1. The total protein or the DNA can be used for normalization, the latter is more accurate in representing total number of cells.

2. Yes. The extent of change will depend on the degree and duration of stimulation.

Dear Dr. Alam,

Thanks for your reply.

Another question comes to me. I am not quite clear about how to analyze the data from GSIS, method from various papers differs. Are normalized insulin secretion under lower glucose (3.3mM) and high lower glucose (16.7mM) stimulation enough? Or the their ratio? And for normalization, using DNA better, considering the released insulin may reduce the total protein concentration.

I was wondering whether you could illustrate for me? I am grateful for your help.

Yours sincerely,

Yingyun Gong

To ensure that your results are meaningful, you need a measure that corresponds to a constant unit, for example per fixed number of cells (usually million) in this case. Yes, you need to divide the numbers at low as well as high glucose by cell number.

As for the ratio it is a matter of simple math, irrespective of whether you divide high by low, prior to or after normalizing them for cell numbers, the ratio will be the same.

@Tausif Alam  Thanks for your explanation, and you made it quite clear!

Dear Yingyun Gong, 

I also agree with Tausif Alam. I have recently been working on normalisations for insulin secretion. From my knowledge and what others do in the field many different normalisation parameters have been used previously such as total protein, insulin content, islet number and DNA. It is important to recognise that pancreatic islets display heterogeneity in their ability to secrete insulin, so much so that some islets will not secrete any insulin at all in response to a high glucose concentration. It is also worth note that when normalising to protein or insulin content, any treatment may affect these levels. For these reasons DNA is most probably the better parameter to normalise your data, however in whole islet experiments you must remember that DNA is also recorded from other cells and in any dead cells. Therefore it is possible that your insulin secretion data will be underestimated. To avoid some of these problems people tend to plot the fold change between low and high glucose (GSIS). However, if you have a treatment that affects GSIS you will probably want to find out if the effect is on insulin secretion at low or high glucose and then you will need to normalise your data by one of the other parameters as described before.  

The situation for INS-1 cells is less complicated and I would recommend using something like DAPI as an estimation of cell density. This can be done immediately after your GSIS experiments. 

Q1 - Normalized to total protein or expressed as a percentage of total insulin content.

Q2 - usually not.  only ~1-3% of total insulin is released in islets during GSIS.

INSULIN SECRETION FROM ISOLATED ISLETS:

a. refer to DNA content, and

b. estimate (by immunofluorescence the proportion or number of beta cells. Since hyperglycaemia may inhibit the expresion and content of insulin keep the islets in normoglycaemia overnight  before

c. you may also refer the secretion to the insulin content

INSULIN SECRETION FROM ISOLATED PANCREAS

more difficult to normalize but you could refer to protein concentration, DNA and weight of the organ

"Since hyperglycaemia may inhibit the expresion and content of insulin keep the islets in normoglycaemia overnight..."

I found the latter half of the above to be true. It is fully anticipated due to degranulation of insulin secretory granules caused by prolonged exposure to hyperglycemia. 

Inhibition of insulin expression by hyperglycemia is contrary to my experience. I recall observing an obvious increase in insulin mRNA by in situ hybridization even though insulin contents of beta cells were low. But that is not surprising when hyperglycemia remains uncorrected and hyperglycemia-induced insulin secretion continues to deplete the stored insulin. Under such circumstances, islets try to compensate by increasing insulin synthesis, as evidenced by higher mRNA contents but they are unable to cope with demands. We have published many of these observations over a couple decades ago. 

Dear Jonathan Barlow,

Thanks for your delicate explanation, Your words, "It is important to recognise that pancreatic islets display heterogeneity in their ability to secrete insulin, so much so that some islets will not secrete any insulin at all in response to a high glucose concentration", really remind me of the others' finding that among all the beta-cells in a single islet, the response varies from cell to cell. Some cells are sensitive and  others may be slow. Will this also be possible for islets? And how to make it out? During GSIS processing, 8-10 islets per well. Is it necessary do pre-stimulation to recognize unresponsive islets? 

Best wishes,

Yingyun Gong

Dear Dr. Fueger,

Thanks for your kind help!

Best wishes,

Yingyun Gong

Dear Yingyun Gong,

In islets I believe there are 3 populations, low-secretors, medium secretors and high secretors, in terms of insulin being secreted. I am unsure why this is the case but it is possibly due to differences in insulin triggering from the beta cells, maybe due to changes in glucose-stimulated mitochondrial respiration. If you wanted to explore this then you would need a sensitive assay that you could use to measure either insulin, or respiration from a single islet. I am unaware of any assay that is currently used to test for glucose-responsiveness in islets. People have explored mitochondrial respiration in islets to assess islet health for transplantation purposes so it might be worth reading some of that literature. Being selective in the islets you use for experiments based on their ability to secrete insulin is something I would be cautious about doing. Generally if you do enough repeats and pick your islets at random from multiple pancreata then this is normally accepted. As mentioned previously and in the other answers to your initial question, % of insulin content or DNA are the usual and better methods to use for normalising insulin secretion data. Clearly, there are limitations to both of these methods, however, as the samples for both assays are the same you could use both assays in the first instance and then this will give you a better idea of whether there are any significant differences between each method. Ideally it would be better if their was a way to easily and accurately quantify the amount of beta cells specific in each islet preparation and then use absolute beta cell numbers to normalise the data, although I am unsure if such assay exists. 

Best wishes, 

Jonathan

Dear Jonathan Barlow,

Your explanation just raised my interest further, and I want to discuss a bit more with you. In our lab, after separating  and collecting mice islets and culturing overnight, we will pick out those intact islets with similar size for GSIS. Bigger size means more beta-cells. Can this process be called ramdom?

Yingyun

There are several known traits about beta cells in islets which certainly make them far less uniform in their glucose responsiveness:

1. Islets from head and tail of pancreas are different in their relative cellular composition.

2. At any given point, not all beta cells secrete insulin equally. There seems to be a pulsatile response pattern and in time, roles of low versus high secretors are switched. 

Working with isolated islets introduces disruption to normal patterns. Each beta cell has access to efferent and afferent flow in vivo. Once the blood flow is severed during islet isolation, cells are limited in their responses to what they experience through diffusion. That is the reason why there is always a certain amount of cell death at the center of islets that have been allowed to stay in culture. Longer the culture time, larger the islets, more extensive the cell death at the center (in rodents, these are mostly beta cells). 

Of course, it is easy to discern dead cells in a microscope but in reality, the damage and change in normal secretory behavior commences long before one can visually recognize dead cells.

Some work has been done in the past on single islets and also on sets of beta cells within an islet, you may have to do a few literature searches. This is how the importance of "communication" between beta cells within an islet was discovered.

In order for others to have a good idea as to what your results may mean when you publish them, I would suggest that whatever size islets you use in each experiment, try to keep their size distribution profile consistent in each test sample. If you are using only a few islets per test sample, it is all the more important that you use similar size islets. This can be done fairly quickly by estimating the diameter of an islet using a microscope at low magnification.

Dear Dr. Alam,

Thanks for making it easy and clear. May I consult another related question? I got to know that two-photon microscopy is applied to test the fusion events of the insulin granule vesicles with plasma membrane in islets. Fluorescent dye indicates this morphology changes and software counts the events. I wanna to employ it for observing specific gene effect on the beta-cell secretion capacity. Will two-proton speak louder than GSIS?

Yours sincerely,

Yingyun Gong

At cellular level, every seemingly simple and obvious event is quite complex. Active processes like regulated secretion involves considerable cytoskeletal action. The question is what is it that you wish to study! 

The choice of technique depends on what questions do you wish to address. If you are interested in overall performance of insulin secreting cells at macroscopic level (in bulk), doing a sub cellular level analysis will give you the final numbers by summation of all measurements but it will be an unnecessary overkill.

To illustrate the point, let's assume you goal is to weigh 10gm of NaCl accurately. You can weigh the amount quite accurately in one attempt by adding all the needed crystals or you can weigh one crystal at a time on a microbalance, keep adding them to a vessel until you have reached to the sum of 10gm.

Some times one can lose the forest for the trees!  Chose the best method that will answer your question in the simplest and most reliable way. 

Hi Yingyun Gong,

I like to conduct secretion studies in islets with a normalization to basal secretion. This allows each set of islets to act as its own control. I do each condition in triplicate or better. Islets are selected the morning after isolation, avoiding islets that have fused or show any sign of central necrosis or look otherwise unhealthy. Typically for a static incubation, I will select 10 islets per well, choosing 3 small, 3 large and 4 medium, and picking them randomly into the wells, three islets into each well and repeat, in the opposite direction until all wells have the desired number of islets.  Occasionally you will find a set of islets that "secretes" a lot of insulin at both basal and stimulatory glucose concentrations. This indicates that the islets are not healthy and are releasing insulin in an uncontrolled manner. Having enough replicates allows one to throw out data from unhealthy islets, but still have enough data to feel confident of what you see. Likewise, in perifused islets, the basal secretion should be significantly less than the stimulated secretion.

For adherent cells, we normalize to protein. It is reproducible within a given cell line, and if we find unusually low protein, we repeat the experiment, because we have many years experience with our lines and know what the protein levels should be for a million cells plated out for two days.

Cheers!

Dear Elizabeth Oseid,

Thanks for your practical reply. It will be a good idea to do it again by your method.

Yingyun Gong 

Thanks for your great information! I just start working on rat pancreatic islet and do not know much about this research field. Do you have a working protocol on DNA and/or protein extraction protocol after glucose stimulated insulin secretion experiment? Thanks a lot!

Dear Haijun Gao,

At present, we use insulin content to normalize insulin secretion. The total insulin content is extracted by extraction buffer (10ml: HCl 0.14ml, ethyl alcohol 7.4ml, ddH2O 2.46ml) overnight (more than 16hours is prefer).

What's more, I am also interested in the working protocol on DNA and protein extraction.

Good luck!

Yingyun Gong

Thanks, Yingyun. I will be working on DNA and protein extraction as well. Actually I'm concerned about the sonication condition before extraction. If I get positive results, I would like to share the protocol. Hopefully someone else will give us some good protocol.

Hi Yingyun Gong and all, 

I'm new to this GSIS study too. I'm trying to evaluate the insulin content by itself regardless of insulin release so I need to measure either DNA or protein to normalize it. So my question is: for a batch of 15 islets, what is the approximately concentration of protein so that I can use an appropriate volume to measure by protein assay? 

If the protein/DNA is so diluted, is it reasonable to evaporate the ethanol? 

Thanks, 

Hi guys 

can anyone of you please share the formula of insulin normalization  for GSIS samples either by protein and DNA?

many thanks