Whole-cell measurements of insulin secretion

Most data on the secretory function of pancreatic β-cells has been obtained with biochemical measurements of insulin-release from isolated islets or perfused pancreata.

Although Cm measurements are an established technique to study β-cell physiology

4 Discussion

(Kanno et al.,2004) it is occasionally questioned that these measurements are faithfully reporting insulin release.

Mostly it is stated that the insulin release rates assessed by biochemical methods are much lower than the rates obtained from Cmmeasurements. In this study we are reporting a theoretical maximal release rate of 118–133 vesicles s^-1 (400 fF s^-1) (see fig.

3.15, p.57). Albeit being in the range of previously published data for mouse^β-cells (Barg et al.,2001;Göpel et al.,2004) this is more than two orders of magnitude higher than the biochemically measured insulin release rate of a single ^β-cell in isolated rat islets, which is estimated to be∼0.2 granules s^-1 at the peak of the 1^st phase of GSIS (Straub and Sharp,2004).

However, it has to be considered that the theoretical maximum release rate of a responsiveβ-cell decays exponentially and already 40 ms (=_b τ) after stimulus onset this rate is reduced to 37% (¹⁰⁰/e) of its initial value—the depletion of the fast secretory component therefore almost immediately leads to a decrease in the global release rate in response to a constant stimulus. We also used a less artificial stimulation regime by applying repetitive depolarization pulses from a negative holding potential to a Vm

that guarantees maximum Ca²⁺-influx. This train stimulation evoked a total^Σ∆Cmof

∼200 fF (see fig.3.8and3.14, pp.49 and56) after 5 s which corresponds to an average release rate of 12–13 vesicles s^-1 (40 fF s^-1). This rate is still higher than biochemically assessed. However, also a depolarization train with a pulse frequency of 10 Hz can still be considered to be an unphysiological high stimulus: Similar asAntunes et al.(2000) we recorded a spike frequency of 1.5 Hz in aβ-cell at 10 mM glucose (see fig.3.4, p.43) and even during bursting activity the action potential frequency usually did not exceed

∼5 Hz. In addition, one has to take into account that action potentials of rat^β-cells only reach -20 mV (Antunes et al.,2000) resulting in a lower Ca²⁺ influx than in the stimulation protocol used here. Mouse ^β-cells show a 3–4^×higher^∆Cm if depolarized to 0 mV in comparison to -20 mV (Smith et al.,1999;Göpel et al.,2004). It is not straightforward to deduce the hypothetical^∆Cmin response to a stimulation mimicking physiological spiking behavior from this data, but it should approach the rate of insulin measurements while likely being still higher. However, up to now our data only applies for the average responsiveβ-cell and no concession was made for the inherent response heterogeneity. It is well known that the endocrine pancreas is characterized by a high

degree of redundancy. In humans only 50% of the pancreatic^β-cell mass is thought to be responsive to glucose under normal conditions (Clark et al.,2001) and in rats 95%

of the whole pancreas has to be removed to induce hyperglycaemia (Bonner-Weir et al., 1983). Even in isolatedβ-cells the percentage of β-cells releasing insulin in response to 20 mM glucose as estimated with the reverse hemolytic plaque assay is 66.4% (Hiriart and Matteson,1988)—a value strikingly similar to the fraction of responsive^β-cells we report here (see fig.3.8, p.49).

Cmmeasurements in pancreatic^β-cells are criticized in two more points. First, similar to other cells containing secretory granules,^β-cells hold small synaptic like microvesicles (SLMVs) in addition to LDCVs (Reetz et al.,1991;Thomas-Reetz and De Camilli, 1994;Takahashi et al.,1997). These GABA-containing (Reetz et al.,1991) vesicles are capable of undergoing regulated exocytosis (Braun et al.,2004;MacDonald et al.,2005) which could possibly lead to a contamination of the Cmmeasurements. However, it has been demonstrated that the contribution of SLMVs to the total^∆Cmin response to depolarizing stimulation is61% (Braun et al.,2004). Second, Cmmeasurements only detect membrane fusion and therefore the actual release of insulin into the extracellular space cannot be assayed by this technique. For the comparative analysis of the data obtained from healthy and diabetic rats this might indeed lead to a bias, because one study found a higher occurrence of electron-translucent vesicles of LDCV size in islets of GK ratβ-cells (Höög et al.,1997). However, whereas in a more recent study these putatively immature vesicles were also observed in obese fa/fa rats, no ultrastructural difference betweenβ-cells of diabetic GK rats and healthy Wistar rat controls was found (Sondergaard et al.,2003). Taking all this into consideration, it is reasonable to assume that our Cmmeasurements indeed report LDCV secretion—albeit necessarily elicited by a more intense stimulation regime than by nutrient-induced electrical activity (see section3.1.3, p.48).

A further possible objection against our approach regards the use of the standard whole-cell technique. During whole-cell recording, substances are lost from the cell due to the effects of dialysis with the pipette solution (Pusch and Neher,1988). Therefore we performed an experiment to test the influence of whole-cell dialysis on the secretory capacity of the patched cells. Fig. 3.7(p. 47) indeed shows that both∆Cmand Ca²⁺

influx are declining during the recordings. The experiment further shows, however, that

4 Discussion

decrease in^∆Cmis likely attributable to the rundown of Ca²⁺ current and not due to the washout of factors vital for secretion, since the dependency of Cmon Ca²⁺-influx stays within the reported low power relation for depolarization-induced exocytosis (Engisch and Nowycky,1996;Smith and Neher,1997;Mansvelder and Kits,1998;Kits and Mansvelder,2000). Therefore, as long as care is taken to monitor Ca²⁺influx together with ^∆Cm meaningful whole-cell measurements of LDCV secretion can readily be performed in^β-cells in tissue slices. For the comparison of healthy and diabetic rats the use of the controlled whole-cell dialysis is particularly advantageous because it enables us to largely exclude the signaling differences resulting from dissimilar concentrations of diffusible cytosolic factors and focus on the distal steps of Ca²⁺-secretion coupling:

Different resting levels of ATP and cAMP in Wistar and GK ratβ-cells have repeatedly been reported (Hughes et al.,1998;Dachicourt et al.,1997). However, because cytosolic ATP is kept constant and in contrast to single cell preparations cAMP can completely be omitted from the pipette solution this is unlikely to bias our results.

Needless to say that several aspects of comparative physiology of healthy and diabetic β-cells necessarily would have to be studied in metabolically more intact cells. Both long-term monitoring of induced electrical activity and secondary glucose-dependent modulations of Ca²⁺-induced secretion—i.e. the triggering and amplifying pathway—need intact metabolism leading to changes in intracellular metabolites that are unlikely to take place in dialyzed cells. The perforated patch clamp technique would allow these changes to occur and therefore it would be desirable to establish this technique for future experiments studying this subjects.

4.2 Ca

-secretion coupling in healthy vs. diabetic rats

4.2.1 Increased basal electrical activity and higher Ca²⁺ current