Distinct Roles of β-Cell Mass and Function During Type 1 Diabetes Onset and Remission

Successful therapy at the onset of type 1 diabetes (T1D) not only requires an effective block of the pathological autoimmune process, but also the restoration of adequate insulin levels. Most promising for optimal glucose control is endogenous β-cell activity, which even at low levels reduces the risk for complications and hypoglycemic events (1). Therefore, preserved β-cell function and mass at diagnosis and their potential to recover after immune intervention are a crucial aspect of T1D therapy. Initially, β-cell mass was suggested to be almost completely destroyed at T1D onset (2,3), which questioned the justification of immune intervention at this late time point (4). However, more recent data demonstrate preserved β-cell mass and function in patients with newly diagnosed (5–7) and long-standing T1D (8,9). These observations raise the question of whether immune intervention at T1D onset will allow functional and morphological recovery of the residual β-cells. Indications of a potential recovery are the so-called “honeymoon phase” observed after initial insulin treatment (10), as well as the reported detection of β-cell proliferation in patients with T1D (11). However, it is unclear if β-cell mass and function have the capability to recover after immune intervention, and their distinct roles in the remission process have not been shown.

Here we take advantage of a noninvasive in vivo imaging platform (12,13), and the possibility of successful immune intervention in mouse models of T1D (14), to study functional and morphological changes of β-cells and islets during the onset and remission of T1D. Our results demonstrate substantial morphological and functional β-cell plasticity before and after immune intervention. We furthermore show that β-cell mass and function differentially progress during T1D remission, providing potentially important indications for T1D therapy.

All experiments were conducted on 2- to 5-month-old mice of mixed sex and were approved by the Committee on the Ethics of Animal Experiments of the State Directory of Saxony. Rag2^−/−, Rag2^−/−;RIP-HA (15), and NOD.SCID mice were used as recipients. As islet donors served NOD.SCID mice, the F1 generation of RIP-HA mice crossed with mice expressing green fluorescent protein (GFP) under the control of the mouse insulin 1 promoter (MIP) in β-cells (MIP-GFP) (16) (The Jackson Laboratory, Bar Harbor, ME) and RIP-Cre mice (17) crossed with mT/mG mice (18) (The Jackson Laboratory). RIP-Cre;mT/mG mice express membrane-bound tdTomato in all cells except the Cre-expressing β-cells, which switch to the expression of membrane-bound EGFP. For the induction of autoimmune diabetes in RIP-HA mice, we transferred CD4⁺ T cells isolated from TCR-HA mice (19); and for the induction in NOD.SCID mice, we transferred T cells from NOD.BDC2.5 mice (20).

Mouse islet isolation was performed as previously reported (21). Human islets (tebu-bio, Le Perray-en-Yvelines, France) used for transplantation originated from a 23-year-old male donor (BMI 23.8 kg/m², and HbA_1c 5.3%) with no history of diabetes. After shipment, islets rested in CMRL 1066 (Corning cellgro; Mediatech, Herndon, VA) supplemented with 2 g/L human serum albumin, 100 unit/mL penicillin, and 100 µg/mL streptomycin for several hours before transplantation.

For human and mouse islet transplantation into the anterior chamber of the mouse eye, mice were anesthetized by inhalation of 2% isoflurane in 100% oxygen via a face mask. The mouse head was fixed in a head holder, and a 25-gauge needle used to make a small incision into the cornea, close to the corneal limbus. Next, 30–40 islets in PBS were slowly injected into the anterior chamber of the eye using a custom-made beveled glass cannula (outer diameter 0.4 mm, inner diameter 0.32 mm; Hilgenberg GmbH, Malsfeld, Germany). To ensure full engraftment on the iris, we waited at least 4 weeks before conducting the first in vivo imaging of the transplanted islets.

To induce autoimmune diabetes, in vitro preactivated hemagglutinin (HA)-specific T cells (TCR-HA) were injected into immunodeficient mice with rat insulin promoter–controlled HA expression (RIP-HA) (20). For T-cell preactivation, single-cell suspensions from pooled spleen and lymph nodes of TCR-HA mice were stimulated with 10 µg/mL HA107–119 peptide (JPT Peptide Technologies GmbH, Berlin, Germany) in DMEM (Life Technologies, Carlsbad, CA), supplemented with 1 mmol/L sodium pyruvate, 1 mmol/L HEPES, 2 mmol/L GlutaMAX, 100 units/mL penicillin, 100 µg/mL streptomycin, 0.1 mg/mL gentamicin, 0.1 mmol/L nonessential amino acids, 50 μmol/L 2-mercaptoethanol, and 10% FBS. After 5 days, 8 × 10⁶ total cells/PBS (≥75% CD4⁺TCR-HA⁺) were injected intravenously into RIP-HA mice, and nonfasting blood glucose levels of recipients were monitored. Diabetes onset and normoglycemia were defined as nonfasting blood glucose levels of >400 and <200 mg/dL, respectively. The arrest of autoimmune diabetes was achieved by intravenous injections of 10 µg/day anti-CD3 monoclonal antibody (mAb) (Clone 145–2C11; eBioscience, San Diego, CA) for 5 consecutive days starting at diabetes onset in RIP-HA⁺ mice or at day 4 after adoptive transfer in RIP-HA⁻ mice. In case anti-CD3 mAb treatment did not lead to the restoration of normoglycemia within 2 weeks, mice were additionally implanted with two insulin pellets (LinBit; LinShin Canada Inc., Toronto, Ontario, Canada) by subcutaneous injection under short-term isoflurane inhalation anesthesia.

For in vivo imaging, mice were intubated (BioLite; Braintree Scientific, Inc., Braintree, MA) and anesthetized by 2% isoflurane in 100% oxygen with a 270-µL stroke volume at 250 strokes/min for ≤90 min. In vivo imaging was performed as previously published (12,13). To limit pupil dilation and iris movement, a drop of 0.4% pilocarpine (Pilomann; Bausch & Lomb, Rochester, NY) in PBS was placed on the cornea shortly before imaging. For confocal and two-photon imaging, an upright laser-scanning microscope (LSM780 NLO; Zeiss, Jena, Germany) with a two-photon laser (Chameleon Vision II; Coherent, Inc., Santa Clara, CA) and W Plan-Apochromat ×20/1.0 differential interference contrast, and M27 75-mm objective (Zeiss) was used. Repetitive imaging of intraocular islets was performed at indicated time points. Backscattered laser light was detected at 633 nm. Total islet volume was calculated using surface rendering (Imaris version 7.4; Bitplane AG, Zurich, Switzerland). The backscatter index (BI) was quantified as a ratio of islet and iris mean backscatter intensity assessed from maximum intensity projections (Fiji software) (22). For human islets, the BI was calculated from single image planes. GFP was excited by a two-photon laser at 900 nm and was detected at 500–550 nm. GFP islet volume was calculated from median filtered z-stacks using surface rendering (Imaris version 7.4). Vessels were visualized by injecting 0.4 μmol/L Qtracker 655 (Life Technologies) in 100 µL PBS into the tail vein. Qtracker were excited at 900 nm and detected at 635–675 nm. Two-photon laser power did not exceed 25 mW at the level of the specimen. Pixel size in the x- and y-dimension was between 0.461 and 1.186 µm, and z-stacks were acquired with a step size of 1.5 or 2 μm. In RIPCre;mT/mG islets, GFP and Tomato were excited at 930 nm and were detected at 500–550 and 575–610 nm, respectively. Vessels were visualized by injecting 0.8 μmol/L Qtracker 705 (Life Technologies) in 100 µL PBS into the tail vein and detected at 690–730 nm. GFP volume was assessed from Gaussian filtered images as the GFP fraction within the top 80 µm of the islet, and the absolute GFP volume was calculated in relation to the total islet volume obtained from backscatter imaging.

5-Ethynyl-2-deoxyuridine (EdU) labeling was performed by intraperitoneal injections of 5 µg EdU/g body wt (Life Technologies) in PBS daily for 6 consecutive days prior to pancreas extraction. For immunohistochemistry, eyes and pancreas were fixed in 4% paraformaldehyde in PBS for 2 h at 4°C, cryoprotected by sucrose substitution (9%, 18%, and 30% in PBS), embedded in Tissue-Tek OCT Compound (A. Hartenstein GmbH, Würzburg, Germany) and fast frozen in 2-methylbutane in liquid nitrogen. Cryosections (10 µm) were stained with antibodies to insulin (1:200 dilution; Dako, Germany), glucagon (1:200; EMD Millipore, Billerica, MA), CD45 (1:100; BioLegend, San Diego, CA), and DAPI (2.5 µg/mL; Sigma-Aldrich). Immunostaining was visualized by Alexa Fluor 488, 633, 546, or 647 secondary antibodies (1:200 or 1:100; Life Technologies). A TUNEL assay (Roche Diagnostics, Mannheim, Germany) was used to quantify apoptosis, and incorporated EdU was detected (Life Technologies) to assess proliferation. Images were acquired individually in single-track mode for each fluorophore by confocal imaging. Quantification of immunohistochemistry was performed manually using Fiji software. The relative insulin and GFP-positive areas per islet were assessed manually from two to seven sections per intraocular islet with a minimal distance of 30 µm between sections. For the analysis of TUNEL⁺ and EdU⁺ staining, >1,800 nuclei (DAPI⁺ staining) per mouse were evaluated manually.

For intraperitoneal glucose tolerance tests (IPGTTs) mice were fasted for 6 h. Subsequently, mice were injected intraperitoneally with 2 g glucose/kg body wt. Blood glucose was measured at 0, 30, 60, 90, and 120 min after glucose injection using an Accu-Chek Aviva glucometer (Roche, Basel, Switzerland). Blood samples were collected from the tail vein at 0 and 30 min after glucose injection, and plasma was separated by centrifugation at 2,000g at 4°C. Plasma insulin levels were assessed using an Ultra Sensitive Mouse Insulin ELISA Kit (Crystal Chem Inc., Downers Grove, IL).

Data are presented as the mean ± SEM. Results obtained from longitudinal in vivo imaging studies were analyzed by linear mixed models (SPSS version 21; IBM, New York, NY). Repeated measurements of islets were modeled with an autocorrelation structure of order 1, and the host effect was reflected in the model by a random intercept per mouse. All other data were compared by ANOVA or by an unpaired, two-tailed t test (Prism; GraphPad Software, San Diego, CA). Multiple comparisons were adjusted for by Šidák correction when necessary. Significant differences are indicated as follows: *P < 0.05, **P < 0.01, or ***P < 0.005.

To investigate β-cell and islet biology during T1D remission, we used the RIP-HA/TCR-HA adoptive transfer model of autoimmune diabetes (20). Adoptive transfer of HA-specific T cells into RIP-HA mice resulted in massive T-cell infiltration and rapid autoimmune destruction of pancreatic islets (Fig. 1A). This led to the development of overt diabetes within 1 week (5.0 ± 0.8 days; n = 19) (Fig. 1B), followed by sustained severe hyperglycemia (Fig. 1C). Treatment of RIP-HA mice at the onset of hyperglycemia with anti-CD3 mAbs halted autoimmune infiltration in 90% of treated mice, and normoglycemia was restored within 9.7 ± 5.7 days (n = 18) (Fig. 1C). We combined the RIP-HA autoimmune diabetes model with noninvasive in vivo imaging of islet biology by transplanting islets into the anterior chamber of the eye (AC) of mice for longitudinal assessment by laser scanning microscopy (12,23–25). To enable the quantification of islet β-cell volume in autoimmune diabetes, we crossed RIP-HA mice with MIP-GFP mice (16). MIP-GFP mice have been previously used to assess β-cell mass. However, the expression of GFP in the β-cells of these mice has been shown to be heterogeneous and linked to differences in insulin promoter activity (26). Hence, we also used the light-scattering property of islet endocrine cells to measure total islet volume, which includes all cells and the vascular network of the islet. Light scattering of β-cells reflects the size of their granule stores and its intensity provides insight into the secretory activity of β-cells (27,28). Nevertheless, in the case of backscattered laser light, the functional dependence of signal intensities did not influence total islet volume measurements due to the extremely high signal-to-noise ratio of backscattered laser light in the AC and the applied measurement algorithm. Consequently, the simultaneous detection of GFP fluorescence and backscattered laser light enabled us to systematically assess the mass and functional status of β-cells in a longitudinal fashion in vivo.

We verified the feasibility of investigating autoimmune diabetes in the AC platform by transplanting MIP-GFP;RIP-HA⁺ islets into the ACs of immunodeficient recipients. After engraftment, islets displayed normal cellular morphology, a dense vascular network, and a strong backscatter signal (Fig. 2A). By day 3 after adoptive transfer, the vessels of intraocular islets and the adjacent iris showed strong T-cell infiltration (Fig. 2B). The GFP volume of intraocular islets declined rapidly after adoptive transfer (day 3 = 93.9 ± 3.9%, day 5 = 31.9 ± 4.8%, day 7 = 9.5 ± 1.9%, day 9 = 5.7 ± 1.8%, and day 17 = 1.2 ± 0.5%, relative to before adoptive transfer) (Fig. 2C), and the islet backscatter signal disappeared within a short time period (Fig. 2A). Moreover, in response to islet destruction the vascular network was also rearranged (Fig. 2A). Immunohistochemical analysis of intraocular islets revealed massive immune infiltration around dispersed insulin⁺ cells after adoptive transfer (Fig. 2D), confirming the autoimmune destruction of intraocular islets as a result of adoptive transfer.

We next transplanted MIP-GFP;RIP-HA⁺ islets into RIP-HA⁺ recipients. In this setting, adoptive transfer led to the infiltration of intraocular and pancreatic islets generating an autoimmune and hyperglycemic environment (Fig. 3A–E). After adoptive transfer, intraocular MIP-GFP volume significantly decreased (83.9 ± 2.5% on day 3, P < 0.05) (Fig. 3B and D) and islet backscatter intensity started to drop (91.0 ± 14.4% on day 3, P = NS) (Fig. 3C and E), suggesting an advancing destruction and degranulation of β-cells during the developing autoimmune diabetes. The treatment of mice with anti-CD3 mAbs at diabetes onset immediately halted β-cell destruction but did not prevent an almost complete loss of cell backscatter by day 5 after adoptive transfer (Fig. 3C and E). β-Cell backscatter remained low during the subsequent hyperglycemia and began to be restored once the mice had returned to normoglycemia, 1–2 weeks after anti-CD3 treatment (Fig. 3A and C). MIP-GFP volume started to increase during the hyperglycemic phase, reaching a peak of 137.6 ± 7.7% at day 9 after adoptive transfer. Surprisingly, the increase in MIP-GFP volume was transient and decreased after the mice returned to normoglycemia (Fig. 3B and D).

We next sought to identify the role of autoimmune infiltration and hyperglycemia on the observed changes in β-cell mass and functional status. Thereto, we modulated the intraocular islet environment by specific donor-recipient combinations of RIP-HA genotypes (Supplementary Table 1) to create conditions of exclusive autoimmune infiltration or hyperglycemia, respectively (Supplementary Table 1 and Figs. 4 and 5). Autoimmune infiltration alone induced islet destruction and loss of β-cell backscatter rapidly after adoptive transfer (Fig. 4A–E). Anti-CD3 mAb treatment halted the decrease of MIP-GFP volume (76.5 ± 3.8% at day 7 after adoptive transfer), while the backscatter signal of β-cells dropped to a minimum. However, in the absence of hyperglycemia we detected no increase in MIP-GFP volume (Fig. 4B and D) after treatment, and β-cell backscatter recovered quickly after arrest of the autoimmune attack (Fig. 4C and E).

As expected, under exclusive hyperglycemic conditions we observed no destruction of intraocular islets after adoptive transfer (Fig. 5A, B, and D). Nevertheless, at the onset of hyperglycemia islet cells showed decreased backscatter, which rapidly progressed to an almost complete lack of backscatter signal (Fig. 5C and E). MIP-GFP volume increased to 151.3 ± 10.2% during hyperglycemia and returned to prediabetes values in response to decreasing blood glucose levels (Fig. 5B and D). Moreover, islet backscatter recovered after the restoration of normoglycemia (Fig. 5C and E). These data revealed that immune infiltration and hyperglycemia independently induce β-cell degranulation. However, hyperglycemia prolongs the time until the functional recovery of β-cells and represents the critical stimulus for MIP-GFP volume increase after autoimmune intervention.

To determine the underlying mechanisms of changes in MIP-GFP volume, we additionally measured total islet volume independent of GFP expression by backscattered laser light (Fig. 6A). Total islet volume increased after the arrest of the autoimmune attack (145.8 ± 7.7% at day 11 after adoptive transfer), corroborating a morphological β-cell mass increase, as observed by MIP-GFP (Fig. 6A). This was associated with a more than fourfold increase in proliferating β-cells (12.8 ± 1.5 vs. 3.0 ± 0.4%), as revealed by EdU labeling (Fig. 6B), and a significant enlargement of individual β-cell size (96.0 ± 1.5 vs. 76.8 ± 4.0 µm² mean cross-sectional β-cell area) (Fig. 6C).

In contrast to MIP-GFP volume, total islet volume remained elevated after recipient mice returned to normoglycemia (Fig. 6A). In addition, TUNEL staining revealed no increase in the frequency of apoptotic β-cells during MIP-GFP volume decrease in comparison with controls (0.07 ± 0.03 vs. 0.06 ± 0.02%) (Fig. 6D). Immunohistochemistry confirmed a decreased fraction of MIP-GFP–expressing cell area within islets after recovery from hyperglycemia (Fig. 6E). However, the relative insulin area per islet was unchanged (Fig. 6F). Accordingly, the percentage of GFP-expressing insulin⁺ cell area after recovery was decreased in comparison with that of controls (21.9 ± 2.8 vs. 29.4 ± 1.4%) (Fig. 6G). To verify the involvement of reduced insulin promoter activity in diminished MIP-GFP expression, we used mice expressing GFP in β-cells after Cre-mediated recombination (RIPCre-GFP) as donors for islet transplantation. In RIPCre-GFP mice, recombined β-cells expressed GFP under the control of the cytomegalovirus β-actin enhancer-promoter (18), thereby rendering the expression level of GFP less dependent on β-cell function. Comparable to MIP-GFP islets, adoptive transfer–induced hyperglycemia significantly increased RIPCre-GFP islet volume to 131.6 ± 5.5% of the initial volume (Fig. 6H). However, in contrast to MIP-GFP islets, RIPCre-GFP islet volume was found to be significantly elevated after 2 weeks of normoglycemia (118.8 ± 4.9%) (Fig. 6H). Thus, our data show that hyperglycemia after autoimmune arrest induces β-cell mass increase by hypertrophy and hyperplasia, and, while the restoration of normoglycemia leads to reduced insulin promoter activity, β-cell mass remains elevated.

To evaluate the applicability of our findings to autoimmune diabetes in general, we additionally studied the islet response in the NOD mouse model by an adoptive transfer of BDC2.5 T cells to NOD.SCID mice, which were transplanted with intraocular NOD.SCID islets. Our results revealed similar dynamics of islet cell mass and functional status as observed in the RIP-HA model (Supplementary Fig. 1), emphasizing the broad significance of our data.

We next assessed glucose tolerance and insulin secretion during the onset and remission of hyperglycemia to correlate systemic glucose homeostasis with the observed divergent changes in β-cell mass and functional status. As expected, glucose tolerance showed extensive deterioration with the onset of diabetes (Fig. 7A and B). After the remission of hyperglycemia, glucose tolerance remained impaired for an additional 2 weeks before reaching control levels during the third week of normoglycemia (Fig. 7A and B). Thereby, the time course of glucose tolerance recovery correlated with the kinetics of β-cell regranulation (Fig. 7C), suggesting a major role of β-cell functional status on glucose tolerance. This was further supported by the lack of significant glucose-stimulated plasma insulin levels during the first 2 weeks of remission. As β-cell mass does not increase further (Fig. 6A), these data demonstrate that the improvement of glucose homeostasis after the remission of diabetes largely depends on the recovery of the functional status of β-cells.

Following the observation that hyperglycemia is critical for a potential functional and morphological recovery of β-cells after autoimmune intervention, we exposed human islets to the same transient hyperglycemic environment in vivo. We transplanted human islets into the ACs of RIP-HA⁺ mice, and monitored total islet size and backscatter during the onset and recovery from diabetes (Fig. 8). After engraftment, human islets displayed intact morphology by laser backscatter and showed no sign of destruction after adoptive transfer (Fig. 8B and D). Interestingly, and in contrast to mouse islets, human islets displayed a significant increase in islet cell backscatter just before onset of hyperglycemia (136.3 ± 7.8%) (Fig. 8C and D). In response to hyperglycemia, human islet cell backscatter decreased to 62.6 ± 4.6% on day 13 after adoptive transfer and recovered after the host mice returned to normoglycemia. The observed backscatter pattern during hyperglycemia differed notably between human and mouse islets. In islets of both species, individual cells within the islet did not lose backscatter in response to hyperglycemia. In mouse islets, the cells maintaining backscatter signal were in low number, whereas in human islets these cells were more abundant and mostly located around the vessels inside the islet (Fig. 8E). This corresponds to the reported increased frequency of and described the location of α-cells in human islets (29). In response to hyperglycemia, human islets showed only a moderate and delayed increase in total islet volume in comparison with mouse islets, reaching a maximum of 126.4 ± 4.6% after 2 weeks of hyperglycemia (Fig. 8B). Importantly, in contrast with mouse islets, total human islet volume expansion was transient and significantly decreased after the reversal of hyperglycemia (Fig. 8B). Together, these results indicate that human β-cells show similar functional plasticity during the onset of and recovery from hyperglycemia as observed in mice, but exhibit only a minor and transient morphological response.

We provide unprecedented insight into the development and role of β-cell mass and functional status during T1D onset and remission. We demonstrate that intraocular RIP-HA islets reflect the key aspects of the RIP-HA mouse model (30), enabling the study of islets in autoimmune diabetes. The use of this mouse model of autoimmune diabetes facilitated efficient crossing with MIP-GFP mice for the fluorescent detection of β-cells. Similar responses of islet cell mass and functional status were also observed in the NOD.SCID mouse after adoptive transfer of BDC2.5 T cells. This suggests that our findings on the behavior of β-cell mass and functional status in T1D onset and remission reflect a general characteristic of islets in autoimmune diabetes.

In correlation with observations from NOD mouse (31) and human (3) pancreas sections, our data show that β-cell mass is significantly reduced prior to the onset of hyperglycemia. In addition, β-cells display an ongoing loss of scattering property as a sign of an advancing degranulation before T1D onset, indicating the emergence of β-cell exhaustion. Our data illustrate that the degranulation of β-cells is caused by infiltration itself, as has been observed in response to infiltration by islet-specific T-cell clones from NOD mice (32). Additionally, β-cell degranulation is a result of the rising blood glucose levels. These results demonstrate that an autoimmune attack not only destroys β-cells prior to T1D onset but also exacerbates metabolically driven β-cell exhaustion.

Treatment with anti-CD3 mAbs at T1D onset halted the destruction of β-cells but did not prevent further β-cell degranulation. Thus, stopping the autoimmune attack at T1D onset creates a state of hyperglycemia at a reduced mass of functionally exhausted β-cells.

We demonstrate that the subsequent phase of hyperglycemia after immune intervention induces islet mass expansion by hypertrophy and hyperplasia of β-cells, emphasizing the role of glucose as an inducer of β-cell proliferation (33–35). At the same time, the hyperglycemic environment prevents refilling of granular stores and functional recovery. Nevertheless, eventual return to normoglycemia leads to a stop of β-cell mass expansion. Interestingly, mice exhibit impaired glucose tolerance for a prolonged time period after the remission of hyperglycemia and only slowly return to normal glucose homeostasis. As β-cell mass does not further increase during this time period, it indicates a crucial role of functional β-cell recovery in this process. This is supported by the corresponding time course of β-cell regranulation after return to normoglycemia and the simultaneous restoration of glucose-stimulated plasma insulin levels. An additional indication for the functional recovery of β-cells after a return to normoglycemia is the observed decrease in MIP-GFP islet volume without a detectable change in morphological β-cell mass. Our results indicate that MIP-GFP islet volume decreased due to reduced insulin promoter activity, suggesting that partial β-cell rest was a result of T1D remission.

Taken together, our study provides evidence for a distinct role of β-cell mass and function during the remission of T1D after immune intervention. Whereas halting autoimmune destruction enables hyperglycemia-induced β-cell mass expansion up to the remission of hyperglycemia, the subsequent return to normoglycemia facilitates the recovery of β-cell function, leading to improved glucose homeostasis.

Human islets showed similar functional exhaustion and recovery under conditions of transient hyperglycemia. However, human islets exhibit a reduced and transient increase in islet mass in response to hyperglycemia. This is consistent with the low regenerative potential reported for human β-cells under comparable conditions (36,37), and could be related to species or age differences in mouse and human donor islets (38). The observed functional β-cell recovery is likely to be the underlying mechanism of the so-called honeymoon phase of partial or complete transient T1D remission observed in patients after initial therapy (39). On the other hand, the lack of hyperglycemia-induced proliferation of human β-cells might help to explain the difference between mouse and human in reaching euglycemia in response to anti-CD3 mAb treatment at recent T1D onset (40–44). Consequently, T1D remission after immune intervention for the treatment of T1D in humans solely depends on functional recovery of the residual β-cell mass. The remaining β-cell mass at the start of treatment and its ability to functionally recover will be influenced by the duration of the autoimmune attack and the functional intactness of the cells. Therefore, our data provide a mechanistic basis for the improved effect of the anti-CD3 mAb treatment reported for patients who were treated soon after T1D onset (45) and for patients exhibiting an improved metabolic status at the start of treatment (46). Thus, in the absence of a remedy to induce human β-cell proliferation, therapy for human T1D should aim at early immune intervention and β-cell rest.

Acknowledgments. The authors thank Ezio Bonifacio and Michele Solimena, Technische Universität Dresden, for advice and critically reviewing the manuscript. The authors also thank Katharina Hüttner, Chrissy Kühn, Angela Gröbe, and Kerstin Pfützner, Technische Universität Dresden, for excellent technical assistance.

Funding. This work was supported with funds from the Emmy Noether Program of the German Research Foundation (Deutsche Forschungsgemeinschafthttp://dx.doi.org/10.13039/501100001659 [DFG], www.dfg.de), the Center for Regenerative Therapies Dresden (CRTD)—DFG Research Center for Regenerative Therapies Dresden, Cluster of Excellence (www.crt-dresden.de), the DFG-Collaborative Research Center/Transregio 127 and the German Ministry of Education and Research (www.bmbf.de) to the German Centre for Diabetes Research (DZD) and to the Competence Network Diabetes Mellitus (KKnDm).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. H.C. conceived and designed the experiments, performed all in vivo and immunohistochemistry experiments, analyzed the data, and prepared the manuscript. C.M.C., J.A.C., and C.C. analyzed the data and provided intellectual input. C.P. performed the in vitro experiments and provided technical expertise. M.K. and I.R. performed the statistical analysis. K.K. provided intellectual input and technical expertise and prepared the manuscript. S.S. conceived and designed the experiments, supervised the overall project, analyzed the data, and prepared the manuscript. S.S. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.