Welcome back to Article 6 in the 42 Factors Series.
First and foremost, congratulations on making it this far. You have serious tenacity, and the time you’re investing in reading these articles will improve the quality of your work with omic and other data…or at least stimulate your brain. Either way, kudos to you.
Today I’ll shift gears slightly and discuss a factor that doesn’t directly affect quantitative rigor or reproducibility, per se, but it has an enormous impact on biological data interpretation.
Factor – Fasted or Fed Initial State (t=0)
Because time course experiments allow us to elucidate cause-effect relationships, they are an ideal strategy for addressing many biological research questions. For example: 1) Which metabolic reactions modulate a biological process (e.g., cancer progression)? 2) What mechanisms mediate intrinsic or acquired resistance to a drug? 3) What is the mechanism of action of a drug? 4) Which biomarkers identify the early stages of a disease?
In setting up a time course experiment, it is customary to start the clock upon treatment. Even if you’re not studying a drug, the treatment should consist of fresh cell culture medium to provide a basis for examining the ensuing metabolism. But there’s a critical, often overlooked detail of great relevance here—should the t=0 sample model the fasted or fed state? That decision can greatly influence the interpretation of experimental results, as shown in the following example:
The difference in the two sets of results is clear. When the t=0 sample models the fed state, cells appear to consume this metabolite normally, and the drug inhibits its metabolism. When the t=0 sample models the fasted state, this metabolite appears to be present at low levels normally, and the drug positively modulates its synthesis. The conclusions are very different and, hence, depend heavily upon how the t=0 data point was modeled.
So which conclusion is correct? It wouldn’t be correct to say that the fasted state is incorrect, but my preference is strongly in favor of modeling the fed state. In the early days of our metabolomics work, we conducted many experiments with t=0 in the fasted state, and the results generally were more difficult to interpret than results acquired with t=0 representing the fed state. Hence, we recommend the fed state for t=0. For the compulsive investigator, value can sometimes be gained by also collecting a fasted state t=0 sample. Alternatively, if one conducts a time course through 48 hours, the experiment will include data (the period between 24 – 48 hours after feeding; see below) to sufficiently model the fasted state.
What is the recommended protocol for modeling the two states? As you’ll recall from my previous article about trypsin, we generally prefer to seed cells on Monday morning and initiate treatments on Wednesday morning. Wednesday morning is t=0, two days after “seeding” the experiment. Based on our experience in measuring nutrient metabolites (e.g., amino acids) over multiple cycles of cell feeding, we find that cells approach the starvation state approximately 24 hours after feeding. Technically, then, one might say that starvation begins on Tuesday with our schedule, and by Wednesday (48 hours after feeding) the cells are in a late stage of starvation. This is a reason that proper cell culture maintenance involves refeeding cells every 48 hours.
How to model the fed state at t=0 is a slightly trickier question. We’ve treated cells with fresh medium (not containing drug) for as little as a few seconds up to one hour, and anywhere in that range seems to work well. One of my team members feels that the longer (one hour) pre-feeding time produces better data when studying energy metabolism (glycolysis, TCA cycle, pentose phosphate pathway, etc.), but the jury is still deciding on that. At the other end of the spectrum, it’s remarkable that a few seconds of fresh food can put cells into the fed state, but the data don’t lie—we consistently see immediate elevation of many intracellular metabolites after exposing starved cells to fresh medium for just a few seconds.
What about preclinical and clinical studies? Do the lessons from nonclinical studies presented here extend to the preclinical and clinical settings? The short answer is that we don’t know yet. There is a rich literature on the topic that deserves careful review, and I think the topic of fasting vs. fed state measurements in patients deserves an entire article of its own, so I’ll get back to that in a future article.
Take-home messages: 1) time course experiments represent an ideal strategy for elucidating cause-effect relationships in the study of metabolism; 2) the choice of fed vs. fasted state for the t=0 sample can greatly influence biological interpretation of results; 3) to model the fasted state at t=0, essentially do nothing prior to harvesting the sample; 4) to model the fed state (our preferred choice) at t=0, treat cells with fresh medium for at least a few seconds (5 s is plenty) up to one hour if desired.
Thanks to present and former members of my team, especially Lin Tan, Marc Warmoes, Wai Kin Chan, and Leona Martin for their efforts toward studying this factor.
Do you have experience modeling fed vs. fasted states in biological systems? Or any pressing metabolomics questions? Leave a comment.
0 comments