Skip to main content
NCBI home page
Endocrinology logoLink to Endocrinology
. 2021 Feb 4;162(10):bqab019. doi: 10.1210/endocr/bqab019

FSTL3-Neutralizing Antibodies Enhance Glucose-Responsive Insulin Secretion in Dysfunctional Male Mouse and Human Islets

Melissa L Brown 1,✉, Alexa Lopez 2, Nolan Meyer 2, Alden Richter 2, Thomas B Thompson 3
PMCID: PMC8384134  PMID: 33539535

Abstract

Diabetes is caused by insufficient insulin production from pancreatic beta cells or insufficient insulin action, leading to an inability to control blood glucose. While a wide range of treatments exist to alleviate the symptoms of diabetes, therapies addressing the root cause of diabetes through replacing lost beta cells with functional cells remain an object of active pursuit. We previously demonstrated that genetic deletion of Fstl3, a critical regulator of activin activity, enhanced beta cell number and glucose-responsive insulin production. These observations suggested the hypothesis that FSTL3 neutralization could be used to therapeutically enhance beta cell number and function in humans. To pursue this possibility, we developed an FSTL3-neutralizing antibody, FP-101, and characterized its ability to prevent or disrupt FSTL3 from complexing with activin or related ligands. This antibody was selective for FSTL3 relative to the closely related follistatin, thereby reducing the chance for off-target effects. In vitro assays with FP-101 and activin revealed that FP-101-mediated neutralization of FSTL3 can enhance both insulin secretion and glucose responsiveness to nonfunctional mouse and human islets under conditions that model diabetes. Thus, FSTL3 neutralization may provide a novel therapeutic strategy for treating diabetes through repairing dysfunctional beta cells.

Keywords: diabetes, beta cell, FSTL3, activin

Diabetes continues to result in a major public health burden worldwide. Type 1 diabetes (T1D) results from an autoimmune destruction of beta cells. Type 2 diabetes (T2D) results from a loss of beta cell function and/or mass that results in production of insufficient insulin to maintain normal blood glucose levels, in combination with insulin resistance (1-4). While a number of therapies are available to treat the symptoms of diabetes, including drugs that enhance insulin sensitivity or production, addressing the root cause of diabetes remains elusive. Active research efforts include enhancing human beta cell replication rates (5), production of beta cells from human embryonic or induced pluripotent stem cells (6, 7), and induction of beta cell regeneration from endogenous precursor cells or other islet cell types (8-10).

Activin is a growth factor in the transforming growth factor β (TGFβ) superfamily with many activities in embryos and adults that are regulated by extracellular antagonists like follistatin and follistatin like-3 (FSTL3) (11, 12). We previously demonstrated that deletion of Fstl3 in mice (FSTL3 knockout [KO] mice) produced enlarged pancreatic islets containing increased numbers of beta cells, greater insulin production, and enhanced glucose tolerance (13). Interestingly, beta cell proliferation and apoptosis were not altered in FSTL3 KO mice raising the question of the source of these additional beta cells (14).

Direct alpha to beta cell transdifferentiation was recently demonstrated using ectopically expressed transcription factors (15). This process also appears to be part of a homeostatic response to loss of beta cells since complete destruction of existing beta cells induced alpha to beta cell transdifferentiation even in the absence of genetic drivers (16). By crossing alpha cell lineage tracing mice with FSTL3 KO mice, we identified twice as many yellow fluorescent protein (YFP)-labeled cells containing insulin compared with wild-type littermates, suggesting that the FSTL3-activin pathway regulates a natural process of alpha to beta cell transdifferentiation that provides new beta cells and accounts for the enlarged islets in FSTL3 KO mice (17).

These observations suggested that FSTL3 neutralization could be used to enhance beta cell function as well as increase beta cell number through enhanced alpha to beta cell transdifferentiation. To pursue this possibility, we first developed an FSTL3-neutralizing antibody that enhances activin activity as our primary objective. Then, in order to determine the translational potential and justification for further development as a therapeutic agent, in vitro functionality studies with isolated islets were performed. The FSTL3-neutralizing antibody enhanced glucose-stimulated insulin secretion (GSIS) in dysfunctional mouse and human islets. Thus, FSTL3 neutralization may provide a novel therapeutic strategy for treating diabetes through repairing and/or replacing lost or dysfunctional beta cells that contribute to diabetes development (18).

Research Design and Methods

Production of Antibodies to Human FSTL3

Human FSTL3 (hFSTL3) was produced by recombinant expression as previously described (19). Recombinant hFSTL3 was used as immunogen for antibody production in mice by Abclonal Technology (Woburn, MA). Clone supernatants were screened for FSTL3 neutralization using the FSTL3 neutralization assay described below. Positive colonies were subcloned, expanded, and antibody purified by Protein-A chromatography (AKTA, GE Lifescience) according to manufacturer recommendations. Protein concentration was determined by BCA assay (Pierce). Additional FP-101 antibody was produced and purified by Abclonal (Lot 2019-3).

Protein Production

Mouse and rhesus FSTL3 and human follistatin (hFST) were initially produced in house as described above for hFSTL3. Additional FSTL3 from all 3 species was produced and purified by Lake Pharma (Belmont, CA). Human activin A and B, GDF11, and myostatin were purchased from R&D Systems (Minneapolis, MN). Exendin 4 was purchased from Sigma (E7144).

FSTL3 Neutralization Assay

HEK293 cells stably expressing an activin-responsive luciferase reporter (20, 21) were maintained in growth medium (DMEM [Corning; MT10017CV], penicillin-streptomycin [Corning; MT30002CL], 10% fetal bovine serum [Gibco 16140071], 100 μg/mL Geneticin [Gibco; 11811031]) in 5% CO2. For the assay, 20 000 cells/well were cultured in 100 μL of growth medium in 96-well Poly-L-Lysine (R&D Systems Cultrex) coated microtiter plates. The next day, treatments (50 μL/well) were prepared containing 0.012nM to 0.02nM Activin A, 0.0625nm to 0.25nM FSTL3, and other test materials such as antibodies in test medium (DMEM plus antibiotics and 0.1% bovine serum albumin). Samples were tested in triplicate and sufficient treatments for each triplicate were prepared in a single tube for 2 hours before adding to the cells. After 16 to 24 hours of treatment, 50 μL One-Glo (Promega) was added to each well for 3 minutes. The treatment medium/One-Glo mixture was transferred to a white isoplate (Perkin Elmer; 6005030) and read immediately in a 96-well luminescence plate reader. Antibody neutralization of FSTL3 was assessed by adding hybridoma supernatant (100 μL for initial screen) or purified antibody as indicated in the figures to medium containing 0.015nM activin A and 0.0625nM FSTL3. Treatments were then added to cells in triplicate.

Electrophoresis and Western Blot

Human, rhesus, and mouse FSTL3 and hFST were fractionated on precast 10% Tris-Glycine-SDS gels (Bio-Rad). Proteins were transferred to a 0.45-μm PVDF membrane (Millipore), blocked in 3% nonfat milk in TBST (10mM Tris, pH 8.0, 150mM NaCl, 0.5% Tween 20) overnight at 4 ºC or for 1 hour at room temperature. After washing in TBST, membranes were incubated with 1 μg/mL primary antibody (FP-101 or FP-102 or anti-hFST monoclonal [MAB669, R&D Systems]) (22) in 1% milk and TBST solution overnight at 4 ºC or 1 hour at room temperature. After washing 3 times with TBST, membranes were incubated with secondary HRP-donkey anti-mouse antibody (Jackson ImmunoResearch; 715-035-150) (23) for 1 hour at room temperature. Each membrane was then washed 3 times in TBST before developing with ECL substrate (PerkinElmer).

Mouse Islet Isolation and Analysis

Animal experiments were approved by the Baystate Medical Center Institutional Animal Care and Use Committee. Male C57BL/6J mice (~25-30 g, 10-12 weeks of age, Charles River or Jackson Laboratories) (24), were housed on a standard 12-hour light/dark cycle and fed a standard rodent diet (Teklad Global Diet 2918) or 1 of 2 high-fat diets (Teklad 06415 for 45% fat or Teklad 06414 for 60% fat) ad libitum with free access to water for 1 week. Mouse islets were isolated and tested for functionality by GSIS as described previously (25). After an overnight recovery from isolation, 10 similarly sized islets were treated with FP-101, activin B, or exendin 4 in triplicate as indicated in Fig. 5 for 24 hours. Insulin secretion in the low (2.0mM glucose) and high (20mM glucose) treatments were determined by insulin enzyme-linked immunosorbent assay (ELISA) (Mercodia; 10-1247-10) (26). Insulin concentrations were normalized as fold-difference from the average low glucose untreated control so that results from 3 to 10 separate experiments could be combined.

Figure 5.

Figure 5.

Open in a new tab

FP-101 and activin B enhance insulin secretion and glucose responsiveness to dysfunctional islets from HFD mice. Islets were isolated from chow-fed or high-fat diet (HFD)-fed mice and cultured at 10 similarly sized islets per well. Treatments were applied for 24 hours after which a glucose-stimulated insulin secretion assay (GSIS) was used to determine whether treatments altered glucose responsiveness. a) Results for control islets are shown as insulin secreted in the 1-hour GSIS treatment showing that high glucose-stimulated insulin secretion by 2.4-fold in chow islets but was reduced to 1.6 fold and not significant for islets from HFD-fed mice. N = 12. b) Islets from chow-fed mice shown normalized to low glucose in control islets and the stimulation index (SI) was significant, showing that these islets were functioning normally. Treatment with FP-101, activin or exendin 4 had no significant effect on GSIS when compared to untreated although insulin secretion with the lower dose of FP-101 tended to be reduced in high glucose. In comparison, islets from untreated HFD-fed (1 week) mice islets were not glucose responsive. Both FP-101 and activin B treatment enhanced glucose responsiveness to a greater degree than exendin 4, although not to chow levels and insulin secretion in these islets was significantly enhanced relative to untreated control islets (Low = 2.0mM glucose in GSIS; High = 20mM glucose in GSIS). *P < 0.05, **P < 0.01, ***P < 0.001 and NS = not significant. Results in B were normalized to mean of control chow islet low glucose so multiple experiments could be combined. N = 3-13 separate replicates. Red bars = low glucose. Blue bars = high glucose. Abbreviations: bA, activin A; bB, activin B; Ctrl, control; ex, exendin; FP, FP-101, HFD, high-fat diet.

Human Islet Experiments

Human islets were obtained from 6 separate shipments from Prodo Laboratories, Inc (Aliso Viejo, CA, USA). Upon arrival, islets were washed with RPMI-1640 (Corning) with 10% heat-inactivated fetal bovine serum and 1% (vol/vol) penicillin-streptomycin (Corning Cellgro). After overnight culture in fresh complete PIM(S) media (Prodo Laboratories), 10 similarly sized islets were tested for functional integrity by GSIS assay in triplicates as described previously (27).

Half of the islets were cultured for 24 hours in standard 5.8mM glucose PIM(S) medium while the other half was treated with hyperglycemic medium [28.5mM glucose PIM(S)] to induce a dysfunctional state reminiscent of diabetic islets. For the purposes of this study, dysfunctional was defined as a lack of response to glucose stimulation. After this incubation, all islets were washed with PIM(S) before being treated with FP-101, activin B, or exendin 4 as indicated in Fig. 6 for an additional 24 hours. Given the limited number of studies showing the effect in human islets, activin A was also tested, which is in contrast to the growing body of evidence of the effect in mouse islets. All islets were then assessed for beta cell function by GSIS. Insulin secretion during the GSIS was determined by ELISA (Mercodia; 10-113-10) (28). Insulin concentrations were transformed as fold-difference from the average low glucose untreated control so experimental replicates could be combined.

Figure 6.

Figure 6.

Open in a new tab

FP-101 and activin restore insulin secretion and glucose responsiveness to human islets cultured under hyperglycemic conditions. a) Insulin secretion from normal human islets or islets after 24 hours of hyperglycemia, showing that normal islets were functional with an SI of 2.6 whereas hyperglycemia-treated islets lost glucose responsiveness (N = 6). b) Ten similarly sized islets/well from normal donors and not exposed to hyperglycemia or treatments (control) responded to high glucose in a GSIS by significantly increased insulin secretion and a stimulation index (SI) of 2.5. Treatments with FP-101, activin A, or activin B had no detrimental effect on GSIS and were not statistically significant versus control, although insulin secretion in low-dose activin A under high glucose was reduced. In comparison, 10 similarly sized islets from the same normal donors but pretreated with elevated glucose (25mM HyperG) for 24 hours lost their glucose responsiveness while overall insulin secretion was suppressed. Glucose responsiveness and insulin secretion were restored to control islet level by treatment with FP-101 as well as with activin A and B. (Low = 2.0mM glucose in GSIS; High = 20mM glucose in GSIS). *P < 0.05, **P < 0.01, ***P < 0.001 and NS = not significant. For panel B, results were normalized to the mean of low glucose in normal islets to combine data from multiple experiments. N = 3-7 replicates. Red bars = low glucose. Blue bars = high glucose. Abbreviations: bA, activin A; bB, activin B; ex, exendin; FP, FP-101; HG, hyperG.

Statistical Analysis

Data are presented as means ± standard error of the mean. Differences between means were evaluated by 1-way analysis of variance (ANOVA) with Tukey post hoc adjustment for multiple comparisons, or the Student t test when appropriate using GraphPad Prism 5 (GraphPad Software, La Jolla, CA). A P value of ≤0.05 was considered significant.

Results

Selection of Antibody for FSTL3 Neutralization

Neutralization of FSTL3 may represent a novel strategy for a diabetes therapeutic that could directly address the loss and insufficiency of functional beta cell mass that is central to development of diabetes. This concept was based on our observation that inactivation of the Fstl3 gene in mice increased alpha to beta cell transdifferentiation as well as enhanced insulin production and glucose control (13, 17), In addition, because the structure and ligand binding pockets of follistatin and FSTL3 are similar (12, 29-31), an antibody approach to specifically neutralize FSTL3 had a greater probability of avoiding off-target effects that might occur if follistatin was also neutralized.

Thirty-seven mouse monoclonal antibodies to human FSTL3 demonstrated strong binding to human FSTL3 and were further evaluated for FSTL3 neutralization using a functional assay based on the biological activity of FSTL3 to inhibit activin signaling (21). Three neutralizing clones were expanded and purified antibody from each was compared for FSTL3 neutralization activity. In this assay, 0.012nM activin A alone stimulates between 80 000 and 100 000 relative light units (RLU) (Fig. 1 brown bar) which is reduced to 2000 RLU by addition of 0.625nM hFSTL3 (green bar). Purified FP-101 and FP-102 both retained neutralization activity, with FP-101 being consistently superior to the others (Fig. 1) while purified FP-103 had activity identical to a control antibody and was not analyzed further.

Figure 1.

Figure 1.

Open in a new tab

Identification of monoclonal antibodies to hFSTL3 with neutralizing activity. Monoclonal antibodies to hFSTL3 were screened for ability to neutralize hFSTL3 binding to activin, which releases bioactive activin from inhibition by FSTL3. This activin was quantified using HEK293 cells stably expressing the CAGA-Luc activin reporter as previously described (20). Signaling by activin alone is shown in the brown bar and inhibition by FSTL3 shown in green. Increasing concentrations of FP-101 (red) neutralize the ability of FSTL3 to bind activin as demonstrated by increasing activin bioactivity. FP-102 and 103 were less active. A non-neutralizing monoclonal antibody was used as an immunoglobulin control. Shown is mean of 3 experiments.

Characterization of Neutralizing Antibodies

To prevent possible unanticipated effects, it is critical that a therapeutic antibody designed to neutralize FSTL3 avoid binding and neutralization of the closely related follistatin (32). As shown in Fig. 2A, both FP-101 and FP-102 bound to nonreduced hFSTL3 but failed to bind to an identical amount of follistatin, demonstrating that both FP-101 and FP-102 were selective for FSTL3.

Figure 2.

Figure 2.

Open in a new tab

Characterization of FP-101 and 102 specificity for FSTL3. a) Equal amounts (100 ng) of recombinant hFSTL3 and the closely related human follistatin (hFST) were electrophoresed under nonreducing conditions and analyzed by Western blot. Both FSTL3 antibodies detect hFSTL3 while neither bound to hFST. An hFST antibody bound hFST as a control. b) Western blot demonstrating that FP-101 bound equally well to 100 ng hFSTL3 and rhesus FSTL3 (rhFSTL3) but bound less well to 200 ng mouse FSTL3 (mFSTL3). FP-102 bound hFSTL3 but did not bind to rh- or mFSTL3. c) Solid phase ELISA with 100 ng/well of mouse or human FSTL3 passively adsorbed to a 96-well plate. FP-101 bound hFSTL3 about 10-fold better than mFSTL3. Shown are representative results of 3 replicates.

The amino acid sequence of human and mouse FSTL3 (hFSTL3 and mFSTL3 respectively) is 80% identical while rhesus monkey (rhFSTL3) and human are 99% identical, suggesting that anti-human FSTL3 antibodies may recognize rhFSTL3 better than mFSTL3. By Western blot, FP-101 bound equally to 100 ng of human and rhesus FSTL3 while binding much less to 200 ng of mFSTL3. In contrast, FP-102 only bound hFSTL3 while binding to rh- and mFSTL3 was undetectable (Fig. 2B). When tested in ELISA format, FP-101 bound well to hFSTL3 adsorbed to a 96-well plate but much less so to mFSTL3 similarly immobilized, supporting the results from Western blotting (Fig. 2C). These results suggest that both neutralizing antibodies recognize hFSTL3 using at least some nonoverlapping epitopes and that FP-102 is less tolerant of nonhuman amino acid substitutions relative to FP-101.

FSTL3 has been demonstrated to antagonize the activity of 4 related TGFβ family ligands, including activin A and B, GDF11, and myostatin (GDF8) (19, 33, 34). To determine if antibodies that neutralized FSTL3 antagonism of activin A would also block antagonism of the related ligands, we substituted activin B, GDF11, or myostatin for activin A in the CAGA-luc assay and compared neutralization of FSTL3 binding to each ligand with activin A. To assist with the comparison, results are expressed relative to each ligand’s induction of RLU when tested alone. Both FP-101 and FP-102 completely neutralized hFSTL3 binding to all 4 ligands (Fig. 3), while a control antibody that binds hFSTL3 but does not release activin A also did not release the other ligands. Therefore, binding of FSTL3 to any of these ligands would be reduced in the presence of FP-101 or FP-102 so that the bioactivity of all 4 ligands would be enhanced in vivo in the presence of sufficient FP-101 or FP-102, which could therefore induce an outcome similar in fashion to what we observed with the FSTL3 KO mouse including enhanced insulin production and beta cell formation (13, 14, 17).

Figure 3.

Figure 3.

Open in a new tab

Antibody neutralization of related activin/TGFβ superfamily members. Using the same CAGA-Luc assay as in Fig. 1, the ability of FP-101 and FP-102 to neutralize binding to activin and related ligands was assessed. Results are expressed as a percent of ligand only wells (max signaling) to normalize for the different stimulation levels of each ligand. Both FP-101 (red) and FP-102 (blue) (66.6nM) neutralized FSTL3 inhibition of activin B (panel b), GDF11 (panel c), and myostatin (panel d) in addition to activin A (panel a). Control is a non-neutralizing mouse monoclonal antibody. Mean of 3 experiments. **P < 0.01, ***P < 0.001.

Antibody-Induced Disruption of Preformed Activin-FSTL3 Complexes in Islets

So far, FSTL3 neutralization by FP-101 and FP-102 has been characterized by adding activin, FSTL3 and antibody to culture medium at the same time and after 2 hours applying the mixture to cells for quantification of free activin (see Fig. 4). To investigate possible mechanisms for antibody neutralization of FSTL3, delayed addition of one reagent was utilized to determine if the antibody could disrupt preformed FSTL3-activin complexes, keeping in mind that we had previously demonstrated such complexes to have a pM affinity with a very slow dissociation constant, rendering the complex nearly irreversible (32, 35). This is critical since it is likely that most activin and related ligands within tissues such as islets exists bound to antagonists like FSTL3 or follistatin and a therapeutic might be more potent if it could liberate activin from these preexisting complexes (36).

Figure 4.

Figure 4.

Open in a new tab

FP-101 can disrupt preformed FSTL3-activin complexes. Using the same CAGA-Luc assay as in Fig. 1, the ability of FP-101 to disrupt preformed FSTL3-activin complexes was assessed. a) Total activin signaling is shown by the black bar and inhibition by FSTL3 shown by the shaded bar, both of which were added to all wells. b) Neutralization by FP-101 when all 3 components are added simultaneously is shown in the green bars. In the delayed activin condition (red bars), antibody and FSTL3 are added simultaneously and after 2 hours, activin is added and the treatment is added to cells. This represents maximal neutralization since maximal FSTL3 is bound by antibody. In the delayed FSTL3 treatment (blue bars), activin and antibody are added simultaneously but don’t interact, so neutralization doesn’t begin until FSTL3 is added 2 hours later. In the delayed antibody condition (orange), FSTL3 and activin are added simultaneously and form neutralized complexes while FP-101 is added after 2 hours. In this case, FP-101 can release 90% of the activin activity released under the other conditions, demonstrating that FP-101 can disrupt preformed FSTL3-activin complexes. This property is specific for FP-101, since FP-102 and nonimmune IgG do not neutralize FSTL3 under any of those conditions. Mean of 3 replicates shown. ***P < 0.001.

Typically, we assess neutralization by antibody with simultaneous addition of treatments as shown in Fig. 4. However, delayed addition of activin allows assessment of maximal FSTL3 neutralization, since FSTL3 and antibody have 2 hours to bind before activin is added to the mixture. This resulted in nearly all activin being released from FSTL3 by FP-101 since activin activity was not statistically different from simultaneous addition of all 3 components. When FSTL3 addition was delayed, leaving only FP-101 and activin in the preincubation mixture, neutralization was also maximal since FP-101 does not bind directly to activin. Interestingly, when FP-101 itself was delayed for 2 hours, allowing activin to first complex with FSTL3, activin activity nevertheless reached 90% of, and was not statistically different from, simultaneous addition and delayed activin. FP-102 and a nonimmune mouse IgG had essentially no binding or neutralization activity, demonstrating that FSTL3 neutralization is specific to FP-101. Moreover, these results demonstrate that FP-101 can disrupt nearly all preformed, and otherwise nearly irreversible, activin-FSTL3 complexes.

FP-101 Treatment Enhances Glucose Responsiveness to Dysfunctional Mouse Islets

We previously demonstrated that mouse islets produce activin A and B and myostatin RNA and activin A and B protein mostly in alpha cells, while FSTL3 mRNA was detected primarily in beta cells (25). Furthermore, activin treatment had no detectable effect on normal, functional mouse islet GSIS but did enhance GSIS in normal, functional rat islets (25) and restored GSIS in human islets from T2D donors (27). We hypothesized that FP-101 treatment of dysfunctional islets from diabetic mice or humans would restore GSIS islets through neutralizing endogenous FSTL3 which would presumably enhance activin bioactivity and restore glucose-responsive insulin secretion.

Islets were isolated from chow-fed or high-fat diet (HFD)-fed mice, with 1 week of HFD having been shown to be sufficient to suppress islet function (37, 38). Chow-fed mouse islets responded normally to a GSIS challenge with a significant stimulation index of 2.4 while islets isolated from mice after 1 week of HFD had reduced insulin secretion in the high glucose challenge (Fig. 5A) that was not statistically different from low glucose, that is, no longer functional. Neither FP-101 nor activin B had a significant effect on GSIS in functional islets from chow-fed mice although insulin secretion in low-dose FP-101-treated islets was reduced (Fig. 5B). Treatment with FP-101, activin B, or exendin 4 all restored some functionality to the dysfunctional HFD islets. Specifically, high-dose FP-101 and activin B insulin secretion was statistically greater than untreated HFD islets, although not to the same level as control (chow) islets. These results suggest that FP-101, presumably acting through release of bioactive activin, enhanced glucose-responsive insulin secretion to islets that were rendered dysfunctional by HFD treatment. Since HFD treatment has been utilized to model T2D in rodents (37, 38), these results support the hypothesis that FP-101 treatment could enhance beta cell function in vivo in models of T2D.

FP-101 Rescues GSIS in Nonfunctional Human Islets

Human islet samples from donors with documented T2D are relatively rare and the variability between donors and islet batches is substantial, which creates a challenging experimental environment. Hyperglycemia has been proposed as a leading cause of beta cell dysfunction leading to T2D (39-42). To model glucotoxic reduction in beta cell function, we treated normal, functional human islets with elevated glucose (28.5mM glucose) for 24 hours. As shown in Fig. 6A, this treatment was sufficient to make normally functional islets with a stimulation index of 2.5 (P < 0.01) lose their response to elevated glucose, that is, become nonfunctional while absolute insulin secretion was also reduced.

Treatment of normal (no hyperglycemic treatment) islets for 24 hours with either dose of FP-101 had little effect on overall insulin secretion or responsiveness to glucose (Fig. 6B). Although the effects of activin A or B on these islets were similar to FP-101, the lower dose of activin A appeared to inhibit response to elevated glucose. In contrast, 24 hours of hyperglycemic treatment of the same batch of islets eliminated glucose responsiveness in these islets. Both doses of FP-101 treatment restored insulin secretion and glucose responsiveness, although only the higher FP-101 dose was statistically significant (P < 0.05). In addition, the higher dose of both activin A and B restored normal insulin secretion and statistically significant glucose responsiveness (P < 0.01 and P < 0.05, respectively) while the lower activin A dose trended in the same direction.

Taken together, these results demonstrate that FP-101 treatment restored insulin secretion and glucose responsiveness to nonfunctional human islets with effects similar to higher doses of activin A or B, consistent with FP-101 presumably working through neutralization of endogenously produced FSTL3 and activin A or B (27). These results also demonstrate that FSTL3 regulation of activin and/or related ligand bioactivity is functional in human as well as mouse islets and further, that neutralization of that FSTL3 has beneficial effects that enhance both insulin secretion and glucose responsiveness to nonfunctional mouse and human islets.

Discussion

Individuals with T2D have a limited ability to compensate for insufficient insulin production or sensitivity to control blood glucose (18, 39). Moreover, abnormally elevated glucose is seen as either a primary or a continuing insult that further hampers beta cell function and survival (40). Therefore, sustainable replacement of functional beta cell mass remains the primary focus for curing both type 1 and type 2 diabetes. The long-term effort to delineate transcriptional control of beta cell development and differentiation (43) has led to establishment of protocols to create functional and durable human beta cells from stem cells. These cells could be used to restore normal glucose-responsive insulin production if they could be protected from autoimmune (T1D) or immune mismatch (T1D, T2D) rejection after transplantation, a remaining challenge for this solution (6, 7, 44). Enticing existing human beta cells to replicate is another potential solution and several pathways have been identified that work impressively in mice and transplanted human islets (5). The results presented here provide another potential solution, namely that neutralization of the activin antagonist FSTL3 results in increased activin activity that can enhance glucose-responsive insulin secretion in mouse and human islets in vitro. This pathway is based on the observation that deletion of Fstl3 in mice resulted in expanded islet size that contained more beta cells resulting in improved glucose homeostasis (13, 14). Coupled with the observation that alpha to beta cell transdifferentiation is enhanced in FSTL3 KO mice (17), an FSTL3-neutralizing therapy could have 2 potential benefits for treating diabetes: an acute benefit of restoring beta cell function and longer-term induction of new beta cell formation. Together, this therapy has the potential to restore natural insulin regulation of blood glucose.

Since FSTL3 is structurally and functionally closely related to the activin antagonist follistatin (32), we developed an antibody, FP-101, that could neutralize FSTL3 while avoiding follistatin, thereby reducing the probability of off-target effects in vivo. Moreover, this antibody released activin from otherwise nearly irreversible FSTL3-activin complexes, suggesting that in vivo potency could be enhanced by increasing activin bioavailability via preventing or reversing formation of activin-FSTL3 complexes. FP-101 can also release closely related TGFβ family ligands that are regulated by FSTL3, including myostatin and GDF11 in vitro, although we did not observe obvious effects of excessive myostatin or GDF11 activity in FSTL3 KO mice, such as reduced muscle mass (13). It is possible that the other ligands primarily bind follistatin which was not affected by Fstl3 deletion or FP-101 treatment. Nevertheless, it remains possible that some of the beneficial effects of FP-101 could be due to increased myostatin or GDF11 activity, in addition to activin A and B, in islets or other tissues.

While our observation that FP-101 and activin B treatment of dysfunctional mouse islets enhanced glucose responsiveness is consistent with improved beta cell function in the FSTL3 KO mouse (13), as well as activin and FSTL3 biosynthesis in alpha and beta cells, respectively (25), our finding that this is also true of human islets suggests that the activin-FSTL3 pathway is functional in humans as well. This is consistent with our previous demonstration comparing human islets from T2D and normal donors, in which activin A was the most highly expressed TGFβ family ligand in normal functional islets but was vastly reduced in nonfunctional T2D islets (27). Furthermore, FSTL3 was expressed in human beta cells and was elevated in T2D islets suggesting that reduced activin signaling might be related to reduced beta cell function (27). Support for this concept was obtained by treating T2D islets with activin A, which restored GSIS in these islets (27). In the present study, FSTL3 neutralization through FP-101 treatment of hyperglycemia-induced dysfunctional human islets restored beta cell function similarly to activin A treatment. This is consistent with FSTL3 being an important regulator of beta cell function in humans and that neutralization of FSTL3 releases activin and/or related ligands to restore GSIS in nonfunctional human islets. Coupled with the ability to induce alpha to beta cell transdifferentiation observed in mice (17), which may also be functional in humans (9, 45, 46), FP-101 treatment has the potential to improve insulin production in the short term and to replace lost beta cells in the longer term.

While detailed epitope mapping is a continuing line of research, our results demonstrating that FP-101 can disrupt preexisting FSTL3-activin complexes that are otherwise nearly irreversible suggests that binding of FP-101 to FSTL3 induces a conformational change that releases 1 or more FSTL3 molecules from a complex consisting of an activin dimer surrounded by 2 FSTL3 molecules (47). Moreover, our finding that FP-101 binds rhesus FSTL3 nearly as well as the human molecule, while binding mouse FSTL3 at least 10-fold less, suggests that at least one of the epitopes utilized by FP-101 is in a region where the mouse sequence differs from human and rhesus, located outside the activin binding pocket of FSTL3, since the pocket sequence is highly conserved across species. The fact that identical doses of FP-101 had similar effects on mouse and human islets in vitro despite FP-101 having substantially lower affinity for mouse FSTL3 suggests that the uM concentrations of FP-101 used were in excess of the FSTL3-activin complexes produced by either mouse or human islets so that effective FSTL3 concentrations might be achievable in mice. On the other hand, the reduced affinity of FP-101 for mouse FSTL3 relative to human or rhesus suggests that mouse models of diabetes could be less than ideal for investigating efficacy and will at the least require higher concentrations of antibody while a smaller rhesus study with FP-101 could be more successful.

Limitations

The data presented herein show only that FSTL3 neutralization results in increased activin activity and enhanced glucose-responsive insulin secretion but do not yet identify the mechanism or prove beta cell specificity; therefore, caution should be taken when interpreting the results. These results should be considered observational at this time. Given the complicated signaling mechanisms involved, including autocrine, paracrine, and endocrine action of the TGFβ family of proteins, further experiments will be required to fully characterize the mechanistic actions of FP-101. It is not yet known whether the FP-101 antibody acts directly on the beta cell or indirectly through another cell type. Experiments on single cell-type populations, (such as alpha cells and beta cells), activin receptor inhibitors, and in vivo large animal studies will be needed to provide clarity as to how FP-101 acts. Toxicology data will also be needed prior to clinical use in humans.

The utility of an antibody designed to neutralize FSTL3 must be determined within the broader scope of the many physiologic roles that the TGFβ family serve beyond the impact on the beta cell. For instance, members of the TGFβ family serve as key regulators of adipose and muscle tissue, the reproductive system, the immune system and cancer. While a full discussion of the impact of FSTL3 neutralization on other activin-sensitive cell types is outside of the scope of this manuscript, 2 recent reviews help to shed light on potential effects of deregulation of TGFβ signaling (48, 49). A more in-depth look is provided on the impact on other organs, tissues, and cells as well as how TGFβ signaling and activin A can alter the maturity of the beta cell (50), which may occur as part of a compensatory mechanism. These potential effects need to be investigated in the future.

The primary focus of this study was to develop the FSTL3-neutralizing antibodies, identify the most effective antibody through a series of experiments investigating activin activity, determine specificity for FSTL3 and neutralization of related TGFβ superfamily members, as well as ability to disrupt preformed FSTL3/activin complexes. Further, GSIS assays were utilized to obtain verification that the selected antibody would have an impact on dysfunctional islets as a measure of translational potential and justification for pursuing further development. In conclusion, the data presented here support continued development of the FSTL3 neutralization strategy through in vivo testing of FP-101 and more in-depth mechanistic studies to address an underlying cause of diabetes.

Acknowledgments

In Memoriam: This manuscript is dedicated to Dr. Alan Schneyer, CEO, Co-Founder and Co-Owner of Fairbanks Pharmaceuticals Inc. We regret to inform the readership of Endocrinology of the unexpected passing of our original corresponding author, Dr. Alan Schneyer. Dr. Schneyer committed his life’s work to science, first in reproductive endocrinology, and then later, toward the development of a cure for diabetes. Dr. Schneyer was an active member of the Endocrine Society and frequent speaker at Endo Conferences and expert sessions. We remember Dr. Schneyer as the indomitable optimist, an individual who could see opportunities, as well as the hurdles. Dr. Schneyer represented the best of science and the effort to use science to improve the lives of people. The scientific world in general is mourning the great loss of Dr. Schneyer as one of the driving forces behind the discoveries involving the role of the TGFβ superfamily of protein in regulation of pancreatic islet cells.

The authors would also like to thank Drs. Laura Alonso (Weill Cornell Medicine, NY), Daniel Bernard (McGill University, Quebec) and Joe Jerry (University of Massachusetts—Amherst) for critical review of the manuscript.

Financial Support: This research was supported by the National Institutes of Health/National Institute of Diabetes and Digestive and Kidney Diseases, Small Business Innovation Research and Small Business Technology Transfer, Phase II grant R44 DK107018-03A1 (MLB; originally awarded to A. Schneyer) and National Institute of General Medical Sciences grant R35GM134923 (TT).

Glossary

Abbreviations

ELISA

enzyme-linked immunosorbent assay

FSTL3

follistatin like-3

GSIS

glucose-stimulated insulin secretion

HFD

high-fat diet

hFST

human follistatin

hFSTL3

human FSTL3

KO

knockout

mFSTL3

mouse FSTL3

rhFSTL3

rhesus FSTL3

RLU

relative light units

T1D

type 1 diabetes

T2D

type 2 diabetes

TGFβ

transforming growth factor β

Additional Information

Disclosures: A.L., N.M., and A.R. are employees of Fairbanks Pharmaceuticals Inc and have been granted options. M.L.B. has been granted options in Fairbanks. T.T. has no financial conflicts to disclose.

Data Availability

Some or all datasets generated during and/or analyzed during the current study are not publicly available but datasets and critical resources are available from the corresponding author on reasonable request and MTA execution.

Prior Presentation: Portions of this study were presented in an abstract at the 2020 Endocrine Society Annual Meeting (Schneyer A, Brown ML, Meyer N, et al. OR14-01 FSTL3 Neutralizing Antibodies Restore Function to Diabetic Mouse and Human Islets: A New Approach for Treating Diabetes. J Endoc Society. 2020:4(Suppl_1):OR14–01. https://doi.org/10.1210/jendso/bvaa046.1862).

References

  • 1. Butler AE, Janson J, Bonner-Weir S, Ritzel R, Rizza RA, Butler PC. Beta-cell deficit and increased beta-cell apoptosis in humans with type 2 diabetes. Diabetes 2003;52:102-110. [DOI] [PubMed] [Google Scholar]
  • 2. Costes S, Langen R, Gurlo T, Matveyenko AV, Butler PC. β-Cell failure in type 2 diabetes: a case of asking too much of too few? Diabetes. 2013;62(2):327-335. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3. Meier JJ, Bhushan A, Butler AE, Rizza RA, Butler PC. Sustained beta cell apoptosis in patients with long-standing type 1 diabetes: indirect evidence for islet regeneration? Diabetologia. 2005;48(11):2221-2228. [DOI] [PubMed] [Google Scholar]
  • 4. Cohrs CM, Panzer JK, Drotar DM, et al. Dysfunction of persisting β cells is a key feature of early Type 2 diabetes pathogenesis. Cell Rep. 2020;31(1):107469. [DOI] [PubMed] [Google Scholar]
  • 5. Ackeifi C, Wang P, Karakose E, et al. GLP-1 receptor agonists synergize with DYRK1A inhibitors to potentiate functional human beta cell regeneration. Sci Transl Med. 2020;12:12-530. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6. Pagliuca FW, Millman JR, Gürtler M, et al. Generation of functional human pancreatic β cells in vitro. Cell. 2014;159(2):428-439. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7. Helman A, Melton DA. A stem cell approach to cure Type 1 diabetes. Cold Spring Harb Perspect Biol 2021:13(1):a035741. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8. Liu J, Banerjee A, Herring CA, et al. Neurog3-independent methylation is the earliest detectable mark distinguishing pancreatic progenitor identity. Dev Cell 2019;48:49-63. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9. Furuyama K, Chera S, van Gurp L, et al. Diabetes relief in mice by glucose-sensing insulin-secreting human alpha-cells. Nature 2019;567:43-48. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10. Osipovich AB, Stancill JS, Cartailler JP, Dudek KD, Magnuson MA. Excitotoxicity and overnutrition additively impair metabolic function and identity of pancreatic β-cells. Diabetes 2020;69(7):1476-1491. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11. Xia Y, Schneyer AL. The biology of activin: recent advances in structure, regulation and function. J Endocrinol. 2009;202(1):1-12. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12. Schneyer A, Sidis Y, Xia Y, et al. Differential actions of follistatin and follistatin-like 3. Mol Cell Endocrinol. 2004;225(1-2):25-28. [DOI] [PubMed] [Google Scholar]
  • 13. Mukherjee A, Sidis Y, Mahan A, et al. FSTL3 deletion reveals roles for TGF-beta family ligands in glucose and fat homeostasis in adults. Proc Natl Acad Sci U S A. 2007;104(4):1348-1353. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14. Brown ML, Bonomi L, Ungerleider N, et al. Follistatin and follistatin like-3 differentially regulate adiposity and glucose homeostasis. Obesity (Silver Spring) 2011;19:1940-1949. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15. Collombat P, Xu X, Ravassard P, et al. The ectopic expression of Pax4 in the mouse pancreas converts progenitor cells into alpha and subsequently beta cells. Cell 2009;138:449-462. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16. Thorel F, Nepote V, Avril I, et al. Conversion of adult pancreatic alpha-cells to beta-cells after extreme beta-cell loss. Nature 2010;464:1149-1154. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17. Brown ML, Andrzejewski D, Burnside A, Schneyer AL. Activin enhances alpha- to beta-cell transdifferentiation as a source for beta-cells in male FSTL3 knockout mice. Endocrinology 2016;157:1043-1054. [DOI] [PubMed] [Google Scholar]
  • 18. Chen C, Cohrs CM, Stertmann J, Bozsak R, Speier S. Human beta cell mass and function in diabetes: Recent advances in knowledge and technologies to understand disease pathogenesis. Mol Metab. 2017;6(9):943-957. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19. Sidis Y, Mukherjee A, Keutmann H, Delbaere A, Sadatsuki M, Schneyer A. Biological activity of follistatin isoforms and follistatin-like-3 is dependent on differential cell surface binding and specificity for activin, myostatin, and bone morphogenetic proteins. Endocrinology. 2006;147(7):3586-3597. [DOI] [PubMed] [Google Scholar]
  • 20.RRID:CVCL_ZD63. [Google Scholar]
  • 21. Cash JN, Angerman EB, Kirby RJ, et al. Development of a small-molecule screening method for inhibitors of cellular response to myostatin and activin A. J Biomol Screen. 2013;18(7):837-844. [DOI] [PubMed] [Google Scholar]
  • 22.RRID:AB_2106709. [Google Scholar]
  • 23.RRID:AB_2340770. [Google Scholar]
  • 24.RRID:MGI:5657312. [Google Scholar]
  • 25. Brown ML, Kimura F, Bonomi LM, Ungerleider NA, Schneyer AL. Differential synthesis and action of TGFß superfamily ligands in mouse and rat islets. Islets. 2011;3(6):367-375. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.RRID:AB_2783837. [Google Scholar]
  • 27. Brown ML, Ungerleider N, Bonomi L, Andrzejewski D, Burnside A, Schneyer A. Effects of activin A on survival, function and gene expression of pancreatic islets from non-diabetic and diabetic human donors. Islets. 2014;6(5-6):e1017226. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.RRID:AB_2783838. [Google Scholar]
  • 29. Cash JN, Angerman EB, Keutmann HT, Thompson TB. Characterization of follistatin-type domains and their contribution to myostatin and activin A antagonism. Mol Endocrinol. 2012;26(7):1167-1178. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30. Stamler R, Keutmann HT, Sidis Y, Kattamuri C, Schneyer A, Thompson TB. The structure of FSTL3-+activin A complex. J Biol Chem. 2008;283(47):32831-32838. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31. Thompson TB, Lerch TF, Cook RW, Woodruff TK, Jardetzky TS. The structure of the follistatin:activin complex reveals antagonism of both type I and type II receptor binding. Dev Cell. 2005;9(4):535-543. [DOI] [PubMed] [Google Scholar]
  • 32. Tortoriello DV, Sidis Y, Holtzman DA, Holmes WE, Schneyer AL. Human follistatin-related protein: a structural homologue of follistatin with nuclear localization. Endocrinology. 2001;142(8):3426-3434. [DOI] [PubMed] [Google Scholar]
  • 33. Schneyer AL, Sidis Y, Gulati A, Sun JL, Keutmann H, Krasney PA. Differential antagonism of activin, myostatin and growth and differentiation factor 11 by wild-type and mutant follistatin. Endocrinology. 2008;149(9):4589-4595. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34. Schneyer A, Schoen A, Quigg A, Sidis Y. Differential binding and neutralization of activins A and B by follistatin and follistatin like-3 (FSTL-3/FSRP/FLRG). Endocrinology. 2003;144(5):1671-1674. [DOI] [PubMed] [Google Scholar]
  • 35. Schneyer AL, Rzucidlo DA, Sluss PM, Crowley WF Jr. Characterization of unique binding kinetics of follistatin and activin or inhibin in serum. Endocrinology. 1994;135(2):667-674. [DOI] [PubMed] [Google Scholar]
  • 36. Schneyer AL, Wang Q, Sidis Y, Sluss PM. Differential distribution of follistatin isoforms: application of a new FS315-specific immunoassay. J Clin Endocrinol Metab. 2004;89(10):5067-5075. [DOI] [PubMed] [Google Scholar]
  • 37. Heydemann A. An overview of murine high fat diet as a model for Type 2 diabetes mellitus. J Diabetes Res. 2016;2016:2902351. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38. Stamateris RE, Sharma RB, Hollern DA, Alonso LC. Adaptive β-cell proliferation increases early in high-fat feeding in mice, concurrent with metabolic changes, with induction of islet cyclin D2 expression. Am J Physiol Endocrinol Metab. 2013;305(1): E149-E159. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39. Ebrahimi AG, Hollister-Lock J, Sullivan BA, Tsuchida R, Bonner-Weir S, Weir GC. Beta cell identity changes with mild hyperglycemia: Implications for function, growth, and vulnerability. Mol Metab. 2020;35:100959. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40. Weir GC. Glucolipotoxicity, β-cells, and diabetes: the emperor has no clothes. Diabetes. 2020;69(3):273-278. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41. Ottosson-Laakso E, Krus U, Storm P, et al. Glucose-induced changes in gene expression in human pancreatic islets: causes or consequences of chronic hyperglycemia. Diabetes. 2017;66(12):3013-3028. [DOI] [PubMed] [Google Scholar]
  • 42. Eizirik DL, Korbutt GS, Hellerström C. Prolonged exposure of human pancreatic islets to high glucose concentrations in vitro impairs the beta-cell function. J Clin Invest. 1992;90(4):1263-1268. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43. Oliver-Krasinski JM, Stoffers DA. On the origin of the {beta} cell. Genes Dev. 2008;22:1998-2021. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44. Veres A, Faust AL, Bushnell HL, et al. Charting cellular identity during human in vitro β-cell differentiation. Nature. 2019;569(7756):368-373. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45. Mezza T, Muscogiuri G, Sorice GP, et al. Insulin resistance alters islet morphology in nondiabetic humans. Diabetes. 2014;63(3):994-1007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46. Katahira T, Miyatsuka T, Miura M, et al. Conversion of pancreatic α cells into insulin-producing cells modulated by β-cell insufficiency and supplemental insulin administration. Biochem Biophys Res Commun. 2020;521(1):178-183. [DOI] [PubMed] [Google Scholar]
  • 47. Stamler R, Keutmann HT, Sidis Y, Kattamuri C, Schneyer A, Thompson TB. The structure of FSTL3.activin A complex. Differential binding of N-terminal domains influences follistatin-type antagonist specificity. J Biol Chem. 2008;283(47):32831-32838. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48. Brown ML, Schneyer A. A decade later: revisiting the role of the TGFβ family’s role in diabetes. TEM 2021;32(1):36-47. [DOI] [PubMed] [Google Scholar]
  • 49. Lee J-H, Lee J-H, Rane S. TGF-β signaling in pancreatic islet β cell development and function. Endocrinology. 2020;162(3):bqaa233. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50. Szabat M, Johnson JD, Piret JM. Reciprocal modulation of adult beta cell maturity by activin A and follistatin. Diabetologia. 2010;53(8):1680-1689. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

Some or all datasets generated during and/or analyzed during the current study are not publicly available but datasets and critical resources are available from the corresponding author on reasonable request and MTA execution.

Prior Presentation: Portions of this study were presented in an abstract at the 2020 Endocrine Society Annual Meeting (Schneyer A, Brown ML, Meyer N, et al. OR14-01 FSTL3 Neutralizing Antibodies Restore Function to Diabetic Mouse and Human Islets: A New Approach for Treating Diabetes. J Endoc Society. 2020:4(Suppl_1):OR14–01. https://doi.org/10.1210/jendso/bvaa046.1862).

Articles from Endocrinology are provided here courtesy of The Endocrine Society

RESOURCES