimmune diabetes by inducing Tregs, J. Clin. Investig. 116 ulocyte-macrophage colony-stimulating factor prevents diabetes develop-
(2006) 1371e1381. ment in NOD mice by inducing tolerogenic dendritic cells that sustain the
[34] M.G. von Herrath, B. Coon, T. Wolfe, L. Chatenoud, Nonmitogenic CD3 anti- suppressive function of CD4þCD25þ regulatory T cells, J. Immunol. 179
body reverses virally induced (rat insulin promoter-lymphocytic choriome- (2007) 3638e3647.
ningitis virus) autoimmune diabetes without impeding viral clearance, [57] S. Xue, C. Wasserfall, M. Parker, S. McGrail, K. McGrail, M. Campbell-Thomp-
J. Immunol. 168 (2002) 933e941. son, et al., Exendin-4 treatment of nonobese diabetic mice increases beta-cell
[35] P.P. Pagni, D. Bresson, T. Rodriguez-Calvo, A. Bel Hani, Y. Manenkova, proliferation and fractional insulin reactive area, J. Diabetes Complicat. 24
N. Amirian, et al., Combination therapy with an anti-IL-1beta antibody and (2010) 163e167.
GAD65 DNA vaccine can reverse recent-onset diabetes in the RIP-GP mouse [58] S. Rutti, N.S. Sauter, K. Bouzakri, R. Prazak, P.A. Halban, M.Y. Donath, In vitro
model, Diabetes 63 (2014) 2015e2025. proliferation of adult human beta-cells, PloS One 7 (2012) e35801.
[36] A.P. Sutherland, N. Joller, M. Michaud, S.M. Liu, V.K. Kuchroo, M.J. Grusby, IL-

n, et al., Anti-IL-21 monoclonal antibody combined with liraglutide effectively reverses established
Please cite this article in press as: A.K. Ryde
hyperglycemia in mouse models of type 1 diabetes, Journal of Autoimmunity (2017), http://dx.doi.org/10.1016/j.jaut.2017.07.006