Chronic fatigue syndrome and mitochondrial dysfunction

Mitochondrial energy metabolism

In each cell glucose is broken down to pyruvate with the production of some ATP (2 molecules net per molecule of glucose). The pyruvate and also fatty acids enter the mitochondria of each cell, shown schematically in Figure 1, where two coordinated metabolic processes take place: the tricarboxylic acid (TCA) cycle, also known as the Krebs' citric acid cycle, which produces some ATP, and the electron transport chain (ETC, also called the Respiratory Chain because it uses most of the oxygen we breathe in) which regenerates ATP from ADP by the process of oxidative phosphorylation (ox-phos). Altogether some 30-odd molecules of ATP are produced per molecule of glucose and these constitute the main cellular energy packets used for all life processes. As well as food and oxygen the metabolic pathways require all the nutrients involved in the production of the large number of enzymes which control the many biochemical reactions involved and all the cofactors needed to activate the enzymes [31-33]. Most of the enzymes are coded by nuclear DNA (nDNA) in the cell's nucleus and a few are coded by mtDNA. Some of the enzymes rely on other organs. For example, thyroid hormone is needed in the TCA cycle. On the other hand hyperthyroidism can uncouple the ox-phos process [34], so a thyroid problem can lead to fatigue and this can be tested for. The human body contains typically less than 100 g of ATP at any instant, but can consume up to 100 kg per day. Thus the recycling ox-phos process is extremely important and it produces more than 90% of our cellular energy. The main features and processes are illustrated in a simplified form in Figure 1 (further details can be found in all college-level textbooks on biochemistry, e.g. [6], and in secondary school advanced-level biology textbooks, e.g. [35]). The ETC culminates with the protein complex ATP synthase which is effectively a reversible stepping motor in which 3 ATP molecules are produced from ADP and inorganic phosphate (P_i) every revolution [36]. Because of evolutionary history ATP is made inside the mitochondrial inner membrane but used outside in the cytosol where it releases energy by converting to ADP and P_i. The Pi as a negative ion is co-transported back inwards together with H⁺, while ADP³⁻ is transported inwards through the Translocator protein adenosine nucleotide translocase (TL or ANT) in exchange for ATP⁴⁻ moving out into the cytosol. There are potential problems here because it is known that some specific molecules (e.g. atractyloside) block the transfer inwards and certain others can block transfer outwards [37], and there is the possibility that there may be other molecules including environmental contaminants which can block transfers.

What happens if some part of these cellular metabolic pathways goes wrong? If the mitochondrial source of energy is dysfunctional many disease symptoms may appear [38] including the symptoms of CFS.

Suppose that the demand for ATP is higher than the rate at which it can be recycled. This happens to athletes during the 100 meters sprint. The muscle cells go into anaerobic metabolism where each glucose molecule is converted into 2 molecules of lactic acid. This process is very inefficient (5.2% energy production compared to the 100% of complete oxidation) and can last for only a few minutes. The increased acidity leads to muscle pain. Also, when the concentration of ADP in the cytosol increases and the ADP cannot be recycled quickly enough to ATP, another chemical reaction takes place. This becomes important if there is any mitochondrial dysfunction. Two molecules of ADP interact to produce one of ATP and one of AMP (adenosine monophosphate). The AMP cannot be recycled [6] and thus half of the potential ATP is lost. This takes some days to replenish and may account for the post-exertional malaise symptom experienced by patients [39-43].

Thus, mitochondrial dysfunction resulting in impaired ATP production and recycling is a biologically plausible hypothesis, and there is considerable evidence that it is a contributory factor in CFS, at least for a subset of patients. Our study may be considered to be a test of this hypothesis.

Participants

Seventy-one patients, 54 female of average age 47 (range 14 to 75) and 17 male of average age 52 (range 20 to 86), were selected from a total of 116 consecutive patients attending a private medical clinic specializing in CFS/ME. Patients were excluded only if they did not meet the CDC diagnostic criteria for CFS [19] or if the “ATP profile” test had been made before they had been seen clinically. Evaluations, tests and interventions, where appropriate, were carried out for diet and sleep problems, allergies, and thyroid and adrenal problems. Advice on pacing was also given. After this stage a meeting was held with each patient at which an agreed numerical Ability was assigned and recorded in the clinical notes. The integral CFS Ability Scale [44] runs from 0 to 10 and is given in Appendix A. It was proposed to those patients who had not improved to an acceptable clinical level after these interventions that they have the “ATP profile” test done. All the participating patients had scores of 7 or less on the CFS Ability Scale.

The nature of the test was explained and each patient agreed (and paid) for the “ATP profile” test (needing a 3-ml venous blood sample) to be performed. The laboratory carrying out the tests (Biolab Medical Unit, www.biolab.co.uk) was blinded to the Ability associated with any blood sample. As tests were carried out on more patients, it became clear that the “ATP profile” results were providing helpful information, and patients were asked to give written, informed permission for their test data to be used anonymously. All patients have done this.

Blood samples from fifty-three normal, healthy volunteers were obtained by one of us (JMH) as Laboratory Director of Biolab until retirement from that position in 2007. Biolab obtained written permission with informed consent from each volunteer. The samples from the patient group and the normal (control) group were processed in the same way. The control group consisted of 40 females of average age 36 (range 18 to 63) and 13 males of average age 35 (range 18 to 65).

For both groups all procedures were consistent with the Declaration of Helsinki (2000) of the World Medical Association (www.wma.net) and this report follows the guidelines of the International Committee of Medical Journal Editors (icmje.pdf available at www.icmje.org).

Procedures

ATP is present in cells mainly as a complex with magnesium and is hydrolysed to the diphosphate (ADP) as the major energy source for muscle and other tissues. ADP conversion to ATP within mitochondria can be blocked or partially blocked by some environmental contaminants. Specifically, the TL in the mitochondrial membrane that controls the transfer of ADP from the cytosol and ATP to the cytosol may be chemically inhibited and its efficiency is also pH dependent. Changes in acid:base balance, magnesium status, and the presence of abnormal metabolic products can have similar effects to xenobiotic inhibition of the TL.

A number of methods have been developed for assaying ATP. Methods such as magnetic resonance spectroscopy (MRS) of ³¹P require the patient to be at a facility which is available only at major hospitals or research institutes. Biopsies of skeletal muscles can be taken, but not of vital organs such as the heart, brain or liver. Methods using blood samples (specifically neutrophils) are relatively non-invasive and are amenable to routine testing. In addition the blood stream reaches almost every cell in the body and carries much information concerning what is going on. The method of measuring ATP used in the “ATP profile” dates from 1947 when McElroy poured a solution of ATP onto ground-up firefly tails and observed bright luminescence and found that the amount of light produced was proportional to the ATP concentration [45]. Thus he showed that the energy contained in ATP can produce light and this led the way to the development of bioluminescent measurements which can be carried out routinely and reproducibly with commercially available biochemical assay kits and bioluminescence equipment [46-49]. Light is produced when ATP reacts with D-luciferin and oxygen in the presence of Mg²⁺ and the enzyme luciferase. When ATP is the limiting reagent, the light emitted is proportional to the ATP present.

The “ATP profile” test

The “ATP profile” test yields 5 independent numerical factors from 3 series of measurements, (A), (B), and (C) on blood samples (neutrophils). Details of the measurements made and how the numerical factors were calculated are given in Appendix B. The 3 series are:

1. ATP concentration in the neutrophils is measured in the presence of excess magnesium which is needed for ATP reactions. This gives the factor ATP in units of nmol per million cells (or fmol/cell), the measure of how much ATP is present. Then a second measurement is made with just endogenous magnesium present. The ratio of this to the one with excess magnesium is the ATP Ratio. This tells us what fraction of the ATP is available for energy supply.

2. The efficiency of the oxidative phosphorylation process is measured by first inhibiting the ADP to ATP conversion in the laboratory with sodium azide. This chemical inhibits both the mitochondrial protein cytochrome a3 (last step in the ETC) and ATP synthase [50]. ATP should then be rapidly used up and have a low measured concentration. Next, the inhibitor is removed by washing and re-suspending the cells in a buffer solution. The mitochondria should then rapidly replete the ATP from ADP and restore the ATP concentration. The overall result gives Ox Phos, which is the ADP to ATP recycling efficiency that makes more energy available as needed.

3. The TL switches a single binding site between two states. In the first state ADP is recovered from the cytosol for re-conversion to ATP, and in the second state ATP produced in the mitochondria is passed into the cytosol to release its energy. Measurements are made by trapping the mitochondria on an affinity chromatography medium. First the mitochondrial ATP is measured. Next, an ADP-containing buffer is added at a pH that strongly biases the TL towards scavenging ADP for conversion to ATP. After 10 minutes the ATP in the mitochondria is measured. This yields the number TL OUT. This is a measure of the efficiency for transfer of ADP out of the cytosol for reconversion to ATP in the mitochondria. In the next measurement a buffer is added at a pH that strongly biases the TL in the direction to return ATP to the cytosol. After 10 minutes the mitochondria are washed free of the buffer and the ATP remaining in the mitochondria is measured and this gives the number TL IN. This is a measure of the efficiency for the transfer of ATP from the mitochondria into the cytosol where it can release its energy as needed.

The individual numerical factors

Figure 2 shows scatter plots (a point for each patient) of each of the 5 factors vs. CFS Ability. As we will see later it is convenient to divide the data from the 71 patients into 3 categories, “very severe”, “severe”, and “moderate”, which have about the same number of entries (25, 21, and 25). To the right of each scatter plot we show a stacked projection histogram for the 3 categories of Ability, and at the far right a histogram for the normal controls. Looking first at the ATP histogram for the normal controls we see a well defined minimum value with a long tail up to a maximum value of 2.89 fmol/cell. The average value is 2.00 ± 0.05 (SEM (Standard Error of the Mean), n=53) which can be compared with the measurement, 1.9 ± 0.1 (SEM, n=12), made some 25 years ago by the same technique in a study of the energetics of phagocytosis in neutrophils [51]. The stacked histogram for the patients and the Ability plot clearly show that some patients are in the normal region and some are below and they split into two groups with very little overlap. Rather than comparing the patients with the average of the normal control group which is customary, we prefer to compare with the minimum value of the control group which is more cleanly defined. Also, this method permits us to classify patients as being in the normal region or being below the normal region. Clearly this can be changed easily and all the numbers are given in Figure 2. We therefore show as a heavy horizontal dashed line the minimum value of each factor measured for the controls.

Figure 2A, ATP vs. CFS Ability, shows that the majority of the “very severe” and “severe” patients are below the normal minimum but very few are below 75% of this minimum. Note that 3 of the “very severe” patients are well into the normal region; they have problems with one or more of the 4 other factors. Just over 50% of the “moderate” patients are in the normal region. There is a small positive correlation which is indicated by the “trend” crosses. There is not a gentle increase in ATP with Ability, but an increase in the fraction of patients above the normal minimum line.

Figure 2B shows ATP Ratio vs. CFS Ability. The majority of patients in all 3 categories are below the normal minimum, and about 1/3 of “moderate” patients are below 75% of the normal minimum. The correlation with Ability is slightly negative. Values for the normal controls are rather tightly grouped with a minimum of 0.65 and average of 0.69.

The Ox Phos plot in Figure 2C shows a wide range of values and a strong positive correlation for this factor for the patient group. The stacked projection clearly shows that there are two groups – above and below the normal minimum and the upper group spans a similar range to the controls. Note the high value for the sole patient with CFS Ability = 0. This patient also has ATP = 1.26 and ATP Ratio = 0.59 which are not very far below the normal minima.

The TL OUT plot of Figure 2D also shows two groups with a rather sharp peak in the stacked projections just above the normal minimum and this closely matches the projection of the control group. Many patients, particularly the “very severe”, are far below the normal minimum.

The TL IN plot of Figure 2E also has a peak in the normal region. However, some patients have very low values, including the patient with CFS Ability = 0. The product TL OUT × TL IN is only 0.012 for this patient who is very severely ill whereas this product is 0.17 at the normal minima, a factor of 14 larger. If just ATP or Ox Phos had been measured the very severe mitochondrial dysfunction of this patient would not have been detected. Note the strong positive correlation for TL IN.

Most patients are below normal in more than one factor (average [range] is 3.7 [2 to 5] for “very severe”, 3.5 [2 to 5] for “severe” and 2.2 [1 to 4] for “moderate”). Some of the features are summarised numerically in Table 1.

For most of the factors the percentage of patients who are in the normal region increases in going from “very severe” to “severe” and to “moderate”. The exception is the ATP Ratio which gently decreases, but within the statistical errors is constant. Both TL IN and the product TL OUT × TL IN increase by large factors. For patients in the “moderate” category the main influence on their illness appears to be the ATP Ratio.

Table 1 also illustrates the importance of measuring more than one factor. For example, if only ATP had been measured, 28% of all the patients would be classified as normal, and if only Ox Phos had been measured, 32% of the “very severe” patients would be classified as normal.

Correlations between numerical factors

It is also helpful to look at correlations between pairs of numerical factors. The five most relevant examples are shown in Figure 3.

In the scatter plots of Figure 3 the normal region is the rectangular region in the upper right corner defined by the normal minima dashed lines. In the ATP Ratio vs. ATP plot (Figure 3A) most patients are fairly close to the normal region apart from the small cluster at ATP Ratio ∼ 0.35.

In the Ox Phos vs. ATP plot (Figure 3B) there are only a few patients, all “moderate”, in the normal region for both factors. Some of the “very severe” and “severe” patients are in the normal region for Ox Phos and some are far below. Note the apparent negative correlation for the normal controls. This shows that for normal subjects there is a compensatory mechanism, i.e. if ATP is high Ox Phos is low and vice versa. This is expected because the ATP concentration is a major factor in the control of the rate of the ox-phos process and the energy supply is adjusted to meet the energy demand. There is no obvious evidence for this effect in the patient group.

In the Ox Phos vs. TL OUT plot (Figure 3C) only 6 (all “moderate”) patients are in the normal region of both factors. Note that as a function of TL OUT there are 2 groups, a distinct narrow band to the right of the vertical dashed line and a spread-out group to the left of this line. Looking vertically, note that for the first group the Ability of patients (indicated by the 3 categories) is correlated with their value of Ox Phos. There is a large spread in this factor and again there appears to be two groups roughly divided by the horizontal normal minimum line. Some of the patients in all 3 categories are above this line, but they have problems with one or more of the other factors. Some of the “very severe” patients have Ox Phos lower than the normal minimum by an order-of-magnitude.

In the Ox Phos vs. TL IN plot (Figure 3D) there are many more patients in the normal region for both factors, but also many with very low values of one or both factors.

In the TL IN vs. TL OUT plot (Figure 3E) there are two clusters, one in the upper right corner which is the normal region for both factors, and another well below it at TL IN ∼ 0.1. Several patients are far below the normal minimum for both factors.

In the biochemical methods used we might expect some correlation between the TL factors and Ox Phos because they are closely coupled and interacting parts of the ADP to ATP reconversion cycle. However, the plots indicate that the biochemical methods used can separate the Ox Phos and TL factors and measure them individually.

To our knowledge this is the first time that such detailed effects have been observed.

The Mitochondrial Energy Score

The biochemical measurements in the “ATP profile” separate the energy generation and recycling processes into 5 steps. As in any multistep process, for example electrical power production or an assembly line, the efficiency of the overall process is the product of the efficiencies of the individual steps. Any suggestion of relative weighting is irrelevant; it only results in an overall normalization factor. The product of ATP and ATP Ratio is the cellular concentration of ATP complexed with magnesium and this is the available energy supply of ATP. Ox Phos is the efficiency of the ETC which converts ADP into ATP. However, for the recycling of ADP to make more energy available the Translocator protein must efficiently have its binding site facing out to collect ADP (TL OUT) and alternately facing in (TL IN) to efficiently transmit ATP from the mitochondria into the cytosol where its energy can be used.

We have found it useful to calculate the product of the five factors, the overall mitochondrial energy-producing relative efficiency, and call it the Mitochondrial Energy Score. We just multiply the 5 factors together for each patient and each control. The minimum value for the controls is 0.182 fmol/cell. We have chosen this as our normalisation point so we divide all the Energy Scores (for both patients and controls) by this value. Thus all controls have Mitochondrial Energy Score ≥ 1.00.

A scatter plot of the Energy Score for each patient at each value of CFS Ability and each control is shown in Figure 4A. The horizontal dashed line indicates the minimum value for the normal controls and this is our normalization value of 1.00. Only one of the 71 patients has an Energy Score > 1 (namely 1.25 for one of the patients with Ability = 7). However this patient has 2 of the 5 factors below the normal minima.

Note the high degree of correlation between Energy Score and CFS Ability and this is independent of where the mean or minimum of normal subjects is. It is natural to believe that the CFS Ability of patients is more likely to depend upon mitochondrial dysfunction than vice versa, so we should really plot CFS Ability vs. Energy Score. However, the Ability was measured first, and Figure 4A shows convincingly that mitochondrial dysfunction is a major risk factor, and this has not been demonstrated before. Also shown in Figure 4A is the best straight line fit to all 71 entries. The fit is good, but there is no reason that the relationship should be a straight line. Table 2 gives the parameters of the fit. The Standard Error in the slope of the fitted straight line is so small that the probability P of the null hypothesis (i.e. that the slope is zero) is extremely small, P <0.001, when computed from the Student's t-distribution [52]. The 99.9% confidence interval is 0.092 < β < 0.174 where β is the true slope and this lower limit is still several Standard Errors above zero.

In Figure 4B the Energy Scores are plotted as a function of the age of each participant. It is believed that mitochondria play a major role in the aging process [31, 33] so there is the possibility that the younger mean age of the control group as compared to the patient group may influence our results. We have looked at the age dependence of all 5 factors and see no effect, and this is not surprising in view of the wide spread in values of each factor. The Energy Score is a more reliable measure of mitochondrial dysfunction. In Figure 4B there is no evidence for age dependence in the control group but the maximum age is only 65 years (points for three controls have been omitted because their Energy Score is more than 2.00 with a maximum of 2.83). There are six patients of age > 70, one “very severe” with Ability = 2, one “severe” with Ability = 3, and four “moderate” who all have Ability = 4. These four patients have Energy Scores which are below the average (0.42) for this Ability so there may well be a decrease with increasing age. On the other hand there is a 33-year old patient who also has Ability = 4 and is well below the average. Excluding the six patients of age ≥ 70 slightly improves the straight line fit R² = 0.677) but has negligible effect on the other parameters or our conclusions.