Insulin Resistance (17) ROS NOX2 RET and alpha-tocopherol in cell culture

I'd like to just state, as I start this post, that I have absolutely no doubt that adding a modest dose of α-tocopherol to a lard based obesogenic high fat diet is protective against the associated fatty liver and liver damage. The mechanism is not clear to me as yet, but while pondering it a perfectly reasonable explanation for the problems caused by higher doses of α-tocopherol became apparent. That's what this post is about.

I want to discuss the supplementary figure S2:

taken from

α‐Tocopherol suppresses hepatic steatosis by increasing CPT‐1 expression in a mouse model of diet‐induced nonalcoholic fatty liver disease

ALT is a routine indicator of liver damage, more specifically, of hepatocellular injury.

I'm interested in how progressively increasing levels of dietary α-tocopherol produce worsening liver damage on an high linoleic acid background. I've added in, from Fig 1 panel C, the measured plasma levels of vitamin E involved for the two oral dose rates which we are given, plus I've removed those parts of the chart which are not relevant to the discussion:

From other parts of the paper the group realised that CPT1, the main mitochondrial fatty acid uptake protein, is down regulated in α-tocopherol liver damage and that down regulating CPT1 function using an inhibitor restores the toxicity eliminated by α-tocopherol added at 50mg/kg to the high fat diet. I agree that the role of CPT1 down regulation might be important to the hepatotoxicity of high dose α-tocopherol.

So they went to a cell culture model to see if α-tocopherol reduced CPT1 activity in HepG2 cells.

This is their bar chart:

and this is the line I wish to discuss:

People may have noticed that I have ignored the value for 1μM of added α-tocopherol so here's an aside to attempt to justify this fudge. I have several reasons for this. Primarily it doesn't fit my hypothesis, you have been warned. Added to this is that it has a higher standard deviation than all other bars on the chart, especially those containing α-tocopherol. It's very difficult to interpret this because the methods do not specify if the cell cultures were replicated, if repeated aliquots of tissue protein from the cell culture were analysed and averaged or whether Western Blots were repeated and the data presented are the averages of several densiometry measurements. We could tell a fairy tale which looks like this, with tightly grouped results in four of the bars but with one outlier at 1μM in culture:

the elimination of which would allow us to redraw the bar chart to look more like this:

I have to emphasise that this is a COMPLETELY hypothetical situation. The only problem is that it makes sense once we think about mechanisms.

So let's think.

It takes thirty seconds on any search engine to ascertain that insulin controls the transcription and expression of the gene for CPT1, negatively.

The more we facilitate insulin signalling, the lower CPT1 will be. These are the terms in which we need to be looking at the cell culture results.

The cells were cultured, for 24h, in "10% fetal calf serum/Roswell park memorial institute (FCS/RPMI)‐1640 (Sigma)" which not only provides 25mM of glucose and everything else a cell might want, it also provides significant amounts of insulin and IGF1.

They were then transferred to 1% FCS/RPMI‐1640, which also routinely provides 25mM glucose but with no insulin, IGF1 or any appreciable fatty acids. These cells are quiescent and give a stable platform for assessing CPT1 production without the complication of rapid cell growth over 24h.

Anyone who has followed this blog for any number of years will be very aware that glucose, acting via an NADPH oxidase, in this case NOX2, is an insulin mimetic in its own right. A jump from 5mmol/l to 10mmol/l in humans with pharmacologically fixed insulin levels will demonstrate this insulin-like signal.

This is the paper and this is my doodle drawn on somebody else's doodle:

Equally, we can go to this paper and see that glucose at 30mmol/l produces a seriously damaging level of ROS and insulin *resistance*. Here's my doodle:

The full discussion is in this blog post.

In cell culture at 25mM glucose I would posit that there is a serious ROS signal being produced in the complete absence of insulin which looks, from an HepG2 cell's perspective, a lot like supra-peak insulin signalling. And it's stable. The cells are not like mice, they don't do meals. There is a constant ROS signal equivalent to supra-physiological insulin. This "insulin-like" ROS signal, by imitating insulin, sets our CPT1 baseline level in this column of the bar chart:

Our next move is to look what happens when we add 1μM (a small amount) of oleate to that base line ROS signal. Oleate dose nothing to ROS production via NOX2 but does generate its own ROS signal via RET, as per Protons. The increase is enough to go from insulin resistance levels of ROS to even more insulin resisting ROS, more insulin resistance and this allows more CPT1 production, as in the blue oval:

Bear in mind that there is no insulin. At all. These extra ROS will suppress insulin-like signalling but negative feedback limiting insulin signalling is unable to reduce the ROS because the ROS are not, in this case, from insulin.

But ultimately we can reduce the ROS signal by adding α-tocopherol.

Adding 0.1μM of α-tocopherol is just too little to do anything, we can imagine the blue oval moved one column to the right.

The addition of 1μM of α-tocopherol *should* reduce the ROS signal and allow sightly better insulin signalling which might reinstate a little of the suppression of CPT1 by lowering the ROS signal back closer toward to peak insulin-mimesis. In my imagination like this:

As more and more α-tocopherol is added the levels of ROS are stabilised at lower and lower levels until, somewhere between 25μM and 50μM of α-tocopherol, the added ROS from the oleic acid are neutralised and suppression of ROS now equates the control levels of insulin resisting ROS seen in the control cells.

But these control cells are still looking at 25mM of glucose and, at the high levels of ROS this will be generating, there is still more scope to improve the insulin-like signal towards more effective CPT1 suppression by reducing ROS further out of insulin resistance and toward insulin signalling levels. This happens with 20-50μM of α-tocopherol.

That's what I think is happening.

It's cell culture. It's as close to steady state as we can imagine. It almost certainly tells us something interesting about high levels of ROS at or above peak insulin levels and how manipulating these levels downwards has some effect on insulin signalling downstream of ROS generation.

Just before I take a break we can look at this in terms that I've viewed ROS mediated insulin mimesis before.

That's using the concept from

Evidence for Electron Transfer Reactions Involved in the Cu2+-dependent Thiol Activation of Fat Cell Glucose Utilization

and this well worn graph:

Looking at intact organisms I modified this to show an (arbitrary) upper limit to insulin derived ROS set by the negative feedback cells use to resist the action of real, actual insulin. It looked like this:

But in cell culture with glucose fixed at 25M there is no negative feedback and, though insulin signalling does still become obtunded, ROS generation can exceed that mediated by peak insulin and we end up with a graph very much like the one in Czech's seminal paper where ROS were added from a bottle of hydrogen peroxide:

In the current case we are achieving supra-peak ROS generation using NOX2 and 25mM glucose and then even more by adding to this level of ROS that from oleic acid a 1μM like this:

The in vitro data points are capable of reducing insulin-like signalling by raising ROS. We can then add in the likely ROS reductions mediated by α-tocopherol and see the progressive improvement in insulin signalling.

Which, of course, suppresses CPT1 protein production.

Just as a final "dig" at α-tocopherol 1μM, the two places where it could have acted to markedly reduce insulin signalling, via either high or low ROS, to facilitate CPT1 suppression are these:

I don't buy it.

To summarise: In cells generating stable insulin resisting levels of ROS α-tocopherol moves the ROS signal closer towards imitating an effective insulin-like signal. We can put the same information directly on to Czech's graph like this:

Of course, insulin stores fat.

If you store fat in liver cells you get fatty liver. Does this happen in vivo? When plasma levels are 163μmol/l, rather than the 50μM used in cell culture? Under dynamic insulin signalling conditions?

This not as simple as it seems. A live mouse is not a cell in culture with 4.0mM of added hydrogen peroxide out of a bottle or an added ROS signal from metabolic manipulation via culture medium, both of which can be reduced by a simple antioxidant. In-vivo there are limits to ROS generation and it's the movement of these limits which matters.

Peter