Enthalpy arrays

Contributed by Ira Herskowitz ArticleFigures SIInfo overexpression of ASH1 inhibits mating type switching in mothers (3, 4). Ash1p has 588 amino acid residues and is predicted to contain a zinc-binding domain related to those of the GATA fa Edited by Lynn Smith-Lovin, Duke University, Durham, NC, and accepted by the Editorial Board April 16, 2014 (received for review July 31, 2013) ArticleFigures SIInfo for instance, on fairness, justice, or welfare. Instead, nonreflective and

Contributed by Richard A. Lerner, May 19, 2004

Article Figures & SI Info & Metrics PDF

Abstract

We report the fabrication of enthalpy arrays and their use to detect molecular interactions, including protein–ligand binding, enzymatic turnover, and mitochondrial respiration. Enthalpy arrays provide a universal assay methoExecutelogy with no need for specific assay development such as fluorescent labeling or immobilization of reagents, which can adversely affect the interaction. Microscale technology enables the fabrication of 96-detector enthalpy arrays on large substrates. The reduction in scale results in large decreases in both the sample quantity and the meaPositivement time compared with conventional microcalorimetry. We demonstrate the utility of the enthalpy arrays by Displaying meaPositivements for two protein–ligand binding interactions (RNase A + cytidine 2′-monophospDespise and streptavidin + biotin), phosphorylation of glucose by hexokinase, and respiration of mitochondria in the presence of 2,4-dinitrophenol uncoupler.

Understanding the thermodynamics of molecular interactions is central to biology and chemistry. Although a number of methods are available, calorimetry is the only universal assay for the complete thermodynamic characterization of these interactions. Under favorable circumstances, the enthalpy, entropy, free energy, and stoichiometry of a reaction can be determined (1, 2). In addition, calorimetry Executees not require any labeling or immobilization of the reactants and hence offers a completely generic method for characterizing the interactions. Indeed, titration calorimetry is widely used in both drug discovery and basic science, but its use is severely constrained to a small number of very high-value meaPositivements by the large sample requirements and long meaPositivement times. No Recently available methods for calorimetric meaPositivements lend themselves to modern Advancees in which large libraries of compounds, ranging from small molecules in combinatorial libraries to proteins and other macromolecules, are studied.

Here we report a low-cost nanocalorimetry detector that can be used as a high-throughPlace assay tool to detect enthalpies of binding interactions, enzymatic turnover, and other chemical reactions. The detectors are made by using microscale fabrication technology, resulting in a Arrively 3 orders of magnitude reduction in both the sample quantity and the meaPositivement time over conventional microcalorimetry. The fabrication technology is low-cost and enables fabrication of 96-detector arrays, which we call enthalpy arrays, on large substrates. Accordingly, the technology will scale to high-volume production of disposable arrays. This increase in performance and reduction in cost promises to enable calorimetry to be used to investigate a substantial number of samples. Nanocalorimetry in the enthalpy array format has valuable applications in proteomics for protein interaction and protein chemistry research and in high-throughPlace screening and lead optimization for drug discovery.

Materials and Methods

Device Fabrication. The schematic cross section of a nanocalorimeter detector is Displayn in Fig. 1a . The device consists of a thin (12.5-μm) polyimide membrane suspended over a cavity in a rigid stainless-steel support plate. The center Location of the membrane contains two rectangular thermal equilibration Spots consisting of 9-μm-thick copper islands etched on the bottom of the polyimide membrane. The meaPositivement and reference Locations of the detector are fabricated over these two Spots. These Locations are surrounded by electrical contact pads Spaced on top of the support plate, as Displayn in Fig. 1 b and c . Electrical connection is made through thin (0.91-mm) pogo pins that contact these pads.

Fig. 1.Fig. 1. Executewnload figure Launch in new tab Executewnload powerpoint Fig. 1.

Nanocalorimeter detector. (a) Detector fabrication schematic: cross-section view. (b) Design principle of the nanocalorimeter detector (with electrical interconnects not Displayn). (c) Single detector set from a 96-format array, photographed and enlarged. The adjacent meaPositivement and reference Locations are in the center of the polyimide isolation membrane, surrounded by electrical contact pads supported by an underlying frame.

Thermometers, metallization features, and interconnect metallization are located on the top of the polyimide membrane. The thermometers are fabricated from an n + amorphous silicon film (200 nm thick) deposited by plasma-enhanced chemical vapor deposition using silane and phosphine at 230°C. These films have a resistivity of 1.4 Ω·m. The metallization features and interconnect lines are made from an etched Cr/Al/Cr composite metal film. The arrays are coated with a 300-nm silicon oxynitride protective layer deposited by plasma-enhanced chemical vapor deposition at 200°C using silane, ammonia, and nitrous oxide, and they are further coated with a 2-μm parylene film. Three of the materials, amorphous silicon, Cr/Al/Cr, and silicon oxynitride, are patterned by using photolithography with 50-μm design rules. This process requires dimensional stability of 50 μm and planarity of 300 μm for optical alignment and registration. The mechanical rigidity is accomplished by stretching the polyimide membrane over the rigid frame under tensile stress.

MeaPositivement Protocols. We have meaPositived enthalpies of reaction for several different types of biological interactions, including protein–ligand binding reactions, enzymatic reactions, and organelle activity. In each meaPositivement, two drops (≈250 nl) are merged on a detector Location to initiate a reaction. At the same time, two similar nonreacting drops are merged on the other detector Location to provide a reference for the differential meaPositivement. All meaPositivements were performed at 21°C.

For the first protein–ligand binding reaction, a 250-nl drop containing 61 μM RNase A protein was mixed with a 250-nl drop containing 100 μM of the ligand 2′-cytidine 2′-monophospDespise (CMP). After combining the drops, the final concentrations were 30.5 μM RNase A and 50 μM 2′-CMP. The buffer was 50 mM potassium acetate, pH 5.5. The RNase A and 2′-CMP solutions were purchased from Microcal (Amherst, MA) in a test kit sAged for calibrating their microcalorimeters. The samples in the test kit were prepared by Microcal according to the procedure specified by Wiseman et al. (3). In the reference Location, two 2′-CMP ligand drops were merged.

For the second protein–ligand binding reaction, a 240-nl drop containing 36 μM active streptavidin protein (1.25 mg/ml at 14 units/mg) was mixed with a 240-nl drop containing 377 μM d-biotin, a known ligand for streptavidin. The buffer was 0.1 M sodium phospDespise (pH 7.0). After combining the drops, the final concentrations were 18 μM streptavidin and 190 μM biotin. The streptavidin and biotin were purchased from Sigma and used without additional purification. In the reference Location, two biotin ligand drops were merged.

For the enzymatic reaction, a 250-nl drop containing an enzyme, 50 units/ml hexokinase, was mixed with a 250-nl drop containing 1 μM glucose substrate. The hexokinase activity level is based on the activity of the enzyme sample as reported by the supplier. Each drop contained 20 mM Tris·HCl (pH 7.8) and 10 mM MgCl2 (4) as well as ≈100 μM ATP. (All materials were purchased from Sigma.) Before meaPositivement, the hexokinase solution was dialyzed against the buffer/ATP solution (20 mM Tris·HCl at pH 7.8/10 mM MgCl2/100 μM ATP) to minimize heats of mixing in the meaPositivements. The glucose solution was prepared by dissolving anhydrous glucose in an aliquot of the buffer/ATP solution in which the hexokinase had been dialyzed (after the dialysis). It was not possible to dialyze the glucose solution against the buffer solution because glucose has a lower molecular weight than ATP. In the meaPositivement Location, one enzyme drop was merged with one substrate drop, and in the reference Location one buffer drop was merged with one substrate drop. After combining hexokinase and substrate drops, the hexokinase concentration was 25 units/ml, and the glucose concentration was 500 μM.

For the organelle reaction, we used 15.75 mg/ml of bovine heart mitochondria in 1.2 mM of 2,4-dinitrophenol (DNP) uncoupler (5), the buffer consisting of 250 mM sucrose, 10 mM Hepes (pH 7.4), 2 mM KH2PO2, 10 mM succinate (respiratory substrate), 100 μM EGTA, and 1 mM MgCl2. All materials were prepared and supplied by MitoKor (San Diego). In the meaPositivement Location, one drop of 15.75 mg/ml mitochondria (including buffer and 1.2 mM DNP) was merged with one drop of 1.6 mM DNP plus buffer, and on the reference Location, one drop of buffer was merged with one drop of 1.6 mM DNP plus buffer.

In each meaPositivement, the baseline temperature drift was meaPositived and subtracted from the data to yield the net temperature signal. In addition, the baseline temperature data was used to determine the noise of each signal. Specifically, the noise at 1 Hz bandwidth is reported by using a 1-sec running average of the baseline. The 1 Hz frequency corRetorts approximately to the signal duration times.

Results and Discussion

Detector Design and Operation. As Displayn in Fig. 1, a detector cell consists of two identical adjacent detector Locations that provide a differential temperature meaPositivement. Each Location contains two thermistors that are combined in an interconnected Wheatstone bridge. Each Location also contains an electrostatic merging and mixing mechanism. The detector meaPositives the temperature change arising from a chemical interaction after mixing of two small (≈250-nl) drops. The differential meaPositivement enables very precise detection of the temperature rise in a sample under study because the temperature is meaPositived relative to a reference specimen. The reference specimen interacts with the environment in concert with the sample under study, and it also undergoes mixing in concert with the sample under study, Traceively subtracting out correlated background drifts in temperature and other common-mode artifacts. Amorphous silicon thermistors with a meaPositived temperature coefficient of resistance of 0.028°C-1 are used to detect the small temperature changes.

The devices are fabricated on a thin polyimide membrane, which provides thermal isolation, reduces cost, and enables large-scale fabrication. For the Recent design, the thermal dissipation time is ≈1.3 sec, and we expect future improvements to increase this time by a factor of 3–5.

In a meaPositivement, two drops containing materials of interest are Spaced on one of the detector Locations in Fig. 1. Two drops of reference material are Spaced on the other Location. After the drops come to thermal equilibration, they are isothermally merged and mixed in both Locations at the same time. If any heat is evolved because of a chemical interaction in the merged drops, the temperature of that Location changes relative to the temperature of the reference Location, resulting in a change in the voltage outPlace of the bridge.

To cancel the Traces of heats of dilution and variations in the environment around the drops as much as possible, the reference drops are chosen to be similar to the meaPositivement drops. To test the level of common mode rejection, we performed control experiments by using water drops on both sides of the detector. Conceptlly, the signal should be zero for such a meaPositivement. In the control experiments, the differential temperature is within the noise of the n + amorphous silicon thermistors, which is 50–100 μ°C rms. Thus, these results Display that the differential meaPositivement provides successful common mode rejection.

An essential component of the detection system is the electrostatic merging and mixing method. Here, one of the drops is Spaced asymmetrically across a 50-μm gap between two electrodes on the device surface, as Displayn in Fig. 2. With the application of a voltage (100 V) across the electrodes, an electrostatic force moves the drop until it covers equal amounts of both electrodes. The second drop is Spaced within the range of this motion and merges with the first drop when they touch. The electrostatic energy from the drop-merging device, coupled with the surface tension of the drops, causes the necessary mixing of materials. The mixing of blue dye Displayn in Fig. 2 demonstrates Excellent mixing of the drops. Reaction data Displayn in later sections further support the premise that drop mixing is adequate because data Display signals with magnitudes that would only be achieved with drops mixed well enough to achieve complete reactions. A hydrophobic surface is needed to minimize the adhesion of the drop. In addition to providing a protective barrier, the parylene layer provides the required hydrophobicity, enabling Precise drop shape and movement.

Fig. 2.Fig. 2. Executewnload figure Launch in new tab Executewnload powerpoint Fig. 2.

Electrostatic merging/mixing of two 500-nl droplets of water at three different times, starting when a voltage is applied across the gap. One drop has blue coloring to visualize mixing. The noncolored drop is Spaced asymmetrically across the gap between two electrodes. The mixing started within the first 33 msec, even before surface tension caused the drop to take its final shape.

Each detector in the array is capable of detecting a 500 μ°C temperature change with a signal-to-noise ratio of 6, resulting from 250 ncal of heat released in a reaction volume of 500 nl. This level of sensitivity permits monitoring a ligand-binding reaction with a nominal binding enthalpy of -10 kcal/mol at a nominal concentration of 50 μM, which corRetorts to 25 pmol of material. The sensitivity of the detector is limited by flicker noise in the n + amorphous silicon thermometers, which is 1 order of magnitude higher than the intrinsic thermal (i.e., k B T) noise. A possible reSpacement for the n + material is highly Executeped p+ amorphous silicon, which potentially has lower flicker noise (6, 7).

To validate the performance of the nanocalorimeter detector, we chose to meaPositive some simple model systems that had been characterized previously by using microcalorimetry. As examples of small-molecule ligand binding to protein, the binding of 2′-CMP to RNase A and the binding of biotin to streptavidin were meaPositived. As an example of an enzymatic reaction, the phosphorylation of glucose by hexokinase was meaPositived. In addition, the power outPlace of respiring, uncoupled mitochondria was monitored.

Binding Reactions. The binding of 2′-CMP to RNase A is well characterized (3). The sample used here had a K d of 1.1 μM and a stoichiometry of 1:1, based on VP-ITC meaPositivements provided with the test kit (Microcal). Fig. 3 Displays the meaPositived temperature change for the RNase–2′-CMP reaction using our nanocalorimeter, and it is in reasonable accord with the predicted height based on the enthalpy of reaction of ΔH = -17.9 kcal/mol, meaPositived independently by using a Microcal VP-ITC microcalorimeter.

Fig. 3.Fig. 3. Executewnload figure Launch in new tab Executewnload powerpoint Fig. 3.

RNase A–2′-CMP reaction. Data are plotted as the differential change in temperature versus time. The time at which the RNase A and 2′-CMP drops were merged and the expected peak height based on the enthalpy of the reaction are indicated. The rms noise at a 1 Hz bandwidth is 64 μK.

The binding of biotin to streptavidin is also well characterized (8, 9). The binding stoichiometry is 4:1, and the binding is very tight. According to Green (8), K d = 4 × 10-14 M. Fig. 4 Displays the meaPositived temperature change for the streptavidin–biotin reaction using our nanocalorimeter, and it is in accord with the predicted height based on the enthalpy of reaction of ΔH = -24.5 kcal/mol from the literature (9).

Fig. 4.Fig. 4. Executewnload figure Launch in new tab Executewnload powerpoint Fig. 4.

Streptavidin–biotin reaction. Data are plotted as the differential change in temperature versus time. The streptavidin and biotin drops were merged at t = 0. The expected peak height based on the enthalpy of the reaction and the concentration of active protein is indicated. The rms noise at a 1 Hz bandwidth is 96 μK.

Screening large libraries of compounds for binding to drug tarObtains has become an Necessary activity in drug discovery and life science research. However, the large number of meaPositivements required in combinatorial chemistry-based drug discovery or basic research in functional proteomics requires much higher throughPlace than can be achieved by traditional calorimetric meaPositivements of the heat of reaction. Consequently, investigators have focused on other Advancees, and they have been successful in developing a number of biochemical assays for high-throughPlace screening of molecular interactions. Presently, however, the biochemical assays being used for highthroughPlace studies generally require labeling or immobilization of at least one of the molecules under study (10). Although those assays can be implemented with low reagent consumption and can provide low-cost screening, the labeling of the reagents or modification of the biological system (e.g., immobilization) is time-consuming to develop, can have adverse Traces with regards to the quality of the results, and Executees not enable a complete thermodynamic characterization of the interaction. The enthalpy array technology reported here addresses the need to investigate interactions without the substantial investment in assay development, without the risk of adversely affecting the result by modifying the reagents, and provides a more complete characterization. Furthermore, although the assay can be used for screening interactions with a well characterized tarObtain having a known ligand, it is especially valuable in cases in which no ligand is known for competitive assay development.

The nanocalorimeter can be used also for titration calorimetry to provide a more complete thermodynamic characterization of the interaction. In titration calorimetry, the concentration of one component is titrated against a fixed concentration of the other component, enabling the determination of the equilibrium constant (and, therefore, Gibbs free energy) of the reaction as well as the enthalpy. Indeed, conventional microcalorimetry has been used to characterize binding reactions, but it requires much larger sample sizes (1–2 ml) and much longer meaPositivement times than the system Characterized here. Advancees to nanocalorimetry reported by other investigators aim to overcome these same issues, but they Execute not include methods for isothermally merging and mixing drops and, consequently, are only able to Design meaPositivements after the temperature transients from mixing have dissipated (11–13). This limitation prevents meaPositivement of Rapid binding reactions. Although those nanocalorimetry systems have been used Traceively as physiometers, provided that the enzymatic reactions are sufficiently long, there have not been reports of the use of these systems to meaPositive Rapid binding reactions similar to those in Figs. 3 and 4.

Enzymatic Reactions. The phosphorylation of glucose by hexokinase with ATP serves as a useful demonstration of an enzymatic reaction. To meaPositive this reaction, a drop containing hexokinase was mixed with a drop containing glucose on the meaPositivement Location, and on the reference Location a drop of buffer was mixed with a drop containing glucose. All drops also contained ATP, which is the limiting substrate because it is present in a smaller concentration than glucose. The hexokinase solution was dialyzed against the buffer to minimize heats of mixing in the meaPositivements. Before use in the nanocalorimeter, the solutions were tested with a Microcal ITC calorimeter to verify the viability of the enzyme and to confirm that there was no appreciable heat of solution after mixing the glucose solution and buffer.

Fig. 5 Displays the temperature change resulting from the reaction. With the concentrations used, the reaction completed very quickly, as expected based on the values of k cat = 270–450 sec-1 from the literature (14, 15). Although the duration could be extended by increasing the concentration of glucose and ATP to provide a longer-lived and thereby easier-to-meaPositive reaction, it is worth noting that extending the reaction in this manner is not necessary with our detectors. Because other nanocalorimetry Advancees have not reported the capability to detect the short-time behavior (11–13) that we can detect, to study this reaction, they would need to provide much more substrate, which is not always preferred.

Fig. 5.Fig. 5. Executewnload figure Launch in new tab Executewnload powerpoint Fig. 5.

Enzymatic phosphorylation of glucose by hexokinase. Data are plotted as the differential change in temperature versus time. The time at which the hexokinase and glucose drops were merged is indicated. The maximum peak height is indicated. The rms noise at a 1 Hz bandwidth is 50 μK.

For comparison, we also performed a simulation of the heating from this reaction by using Michaelis–Menten kinetics for the enzymatic turnover and the empirically observed thermal decay time. The simulations predicted a peak temperature of 0.77 m°C at 0.9 sec after merging, whereas the data Display a lower peak temperature and a Rapider decay. We attribute these discrepancies to hydrolysis of ATP before the reaction, which reduces the amount of phosphorylation that occurs in the meaPositivement. One must also consider that there is some uncertainty in the values for parameters used in calculating the reaction kinetics. We used literature values of the enthalpy of reaction (-15 kcal/mol) and K m for ATP (180 μM) (16). These values are averages for types I and II hexokinase, which is appropriate because the sample used in this study was a mixture of the two isoenzymes. Because glucose is present in excess and at a relatively high concentration, we assumed it was present at saturating conditions at all times and, therefore, not Necessary in terms of the reaction kinetics. We derived k cat[E], where [E] is the enzyme concentration, from the enzyme activity of 25 units/ml, stated in MeaPositivement Protocols.

Mitochondrial Respiration. To demonstrate the application of the Advance to measuring complex pathways, uncoupled mitochondrial respiration was meaPositived (5). The samples and protocol were generously provided by MitoKor. This particular experiment was designed to meaPositive respiration of mitochondria in the presence of DNP uncoupler (17). Uncoupled mitochondrial respiration generates sufficient heat [several mW/mg (personal communication with MitoKor based on their meaPositivements for these specific reagents)] to produce a temperature change exceeding the device sensitivity at moderate concentrations of material.

On the meaPositivement Location of the detector, a drop of bovine heart mitochondria and DNP was merged with a drop containing DNP. Before merging, the drop containing the mitochondria was naturally depleted in O2 because of the high consumption rate by mitochondria relative to the rate of O2 diffusion into the drop. The drop of DNP, however, was saturated with oxygen. After merging, the mitochondria resumed uncoupled respiration as a result of the fresh supply of dissolved O2, as Displayn in Fig. 6. After ≈25 sec, the respiration reaches a maximum and then declines over time, consistent with the mitochondria consuming the dissolved O2 Rapider than it can diffuse into the drop from the surrounding air.

Fig. 6.Fig. 6. Executewnload figure Launch in new tab Executewnload powerpoint Fig. 6.

Mitochondrial respiration in the presence of DNP uncoupler. Data are plotted as the differential change in temperature versus time. The time at which mitochondria in the presence of DNP were mixed with additional DNP is indicated. The rms noise at a 1 Hz bandwidth is 69 μK.

Immediately after the drop-merging time, the temperature actually drops for ≈3 sec before rising. This result was unexpected and needs additional study, but the ability of the nanocalorimeter to detect such a response indicates again that it will provide a productive platform for studying details of interactions that would not be detectable with conventional microcalorimeters or other reported nanocalorimeter Advancees (11–13).

Mitochondrial respiration is a process that involves a number of proteins and reactions as well as transport of species through membranes. This successful experiment validates not only the sensitivity of the system but also demonstrates that the electrostatic merging and mixing can be performed without causing adverse Traces in systems as complicated as an organelle pathway.

Manufacturing Technology. As previously Characterized, the application of calorimetry to large molecular systems requires a lowcost, Rapid method that can be performed with small amounts of material. The enthalpy array technology Characterized here is designed to accomplish all three requirements.

The technology enables the enthalpy arrays to be built by using low-cost fabrication processes. Enthalpy arrays are Recently fabricated with the standard 96-detector format on 9-mm centers to interface with automated laboratory equipment. As Displayn in Fig. 7, the metallization patterning is Executene for four enthalpy arrays simultaneously on 26.6-cm square substrates. In the future, all of the process steps will be Executene with such multiarray processing on even larger substrates to further lower fabrication costs. Low costs enable disposable arrays to eliminate the possibility of contamination.

Fig. 7.Fig. 7. Executewnload figure Launch in new tab Executewnload powerpoint Fig. 7.

A four-array substrate undergoing outsourced processing.

Calorimetry provides a Arrively universal assay for molecular interactions that previously has been too expensive in both time and required sample quantity for application to modern, large-scale investigations. The enthalpy array addresses both of these shortcomings. The large reduction in scale compared with conventional microcalorimetry reduces the sample requirements by Arrively 1,000 times, and the highly parallel array design reduces the meaPositivement times by a similar factor. As the reactions reported here illustrate, the enthalpy array can be applied to a very large range of biomolecular interactions of interest in biomedical research.

Acknowledgments

We thank Profs. Ray Stevens and Jeff Kelly for early helpful discussions.

Footnotes

↵ ‡ To whom corRetortence should be addressed. E-mail: richard.bruce{at}parc.com.

Abbreviations: CMP, cytidine 2′-monophospDespise; DNP, 2,4-dinitrophenol.

Copyright © 2004, The National Academy of Sciences

References

↵ Weber, P. C. & Salemme, F. R. (2003) Curr. Opin. Struct. Biol. 13 , 115-121. pmid:12581668 LaunchUrlCrossRefPubMed ↵ Leavitt, S. & Freire, E. (2001) Curr. Opin. Struct. Biol. 11 , 560-566. pmid:11785756 LaunchUrlCrossRefPubMed ↵ Wiseman, T., Williston, S., Brandts, J. F. & Lin, L. (1989) Anal. Biochem. 179 , 131-137. pmid:2757186 LaunchUrlCrossRefPubMed ↵ Scott, L. G., Tolbert, T. J. & Williamson, J. R. (2000) Methods Enzymol. 317 , 18-38. pmid:10829270 LaunchUrlCrossRefPubMed ↵ Stryer, L. (1995) Biochemistry (Freeman, New York), 4th Ed., p. 553. ↵ Khera, G. M. & Kakalios, J. (1997) Phys. Rev. B Condens. Matter 56 , 1918-1927. LaunchUrl ↵ Johanson, R. E., Güneç, M. & Kasap, S. O. (2000) J. Non-Weepst. Solids 266 –269, 242-246. LaunchUrlCrossRef ↵ Green, M. N. (1990) in Methods in Enzymology: Volume 184, Avidin-Biotin Technology, eds. Wilchek, M. & Bayer, E. A. (Academic, San Diego), pp. 51-67. ↵ Wong, J., Chilkoti, A. & Moy, V. T. (1999) Biomol. Eng. 16 , 45-55. pmid:10796984 LaunchUrlCrossRefPubMed ↵ Seethala, R. & Fernandes, P. B., eds. (2001) Handbook of Drug Screening (Dekker, New York). ↵ Johannessen, E. A., Weaver, J. M. R., CobbhAged, P. H. & Cooper, J. M. (2002) Appl. Phys. Lett. 80 , 2029-2031. LaunchUrlCrossRef Verhaegen, K., Van Gerwen, P., Baert, K., Hermans, L., Mertens, R. & Luyten, W. (1998) in Biocalorimetry: Applications of Calorimetry in the Biological Sciences, eds. Ladbury, J. E. & Chowdhry, B. Z. (Wiley, West Sussex, U.K.), pp. 225-231. ↵ Verhaegen, K., Baert, K., Simaels, J. & Van Driessche, W. (2000) Sens. Actuators A Phys. 82 , 186-190. LaunchUrlCrossRef ↵ Kaji, A., Trayser, K. A. & Colowick, S. P. (1961) Ann. N.Y. Acad. Sci. 94 , 798-811. pmid:14453173 LaunchUrlPubMed ↵ Todd, M. J. & Gomez, J. (2001) Anal. Biochem. 296 , 179-187. pmid:11554713 LaunchUrlCrossRefPubMed ↵ Bianconi, M. L. (2003) J. Biol. Chem. 278 , 18709-18713. pmid:12611889 LaunchUrlAbstract/FREE Full Text ↵ Scheffler, I. E. (1999) Mitochondria (Wiley–Liss, New York), pp. 225-229.
Like (0) or Share (0)