The papers by Van Voorhies in Free Radical Biology & Medicine (33, 587–596, 2002) and in this Journal claim that the major longevity-extending mutations in C. elegans essentially act by reducing metabolic rate as predicted by the rate-of-living theory, and do not alter any metabolically independent mechanism specific to aging. In contrast, we found no evidence of a reduction in metabolic rate in these mutants using different experimental approaches. Now, Van Voorhies challenges the accuracy of our experimental results.

Some of the criticisms raised by Van Voorhies point to misunderstanding. For example, he is puzzled by the fact that the energy value of the oxygen consumed fails to match the measured heat flux. The simple answer is that this conversion is inappropriate since different instruments involving different environmental conditions are used, as explained in our contribution. Van Voorhies also assumes that the ATP content in the worms is bound to remain constant because of the well known principle that ATP is made as needed, and he infers that our method must be inappropriate ‘because significant amounts of ATP can hydrolyse before heat inactivation of ATPases occurs’. The fact is that we submerge frozen (−75 °C) microvials, containing no more than 100 µL worm suspension, in boiling water. It takes little time for the sample to reach boiling temperature. The ATP concentrations thus measured still decrease exponentially with age. Do we really lose ATP as an exponential function of age? Van Voorhies suggests that ATPase activity can be responsible for this effect. Because ATP shows a 7–10-fold (not 30-fold as mentioned in the Van Voorhies paper) decrease and oxygen consumption decreases a mere 3-fold, this implies that ATPase activity should increase 3-fold with age. This would be a major discovery. However, in our opinion it is very unlikely that ATPases, an impressive group of enzymes involved in a myriad of cellular functions, would increase their activity 3 times (on average) over life span. The alternative explanation is that, in C. elegans, ATP effectively decreases with age in vivo, although the mechanism responsible for this change remains unknown. We hypothesize that the efficiency of ATP production may decline, which would also explain the faster decline of ATP relative to oxygen consumption and heat output.

Van Voorhies criticises the lucigenin-mediated light production assay. As explained in our contribution, this assay measures the light that is produced when freeze-thawed worms are suspended in assay medium containing lucigenin, NADH, NADPH and KCN (blocking cytochrome oxidase and Cu/ZnSOD). The nicotineamide coenzymes drive reactions that produce superoxide to a maximal speed limited by the activity of the biochemical reactions involved. Superoxide thus produced reacts with lucigenin, resulting in luminescence. Superoxide is produced at several sites, but we have estimated that respiration contributes well over 70% of all superoxide produced (Braeckman et al., Mech. Aging Dev., 123, 105–119, 2002). It should be stressed that this assay measures a capacity, a maximum metabolic (mostly mitochondrial) output under the artificial assay conditions. The decrease of luminescence with age reflects a progressive reduction of metabolic (mostly mitochondrial) capacity, not metabolic rate. In the case of the mitochondrial contribution, the light production assay reflects the activity of complex I and downstream elements to transmit electrons under uncoupled conditions (resulting from membrane damage during freezing). Since cytochrome oxidase is blocked, the upstream electron transport system will be in a reduced form, resulting in enhanced superoxide production which remains dependent, however, on the flux of electrons from complex I to cytochrome oxidase. This flux declines with age probably because of increasing failure of components of the electron transport chain. We first saw a similar age-specific decline of lucigenin-mediated light production in microsome fractions prepared from fer-15 and fer-15; age-1 mutant worms. Age-dependent decline was attenuated in the Age mutant (Vanfleteren, Biochem. J., 292, 605–608, 1993).

Van Voorhies also sees conceptual problems in the fact that light production levels decrease far more quickly than measures of oxygen consumption. Our explanation is that the light production assay measures a capacity, i.e. an activity in vitro under forced artificial conditions. The performance measured under these conditions can very well decline before any effect becomes visible in vivo, under natural conditions. Similarly, physiological tests can indicate reduction of function well before the first signs of malfunction appear under normal or resting conditions.

Van Voorhies criticises variation of light production rates by worms of similar ages up to 275-fold. The simple reason is that light emission is recorded as arbitrary units, depending on the sensitivity setting of the photomultiplier tube, and influenced by modifications of the assay conditions over the years. We could have precluded these suspicions by expressing all results as percentage of maximum yield. For example, a non-thermostatted Model 2000 Integrating Photometer receiving a single scintillation vial was used initially and was later (from 1999 onward) replaced by the Victor and Victor2 thermostatted Multilabel Counter microplate readers. There is some variation in age-dependent decline of light production when comparing successive publications. This potentially confounding variability is alleviated by running control and experimental strains in parallel, under rigidly controlled experimental conditions. It should also be stressed that we expressed the data as a function of total age in our papers up to 1998. From 1999 onward, we preferred to relate the measurements to adult age, since the time required for development can differ substantially among strains. Thus it is not clear how the data for 1999–2002 (Van Voorhies, Table 2) were derived. Anyway, appropriate comparison of the age-specific declines would involve curve fitting, using all available data points, rather than using ratios of selected time points. As an example, for the last three Braeckman publications (as listed in Van Voorhies Table 2), exponential decrease rates of chemiluminescence were 21.0, 31.2 and 29.3% per day, respectively.

One of the major concerns of Van Voorhies is that the culture conditions of the worms should be optimal and match their natural habitat. He further suggests that culturing worms in liquid is stressing and does not allow the animals to age in a normal way. We recently analysed metabolic features of worms cultured on agar plates with lawns of E. coli and worms cultured in monoxenic liquid culture. The age-dependent patterns matched each other very well (Houthoofd et al., Exp. Gerontol., 37, 1015–1021, 2002). In that context, it is worth correcting Table 2 of the Van Voorhies paper: growth conditions described in the Braeckman papers were all liquid, E. coli (not ‘axenic’) and at 24 °C while measurement conditions were liquid, axenic (instead of ‘no food’) at 25 °C and 1000s of worms were used. Moreover, it is not clear to us how our oxygen consumption values of the 1999 and 2002 papers can differ 2-fold after Van Voorhies’ recalculations. Our 1999 data were replotted in figure 7 of the 2002 paper and showed, for that particular first time point, a 1.5-fold difference with the new data presented in figure 8 of that same paper.

Another area of controversy relates to normalization. In a previous paper, Van Voorhies and Ward (PNAS, 96, 11399–11403, 1999) normalized all measurements to worm number, with the implication that no distinction was made between small and big worms. Van Voorhies claims that only worms of equivalent size and development stage were compared but it is not clear how equivalence for size and development is to be obtained at the same time, when one strain develops slower and is smaller than the other at all ages (e.g. clk-1; daf-2 vs. N2). In his recent paper, Van Voorhies criticises the use of mass-specific units since this approach assumes that metabolic rate varies isometrically with body mass. We agree with this statement, and show in our paper that this problem can be solved by expressing the data in an appropriate allometric relationship.

Finally, Van Voorhies defends CO2 dissipation as a proxy for oxygen consumption arguing that variation in the respiratory quotient is relatively modest, which in turn suggests that the glyoxylate cycle and pentose phosphate shunt are not a significant source of CO2 production. However, there are only two data points referring to the long-lived mutant e1368. This is a class I daf-2 mutant. The long-lived mutant studied previously (Van Voorhies and Ward, 1999) was daf-2(e1370) which is a class II mutant. Measurement of CO2 dissipation is a valuable proxy for respiration assays, but a mutation-specific shift in the biochemical pathways involved remains a potentially confounding problem.