Gypped by GIP

Dr. Johnson’s recent paper is nothing short of a monster.  They did a TON of experiments.  Here, I only want to focus on one aspect.

The insulin-obesity hypothesis in a nutshell (very oversimplified): more insulin = more fat mass and vice versa.

I know I know, it was mice fed standard rodent chow, but also included models relevant to human biology like reduced insulin and caloric restriction, which may reflect certain aspects of ketogenic diets and intermittent fasting… and some of the results actually do reflect what happens to humans.  Some.



divide and conquer

They use two lines of mice: one with the insulin-2 gene deleted (Ins1+/+; Ins2-/-); and another that also had one copy of the insulin-1 gene deleted (Ins1+/-; Ins2-/-).  The latter model has half the copies of insulin and for simplicity, I’ll refer to these as knockouts (KOs) and the others as wild-type (WT).  Further, they also studied caloric restriction (CR) in both models.

Ergo, four groups: ad lib WT, ad lib KO, CR WT, and CR KO.


Interestingly, ad lib KOs had the lowest insulin.  We may have expected CR KOs to have the lowest, but they don’t.  And in agreement with the insulin-obesity theory, the CR KOs with more insulin also have more fat mass.  Say what?  The CR KO mice ate less (by design) but had more fat mass (supports insulin-obesity theory) (remember the fattier mice?).




insulin legend

fat mass and NEFA


However, ad lib KOs had lower insulin than ad lib WT, but the same amount of fat mass.  This either refutes the insulin-obesity hypothesis or, going out on a limb here, may suggest their adipose is more sensitive to insulin, so less insulin is required to induce the same amount of fat storage.

The authors hypothesized elevated GIP levels in ad lib KOs (relative to ad lib WT) may be responsible for this adipose tissue insulin sensitization.  However, WT mice have the same level of insulin and fat mass regardless of ad lib or CR, despite CR having WAY more GIP.  If GIP was responsible for adipose tissue insulin sensitization, then the same amount of insulin should’ve induced more fat growth, right?




GIP in a nutshell: secreted from the upper small intestine in response to nutrients, insulinogenic, and fattening to the adipocyte.

In general, if leptin doesn’t at least roughly mirror fat mass, something’s up.




Something’s up.

In this study, CR KOs have the highest leptin and most fat mass, but CR WTs have more leptin than ad lib WTs (despite similar fat mass)… this may reflect an influence of the semi-starved state, so I’ll take it… #context and all that jazz.

Check out Jane Plain’s take on the GIP story here. Highly recommended.  Infinitely quotable and ties in a lot of different #contexts including rebound weight-gain, intermittent fasting, and potential mechanisms mediating the effects.

It’s great, but the reasoning only seems to apply to ad lib WT and KOs. It doesn’t apply to CR and ad lib WTs, who have relatively the same amount of insulin as CR KOs.  I mean, the difference in GIP between ad lib and CR WT is really big. CR WTs have the highest GIP of all and more insulin than ad lib KOs, but absolutely the same level of fat mass.

That’s TWO factors that should be promoting adipose accumulation, but aren’t.

So, my question is: what’s the #context mediating this disparity?


Also, highly recommend reading the paper — even the discussion section!  Especially if you want to learn a lot about CR, GIP, energy balance, etc., in a short time.

For example, caloric restriction in the context of reduced insulin gene dosage… think: intermittent fasting and LCHF?  or simply the spontaneous reduction in appetite observed by many on a ketogenic diet (yes, this is a form of CR).  According to the Johnson-Plain Hypothesis™, there may be some people strongly predisposed to weight re-gain in these contexts.

Add in the findings of Gardner, Pittas, Cornier, and Ebbeling and we may be able to begin to hone in on who these people might be.


calories proper



Be Sociable, Share!
  • IMO its probably not good to consider them WT, as they aren’t WT. They are genetically modified to reduced insulin, so their anomalous results are likely / possibly found within context of this genetic manipulation.

    With that said, i suspect the anomalous findings are likely held within the intricacy of how insulin receptors function on individual tissues. Its like any system; if you alter 1 property, you might *anticipate* result x, but you get result y, which seems contradictory. However, it would be reasonable if we understood the complexity and related factors of the entire system better. Both rodents were insulin modified animals.

    With that said, I think one reasonable conclusion from this study is that insulin expression in tissues is not simply fattening, but also required for normal energy use. Thus, in context of CR, insulin deleted animals are fatter/defend their adipose better. I think this is a reasonable assumption that we would evolve such machinery, as insulin is a marker for fed state, so deleting insulin while also exposing to the condition of CR (actual starvation) can enhance the energy conservation in adipose tissue.

    Can we extrapolate these findings to humans on diets designed to lower insulin? Probably not at all. None of the rodents were WT, they all were genetically modified to be insulin resistant throughout the body in various receptors. WHen CR was introduced to this condition, they preserved fat mass better. To be honest, its actually reasonable when we consider insulin is a necessary hormone marking fed state – that some level of insulin is required for normal energy use, this becomes more marked in actual starvation.

    My hypothesis how this might be executed. A JJ hypothesizes, clearly GIP is involved in the 75% insulin deletion rodents, and it is already known from other research and the physiological effects of GIP that it is specifically obesigenic. However, there are likely multiple neuroendocrine factors at work, probably the brain: POMC, MSH, sympathetic tone to adipocyte, etc… and a critical level of insulin signalling in content of CR is required to avoid enhancing starvation state. I suspect some critical degree of insulin is required to avoid an exaggerated energy conservation state with paradoxical superior fat retention in context of starvation. One study cant find all of these factors which would account for why the 50% rodents might have the same fat mass , even when one was exposed to CR.

    As stated, do i think this offers insight to normal humans on LC diets? Probably not, we are not genetically modified to have 50% or 75% of our insulin receptors deleted after all. Although it might offer insight into the varied responses to CR: some are super effecient conserving energy and retaining fat mass. Perhaps insulin receptor types and concentrations throughout the body might be explaining this difference.

  • Some random thoughts: Regarding leptin, its actually not paradoxical the differences observed. Insulin signalling in adipose is the mechanism by which leptin is made. This is why leptin levels always correlate with active weight gain and maintained adiposity, and declining leptin correlates with weight loss and maintained reduced adiposity.

    So lets think about what is observed. 75% insulin CRs have less insulin genes than 50% CRs. It is expected the 50% CRs would have less leptin than the 75% (KO) type while CR, *if* CR is fattening to insulin deleted rodents through WAT sensitization to insulin (which it seems to be). These 75% KOs have the greatest susceptibility to GIP induced WAT insulin sensitivity -> abnormal fat mass retention and leptin synthesis.

    Regarding the leptin difference of 50% CR vs 50% ad lib: Since CR seems to induce fat tissue growth in insulin deleted rodents, it makes sense that the 50% rodents have more leptin while CR, and less leptin while ad lib. While ad lib, they aren’t growing as much fat, there is less *specific* insulin signalling in the adipose, so the leptin will be , as expected, lower. The leptin is higher in 50% CR rodents for the same reasons their fat mass is equal to ad lib. It appears in order for 50% rodents to maintain the same adiposity as ad lib 50%, they require specific WAT insulin sensitization , likely through GIP. Thus, higher leptin, with same fat mass, in CR 50%s.

    The amount of insulin is only a fraction of story; how that insulin behaves in the body is much more important. This is true in humans as well. Lean body builder vs massively obese very well might both have same insulin levels, body builders even dope with novolog. If you take a cocktail of sympathetic stimulants and hormones that orient insulin to muscle and block it at the fat, the more insulin the leaner and more built you will be.

    What i think is interesting, this study seems to suggest insulin may somehow constrain the fattening effects of GIP. It suggests, to me, GIP is part of energy conservation and insulin might inhibit its effects and synthesis.

    Really interesting study and its too bad we dont see more research like this, unraveling the mechanistic details of how adiposity actually happens. This is how we will solve obesity, not MRIs (itself now in question) of brain response to “rewarding food”, studies which are based on the long debunked assumption that obesity is a simple question of how to attenuate pathological addiction-like overconsumption.

  • Jack Kruse

    With a light mismatch GIP and insulin are uncoupled from circadian coupling. This is the missing link.

  • Dr.Dan

    Monster paper indeed.

    A problem here is the that the “WT” mice did not display reduced fasting insulin levels when subjected to calorie restriction (Figure 1D & E). So they are not a “WT” phenotype.

    • “They use two lines of mice: one with the insulin-2 gene deleted
      (Ins1+/+; Ins2-/-); and another that also had one copy of the insulin-1
      gene deleted (Ins1+/-; Ins2-/-). The latter model has half the copies of insulin and *for simplicity,* I’ll
      refer to these as knockouts (KOs) and the others as wild-type (WT).”

      • Dr.Dan

        Yep I saw that and I get it. The “problem” I referred to was not your “WT/KO” nomenclature (which was actually very helpful) but rather that the “WT” mice did not seem to respond to CR by having reduced fasting insulin (which from the text of the paper seemed to be what we should expect).