This is another very abstracted study using isolated mouse islets in cell culture to assess the effect of metformin on insulin secretion.
Metformin Inhibits Mouse Islet Insulin Secretion and Alters Intracellular Calcium in a Concentration-Dependent and Duration-Dependent Manner near the Circulating Range
From the Protons perspective the factors which drive insulin signalling are the same ones which drive insulin secretion, certainly at low physiological concentrations. The situation is different under post prandial conditions where, eventually, reverse electron transport increases from low, physiological activating levels of ROS to the high physiological levels which drive insulin resistance rather than activation. Recall this is what I consider to be the cellular repletion signal, the one so easily mistaken for insulin as an anorexic agent. Anyway, here we have metformin acting under 11mmol/l of glucose to suppress insulin secretion.
Just to recap; 20micromolar metformin is therapeutic, 200micromolar is life threatening lactic acidosis and 1mmolar (1000micromolar) is death:
Any findings in the paper using concentrations of 200micromolar or higher can safely be ignored for therapeutic relevance. Except for the confirmation that cells die rather well at 1.0mmol of metformin and are doing rather more apoptosis than you might like at 200micromolar (see Figure 3 in the paper). No surprises there.
Also consider that picking up subtleties of insulin secretion by measuring the concentration in a culture well is a very blunt instrument. But at least they are looking.
So why doesn't metformin cause diabetes on the sort of criminal (up until very recently) diet advised by any diabetologist?
This is partly because the redox changes in the liver suppress gluconeogenesis, though the exact mechanism by which blockade of mtG3Pdh suppresses hepatic glucose production is debatable.
It's also because, certainly in peripheral cells suffering from chronic hyperinsulinaemia-induced lipotoxicity, cessation or reduction of insulin signalling will allow release of fatty acids able to generate their own RET via the oxidation of beta oxidation derived electron transporting flavoprotein at mtETFdh of the electron transport chain and so restore insulin signalling. Or, if there is enough superoxide, insulin resistance. So cells suddenly realise they have a great supply of FFAs, adequate ATP generation and no need for any more caloric ingress. Which generalises to the whole organism as a "no need to eat" state, which might just give weight loss. As metformin does.
Kind of like LC eating in a pill.
Actually metformin might do the same to lipid in the pancreas as it does in peripheral tissues. Loss of accumulated pancreatic lipid is what people like Dr Roy Taylor consider the mechanism by which the hypoinsulinaemia of semi-starvation induces some degree of remission of diabetes, in a few patients, while they can stick to it.