Human Beta Cell Functional Adaptation and Dysfunction in Insulin Resistance and Its Reversibility Open Access

Beta cells play a key role in homeostasis by sensing the concentration of energy-rich molecules, hormones, and neurotransmitters, and secreting insulin in response. Insulin induces the uptake and storage of nutrients in key target tissues when they are appropriately sensitive to insulin. During starvation, insulin level drops and energy reserves are utilized [1, 2]. Maintaining a functional beta cell phenotype with adequate secretion of insulin is crucial for growth, development, and long-term homeostasis of nutrients. Reduced energy consumption and increased energy intake lead to a positive energy balance, triglyceride accumulation in key target tissues, insulin resistance, and metabolic stress in beta cells, followed by a loss of responsiveness to secretagogues at the functional level and dedifferentiation and apoptosis at the morphological level, with consequent hyperglycemia and type 2 diabetes (T2D) [3]. It was long thought that the loss of beta cell function and mass during pathogenesis of diabetes was progressive and irreversible, but in recent years there is increasing evidence that beta cells respond to insulin resistance and direct metabolic stress imposed on them with a number of adaptive mechanisms that increase their sensitivity to glucose, insulin secretion, and can maintain normoglycemia for more than a decade before onset of hyperglycemia [4]. Furthermore, early after the onset of T2D, negative energy balance and maintenance of normal body weight can normalize insulin sensitivity, eliminate metabolic stress on beta cells, and restore their function and structure, at least transiently [5]. In this paper, we briefly review the normal function of human beta cells, their adaptation and dysfunction under conditions of metabolic stress, as well as their reversibility. Within this conceptual framework, the etiopathogenesis of T2D can be regarded as a potentially reversible adaptive response of beta cells to long-term positive energy balance. We do not focus on changes in beta cell mass, as this is beyond the current scope and it seems that beta cell dysfunction may occur before or without any obvious decreases in beta cell mass. In other words, it is currently believed that beta cell dysfunction precedes decreased beta cell mass during pathogenesis of T2D [6, 7]. However, once beta cell mass starts decreasing, this may lead to increases in plasma glucose levels, which exacerbate beta cell dysfunction through glucotoxicity [8]. Additionally, decreased beta cell numbers are typically accompanied by microarchitectural changes, which may importantly disturb normal homologous and heterologous interactions within islets [9, 10]. Finally, both glucotoxicity and microarchitectural changes may disturb intercellular coupling and lead to less synchronous activity, with prolonged activation and more local waves and thus less active cells during the plateau phase of insulin secretion [11, 12]. Whether these changes observed at the islet level are directly related to disrupted first phase of insulin secretion and disappearance of plasma insulin oscillations in patients with T2D needs to be determined [7]. Due to existing gaps in our knowledge on human beta cell stimulus-secretion and intercellular coupling, we briefly point out some similarities and differences between human and mouse beta cells where it is appropriate to highlight both the translational relevance of studies in mice and the need for more studies on human beta cells [2, 13].

Insulin secretion from human beta cells is oscillatory, with a period of approximately 5 min, and results from coupling between stimulation by secretagogues and activation of the secretory machinery [14]. In substimulatory conditions, human beta cells are hyperpolarized due to resting K⁺ conductance to approximately −70 mV. An increase in glucose concentration above 5 mm causes a concentration-dependent depolarization, and when the membrane potential exceeds −60 mV, a typical pattern of electrical activity is triggered [15]. It consists of a transient first phase characterized by continuous spiking activity, followed by a sustained rhythmical pattern that drives pulses of insulin secretion. These are known as bursts of action potentials and are composed of transient membrane potential depolarizations with superimposed high-frequency potential oscillations. This pattern of electrical activity is driven by changes in conductivity to ions. First, glucose uptake and metabolism increase the intracellular ATP concentration, which decreases the conductance of ATP-dependent K⁺ channels, causing depolarization of the plasma membrane. This then activates voltage-dependent Ca²⁺ channels, promoting the influx of Ca²⁺ ions, and an increase in intracellular Ca²⁺ concentration ([Ca²⁺]_IC) activates the secretory machinery and triggers insulin secretion from beta cells (shown in Fig. 1a). Importantly, in human beta cells voltage-dependent Na⁺ channels are also activated upon depolarization [15]. Voltage-dependent Na⁺ channels and voltage-dependent Ca²⁺ channels are responsible for the action potential upstroke, while the repolarization phase occurs due to opening of calcium-dependent K⁺ channels. In humans, bursts are shorter, and individual action potentials are triggered from a more negative membrane potential and reach higher values than in mice. Faster kinetics of voltage-dependent Na⁺ channels is the main reason for the shorter action potentials observed in human beta cells. When glucose concentration exceeds 11 mm, continuous spiking is observed [15‒17]. How the electrical activity is altered in T2D is an open question, since to the best of our knowledge, no one has so far shown the bursting activity of human diabetic beta cells employing intracellular recordings. However, extracellular slow potential recordings, with slow potentials corresponding with bursts of electrical activity, suggest that in human islets cultured at moderately increased glucose (8 mm) to mimic the diabetogenic environment, the electrical activity is enhanced at all glucose concentrations (shown in Fig. 1b), whereas in islets cultured at high glucose (20 mm) it shows a left-shifted dose-response curve at low concentrations and a right-shifted response at higher glucose concentrations [11, 18].

Changes in [Ca²⁺]_IC closely follow the oscillatory electrical activity [2]. The pattern of Ca²⁺ oscillations in human islets has been less well studied than in mice. Although the degree of gap junctional coupling in humans and mice is comparable (see below), the differences in the pattern of Ca²⁺ oscillations are most likely due to the different architecture of human islets, their heterogeneity, and the relatively small number of studies available [13]. Some studies observed globally synchronized Ca²⁺ oscillations similar to the ones in mouse islets [19], whereas others described only locally synchronized oscillations or did not quantify them in detail [2]. Recently, we have reported that Ca²⁺ oscillations in human islets are synchronized across different regions of islets via Ca²⁺ waves, but with a velocity that is an order of magnitude slower than that of fast Ca²⁺ oscillations (or electrical bursts) in mice, but comparable to slow Ca²⁺ oscillations in mice, supporting the hypothesis that they are driven by oscillations in metabolism [20]. Furthermore, Ca²⁺ waves in human islets are much more heterogeneous in size, frequently encompassing only smaller regions, and not spreading across the entire islet. They are however glucose-dependent since higher glucose concentrations increase the activity and cause Ca²⁺ waves to spread more globally across islets [12]. Because changes in [Ca²⁺]_IC drive pulses of insulin exocytosis, it is particularly important to understand how Ca²⁺ oscillations change in T2D [21]. Some studies have reported less coordinated activity in T2D, while others have described the occurrence of regular Ca²⁺ oscillations, but with a decrease in the amplitude of insulin secretion [7]. Furthermore, a reduction in active time has been demonstrated owing to the lower frequency of Ca²⁺ oscillations. While islets from patients with T2D retain their glucose dependence, Ca²⁺ waves in some islets are more locally restricted, and their dysfunction is widely heterogeneous [12]. Importantly, the glucose dependence of Ca²⁺ activity is left-shifted for islets cultured at intermediate concentrations across the whole spectrum of stimulatory glucose concentrations, whereas in islets cultured at high glucose, the Ca²⁺ activity is shifted to the left at lower, but to the right at higher glucose concentrations [18], and may be right-shifted throughout the stimulatory range in islets from donors with T2D [20].

In human islets, insulin secretion is triggered at significantly lower glucose concentrations than in mouse islets, which correlates with lower blood glucose levels in humans and reflects intrinsic differences in beta cell physiology between species [22]. In both mice and humans, insulin secretion occurs in two phases. The first is high-amplitude and transient, the second is low-amplitude, sustained, and oscillatory, with the pulsatile release of insulin from the pancreas being responsible for the fluctuations in plasma insulin with a period of approximately 5 min, which is in the range of the slow membrane potential and Ca²⁺ oscillations [14]. Oscillatory insulin secretion from islets does not depend on external or intrapancreatic neural stimulation but is their intrinsic property. Insulin secreted in pulses is more effective at inhibiting glucose production in the liver than continuous insulin secretion, probably due to potentiated insulin receptor signaling. Furthermore, pulses of insulin secretion with larger amplitudes are more extensively extracted by the liver than those with smaller amplitudes, resulting in relatively smaller insulin levels in the systemic circulation [4]. Insulin secretion has been reported to change in line with changes in electrical and Ca²⁺ activity in human islets cultured at high glucose [18]. Interestingly, insulin secretion has been reported to be enhanced across a wide range of stimulatory glucose concentrations in islets isolated from obese donors (who are insulin resistant but glucose tolerant) [23]. Moreover, insulin secretion seems to be increased at basal glucose concentration in islets in tissue slices obtained from human donors with impaired glucose tolerance, together with a loss in first-phase insulin secretion, whereas in islets from donors with overt T2D, first-phase insulin secretion remains absent, together with a decrease in both basal and stimulated insulin secretion (Fig. 1b, c) [7]. Similar changes have been described in vivo in patients with T2D where the first phase of insulin secretion is lost. In advanced T2D, the second phase of insulin secretion is also impaired, but the biphasic profile of insulin secretion can potentially be reversed by acute negative energy balance [24].

The phenomenon of intercellular depolarization and calcium waves and the mechanisms enabling them are much less well described for human than mouse islets. Specifically, some reports have suggested that either due to electrophysiological or microarchitectural differences spreading of intercellular waves may be much more limited in human islets [25]. However, it has been demonstrated that connexin 36 (Cx36, shown in Fig. 1a), the main coupling protein in mouse beta cells, is also expressed in human beta cells and couples them by gap junctions [26] that Cx36 expression and intercellular coupling decrease under diabetogenic conditions [26‒28], and that this is accompanied by disturbed synchronization between beta cells [29, 30]. Moreover, recent morphological studies have shown that the structure of human islets may not be too different from mouse islets and that at least in many of human islets, most of the beta cells may form a continuous coupled syncytium [13, 31, 32]. Therefore, at present, it seems reasonable to suggest that human islets shall allow for intercellular depolarization waves and diffusion of glycolytic intermediates between cells [2]. Indeed, recent studies suggest that in human islets, gap junctional coupling mediated by Cx36 and possibly additional mechanisms supports intercellular depolarization waves with periods of a few seconds, similarly to mouse islets, which may be crucial for normal biphasic insulin secretion, and synchronization of slow metabolic, membrane potential, and Ca²⁺ oscillations that are most likely paramount for pulsatile insulin release [11, 12].

In islets from donors with T2D and in human islets exposed to diabetogenic conditions, intercellular coupling has universally been reported to be decreased (shown in Fig. 1c), either due to lowered Cx36 expression [29] or due to channel phosphorylation, with a vanishing first phase of secretion and a less well pronounced pulsatility during the second phase [11, 20, 27]. Finally, intercellular coupling may be a parameter of beta cell function that decreases throughout the natural history of beta cell adaptation and dysfunction without a transient compensatory increase during adaptation, but data for this are scant even for mouse islets and non-existing for human islets [33].

T2D has long been regarded as universally progressing and irreversible. However, recent studies have suggested that beta cell dysfunction in insulin resistance may be reversible. These studies have explored the effect of weight loss by caloric restriction or bariatric surgery, physical exercise, and various medications on beta cell function in insulin-resistant individuals [34]. More specifically, pioneering clinical studies employing caloric restriction have shown that remission of human T2D is achievable in a considerable proportion of patients but requires substantial weight loss with a decrease in liver and pancreas fat content [24]. However, a durable remission seems to most critically depend on the capacity of beta cells to recover from dysfunction [5]. Beta cells have previously demonstrated their ability to recover from a dysfunctional phenotype when rested in various experimental settings in mouse models [35]. To our knowledge, there are currently no systematic studies on changes in stimulus-secretion coupling in human beta cells during the transition from adaptation to dysfunction and remission. However, studies on human islets from normal donors cultured under different conditions support the notion that at least the glucose sensitivity of the Ca²⁺ dose-response curve reverses from a right-shifted one back to normal upon culture in normal glucose concentration [18]. In addition, findings in both mouse and human islets suggest that a large part of the progression from an initially left-shifted to an eventually right-shifted response might be due to an initial increase in the activity of glucokinase and later beta cell exhaustion due to overstimulation [4, 18]. Moreover, functional connectivity that is decreased in islets exposed to a diabetogenic environment in vivo or in vitro seems to be able to recover upon caloric restriction in mice [35] and at least in vitro and at least partially upon culture under normal conditions or by use of inhibitors of cytokine-mediated signaling pathways also in human islets [11, 19, 27].

In this context, beta cell rest, either due to fasting or pharmacological intervention directly lowering the activity of beta cells by increasing K_ATP channel activity or indirectly by establishing normoglycemia, seems to hold promise to correct the deranged beta cell stimulus-secretion and intercellular coupling (shown in Fig. 1c), but the capacity for durable beta cell recovery may depend on the degree of existing functional injury [4, 36]. Given that fasting hyperinsulinemia and loss of pulsatility due to oversensitivity in beta cell stimulus-secretion coupling can sometimes be detected a decade or more before onset of T2D, that they are strong predictors of future T2D, and that they may even contribute to insulin resistance, screening for fasting insulin levels in the population may be a viable approach to earlier detect and treat beta cells that have only just started the process of adaptation to increased demand and are not yet dysfunctional [37, 38].

Insulin resistance is an early metabolic alteration in chronic kidney disease (CKD) patients, and it is directly correlated with the loss of glomerular filtration rate (GFR) [39]. During progression of CKD, beta cells follow a path similar to the one observed in T2D. More specifically, in early stages of CKD (stages 1–3, GFR >30 mL/min), beta cell adaptation and a left-shift in dose-dependence appropriately compensate for the loss in insulin sensitivity, resulting in a state of hyperinsulinemic normoglycemia [40, 41]. At ∼30 mL/min GFR, a partial decompensation is typically observed in the form of glucose intolerance despite hyperinsulinemia [42]. At GFR <30 mL/min (CKD stages 4–5), beta cell failure occurs, mediated by the CKD-related (i) loss of filtration, resulting in urea-mediated oxidative stress on beta cells [43], and (ii) loss of renal hormonal function exemplified by vitamin D-dependent decreases in insulin secretion [44]. Recent evidence further suggests that patients in the waiting list for transplantation typically have different degrees of beta cell dysfunction that dysfunctional cells may be especially susceptible to dedifferentiating effects of immunosuppressive therapy, but also that posttransplant T2D can be reversed upon withdrawal of immunosupressants [45]. More specifically, steroids, mammalian target of rapamycin inhibitors, and calcineurin inhibitors have all been associated with an increased risk of posttransplant T2D [45]. Besides their indirect effect on beta cells via inducing insulin resistance, steroids may have direct effects on beta cells. The suppression of insulin secretion is believed to be mediated via alpha-adrenergic signaling, impaired oxidative metabolism in beta cells, activation of repolarizing K⁺ channels, generation of reactive oxygen species, endoplasmic reticulum stress, and decreased sensitivity of the exocytotic apparatus to Ca²⁺, to name only a few possible mechanisms [46, 47]. Further, the mammalian target of rapamycin inhibitor rapamycin may also indirectly increase the beta cell load via insulin resistance, and cause direct reductions in beta cell size, mass, proliferation, and insulin secretion [46]. Finally, the calcineurin inhibitor tacrolimus is associated with an especially high risk of posttransplant T2D and may also act indirectly via increasing insulin resistance, but also have important direct effects in beta cells [45]. Beta cells that are in the phase of compensation or early decompensation seem to be especially vulnerable since tacrolimus has been shown to negatively impact the same intracellular pathways that are responsible for maintenance of beta cell identity and functionality during beta cell compensation and decompensation [48]. Taken together, the above supports the hypothesis that both CKD and immunosuppression can accelerate the pathogenic process leading from beta cell adaptation to dysfunction and that earlier recognition of initial beta cell dysfunction and pharmacological or lifestyle intervention could be advantageous for all patients with insulin resistance, but especially important for patients with CKD.

Similarly to what is observed in mouse models, human beta cells adapt to and compensate for insulin resistance with left-shifted stimulus-secretion coupling. This intracellular signaling cascade, together with intercellular coupling, becomes disrupted during progression toward decompensation and hyperglycemia. During the natural history of CKD, beta cell function similarly evolves from a compensated to a decompensated state. Even after the onset of overt hyperglycemia, beta cell dysfunction seems to be reversible and earlier detection of emerging dysfunction and its reversal shall become primary therapeutic targets in the future. In the present review, we focused on data from studies on human islets and we wish to refer the interested reader to another complementary review on aspects of intra- and intercellular coupling in beta cell adaptation and dysfunction during development of diabetes that focuses on mouse studies [33]. Comparing both papers shall enable one to better understand the functional similarities and differences between mouse and human islets, appreciate the value and translational relevance of mouse studies, and to better plan the important validation studies on inherently limited human samples.

This minireview is based on the CME DIABESITY presentation in Maribor, Slovenia, 16–17 September 2022.

The authors have no conflicts of interest to declare.

We acknowledge support from the Slovenian Research Agency (Programs I0-0029 and P3-0396, as well as projects J3-9289, N3-0170, J3-3077, N3-0133).

Andraž Stožer contributed substantially to the conception and design of the manuscript. Maša Skelin Klemen, Jan Kopecky, Jurij Dolenšek, and Andraž Stožer authors drafted, revised, and approved the final version of the manuscript. Maša Skelin Klemen, Jurij Dolenške, and Andraž Stožer drafted the figure, and Jurij Dolenšek and Maša Skelin Klemen designed the final version.