Repeatability Is Not the Same as Accuracy

Dworak et al.² report a highly consistent set of data apparently showing that there is a “surge” of ATP in certain brain tissues at sleep onset. That surge correlates with the EEG delta power associated with early sleep and is inhibited by sleep deprivation. They interpret these data to conclude that a decline in cell metabolism during NREM sleep alters the energy balance of the neural tissue so that ATP synthesis exceeds ATP utilization. Although this makes a nice story, it cannot be accepted on the basis of the data reported, nor can it be accepted on the basis of what is known about normal maintenance of cell energy charge. Essential components of a study such as this and its published account should: (1) justify the methodology, and (2) put the data in the context of previous studies. Neither of these elements is in this paper, and it turns out that they are critical.

The methodology used by Dworak at al.² was to decapitate the rat, remove the brain, cut coronal slices, and freeze the slices on dry ice. They report that this procedure took 80 ± 9 sec. A number of studies dating back as far back as the 1950s demonstrate the rapid loss of phosphocreatine and ATP following decapitation. Kratzing and Narayanaswami⁴ obtained small pieces of the cortex of young guinea pigs by sacrifice and rapid dissection or by rapid freezing in liquid air. Values from the rapidly frozen brains were around 3.0 mmoles/kg, whereas those from the brain slices were 0.73 mmoles/kg. The time to inactivation in these slices was not mentioned. Various methods for rapidly inactivating the catabolic enzymes reveal that most of this loss occurs in the first minute following the ischemia due to decapitation. Mandel and Harth⁶ compared ATP values from mouse brains rapidly dissected (20 sec) and frozen with brains obtained from mouse heads that were dropped immediately into liquid nitrogen. The values for ATP were 0.38 mmoles/kg for the dissected brains and 1.4 mmoles/kg for the rapidly frozen brains. Lowry et al.⁵ decapitated mice and dropped the heads into liquid Freon 12 at intervals ranging from immediately to 10 minutes. The highest values of ATP obtained were 2.5 mmoles/kg, and they fell to half that value in about 6 seconds and reached a low, rather constant level by about 1 minute. Ponten et al.⁷ proposed a method to freeze the brain rapidly by pouring liquid nitrogen directly onto the skulls of anesthetized rats that had received implanted thermocouples so that the freeze front could be timed. They also timed the cessation of blood flow. At a depth of only 3 mm, blood flow continued for 5 to 10 seconds, at which time the tissue had reached 20°C, and it took another 5-6 sec for the tissue to freeze. Thus, one would infer that decapitating animals directly into liquid nitrogen would involve 10 to 15 sec of ischemia even at the most superficial layers of the brain. In this study, the concentration of ATP measured in the superior parietal cortex (first to freeze) was about 3 mmoles/kg. So, how much ATP is lost in the 15 sec or more following decapitation and enzyme inactivation in these rapid freezing protocols?

A study by Delaney and Geiger¹ answered the question of what are the levels of ATP in tissues in which the catabolic enzymes are instantaneously inactivated, and how rapidly they fall in the seconds following the onset of ischemia. The innovation of Delaney and Geiger was the use of high-energy focused microwave irradiation to bring brain tissue temperature up to an enzyme inactivating level of 85°C within seconds. They found that a power level of 10 kw heated the brain tissue to this temperature in 1 sec, 6 kw in 2 sec, and 3.5 kw in 3 sec. In addition, they sacrificed rats and extracted tissues by the techniques used by Dworak et al.,² as well as by immersion of the decapitated head in liquid nitrogen or the in situ freezing of the brain tissue with liquid nitrogen. The decapitation into liquid nitrogen yielded values around 0.3 mmoles/kg. The decapitation followed by dissection and then freezing yielded values around 0.06 mmoles/kg; and the brains that were exposed to 10 kw microwave radiation yielded values around 3.5 mmoles/kg. (Since the data reported by DeLaney and Geiger were reported in mg/protein, I translated the data assuming that the protein content of the tissue was 12%.) Clearly, the method of tissue extraction has a huge effect on the measured ATP content because of the rapid degradation of ATP in ischemic tissue. The method used by Dworak et al.² yields the lowest values of all the methods tested; and the fact that this is due to the catabolism of ATP is revealed by the energy charge of tissue obtained in this manner, which was reported as 0.12, whereas the energy charge of the tissue inactivated by the 10 kw microwave radiation was 0.80.¹

We therefore would expect that the ATP levels measured by Dworak et al.² to be in the low range in comparison to the values from the literature cited above. The actual values of ATP cannot be determined from most of the data in this paper because they are presented as % change from baseline. However, actual ATP concentration values are shown in Figures 3 and 5, and they are reported to be in the range of 50 × 10⁻⁸ M/mg tissue (value for the cortex in Figure 3B, C). Converting these values into the same dimensions as the other data discussed above, we get 500 mmoles/kg tissue. Clearly the authors made a calculation error or they labeled their graphs erroneously. If we assume their mistake was 3 orders of magnitude, then their values fall closer to values reported from other experiments using the same protocol, and these values would be far below the levels of ATP existing in the tissue at the time blood flow ceases.

Apart from the actual values of ATP measured by Dworak et al.,² their claims for an ATP “surge” between 200% and 400% of basal rates cannot be true. It is well known that the energy charge of cells that are not deprived of fuel or oxygen is well buffered and in the range of 0.8 to 0.95. That means that 80% to 95% of the adenylates in the cell are in the form of ATP, and there is simply no possibility for increases in ATP of the magnitude suggested. Only if the energy charge of the cells was extremely low could there be an increase of ATP on the order of 200% to 400%. Thus, if the energy charge of the cells were around 0.2, it would be possible to have an ATP increase of 400%. However, healthy cells would never have such a low energy charge. Even insect flight muscle after sustained flight has an energy charge above 0.8.^3,8

How can we explain the apparent “surge” in ATP at the onset of sleep as reported by Dworak et al.² and its inhibition by sleep deprivation? The answer seems quite obvious. In all of the tissue extractions, the loss of ATP was considerable, but presumably the onset of sleep is associated with a decrease in cell ATPases. Thus, there was less degradation in the samples extracted during the sleep phase than during the wake phase of the daily cycle. But, if my corrections of their reported values are correct (a 3 orders of magnitude decrease), then even the highest values reported during the “surges” do not come close to the true ATP concentrations in cells with a normal energy charge.

In summary, by rigorously employing a protocol for extraction and analysis of ATP from brain tissues, the authors have produced a set of consistent, repeatable data. However, since that protocol is not capable of measuring what the authors intended to measure, the data are not accurate. That assessment is reinforced by putting the reported data into the context of what is already well known about brain energy metabolism. In conclusion, this paper does not support its basic claim that there is a large increase in ATP concentrations in various brain areas at the onset of sleep.

DISCLOSURE STATEMENT

Dr. Heller has indicated no financial conflicts of interest.