Measurement of oxygen consumption in Tenebrio molitor using a sensitive, inexpensive, sensor-based coulometric microrespirometer

Measuring metabolic rates provides essential quantitative insight into such diverse fields as climate change (Seebacher et al., 2015), food production (Bjøge et al., 2018), environmental quality (Villasenor et al., 2011) and human health (Rising et al., 1992). Metabolism can be measured directly from the heat produced by an organism, or indirectly from O₂ consumption or CO₂ production. If the assumptions of the methods are satisfied and the equipment is adequately sensitive, all of the above methods can accurately measure metabolism (Lighton, 2019). However, most available methods either require expensive and/or cumbersome apparatus (Hansen et al., 2004; Odell, 1998) or do not maintain constant recording conditions (Burggren et al., 2017).

Coulometric respirometry (also known as electrolytic respirometry) is a long-established method for measuring the rate of O₂ consumption (V̇_O2) in small organisms (Heusner et al., 1982; Hoegh-Guldberg and Manahan, 1995; Tartes et al., 1999; Werthessen, 1937). In coulometric respirometry, O₂ consumed by an organism is replaced by electrolytic O₂ production, with the amount of O₂ generated calculated from the charge used for electrolysis (Fig. 1A). In operation, a coulometric respirometer consists of an airtight chamber that contains CO₂-absorbent medium, so that pressure in the chamber drops as O₂ is consumed and CO₂ is produced (and removed). When pressure decreases below a threshold, current is applied across an electrolyte solution to produce O₂, thereby returning chamber pressure to the starting level. The charge used to generate the O₂ can be measured precisely, allowing calculation of the number of moles of O₂ (and therefore the volume) produced. This method is very sensitive and maintains stable O₂ and low CO₂.

Despite its advantages and long history, coulometric respirometry has not been adopted widely among physiologists, possibly because of technical challenges. Chamber pressure and current through the O₂ generator must be measured with high accuracy and precision, the chamber must be gastight, and temperature and pressure must remain highly stable throughout recording. Therefore, the construction of coulometric respirometers has been complex, using multiple chambers for pressure compensation, liquid-filled manometers as pressure sensors, and precisely calibrated capacitors to monitor charge transfer (e.g. Hoegh-Guldberg and Manahan, 1995; Tartes et al., 1999).

Miniature electronic sensors provide new resources for sensing and recording environmental variables. Here, we describe the design and construction of a coulometric respirometer based on a small, inexpensive Bosch BME 280 environmental sensor mounted inside a standard glass stopper that can be connected to a wide range of glass chambers. An Arduino™ microcontroller monitors the sensor, serves as a current source for the O₂ generator, and relays environmental and current data to a computer. The modular design greatly simplifies the construction and use of the respirometer and provides simultaneous data regarding pressure, temperature and humidity inside the chamber.

The microrespirometer was used to measure V̇_O2 in larvae and adults of the mealworm beetle Tenebrio molitor. V̇_O2 of the two stages could be distinguished from one another, and the values for both stages matched those published previously using independent methods. The device was also able to measure the effect of temperature on V̇_O2 of adult T. molitor. Because the equipment is inexpensive, portable, measures multiple environmental variables and can be assembled from inexpensive off-the-shelf components, it is a significant improvement over previous designs, and should lead to more widespread adoption of this versatile and sensitive technique.

Tenebrio molitor L. were obtained locally (Tropical Lagoon Aquarium, Silver Spring, MD, USA), and reared on rolled oats, wheat germ and vegetable scraps at 22°C. In the absence of preliminary data regarding population variation, we used sample sizes from previous studies (Bjøge et al., 2018: n=6 groups of 2 larvae; Hansen et al., 2004: n=8; Odell, 1998: n=6) to establish a sample size of 12 animals per treatment. Beetles and larvae were removed from the culture, weighed, and placed immediately in the recording chamber. Adults weighed 96.2±2 mg (mean±s.e.m., n=61), with no difference between any of the experimental groups (one-way ANOVA, F_4,56=0.1526, P=0.961). Large (121±4 mg) late-instar larvae were selected for larval experiments.

Materials, sources, and costs of the major components of the coulometric microrespirometer are provided in Table S1.

The body of the chamber used for most experiments consisted of a glass tube 22 mm in outer diameter by 100 mm length, with a fitting for a 19/22 stopper (Precision Glassblowing, Centennial, CO, USA). Because the pressure changes are small, and the experiments require hours of recording, the chamber needed to be gastight. Plastics are permeable to O₂ at a magnitude that would confound measurement of V̇_O2 of a single insect (Norton, 1957). Glass is impermeable to O₂, and its transparency allows visual monitoring of the subject.

Chambers were sealed with 19/22 glass stoppers with fittings for 6–7 mm thermometers (Synthware A521922; Fisher Scientific; Fig. 1B), providing a path for wires into the chamber. Six-conductor cable was sealed with epoxy (Loctite E-30CL, Henkel Corporation, Rocky Hill, CT, USA) into a short (∼5 cm) section of 6 mm o.d., 5 mm i.d. borosilicate glass tubing (Fisher Scientific), which formed an airtight seal with the thermometer fitting. Inside the stopper, four of the wires terminated in a four-pin 2.54 mm pitch socket into which the sensor (Bosch BME 280, DIYmall, Guangdong, China) was connected via a right-angle header.

The BME 280 provided data regarding temperature (accuracy 0.5°C; resolution ±0.01°C), pressure (accuracy ±1.0 hPa; resolution 0.18 hPa), and humidity (accuracy 3% relative humidity, RH; resolution 0.008% RH). When acquiring all three channels with 1× oversampling, response time is <8 ms.

The wires exiting the chamber terminated in a waterproof six-pin circular connector (Switchcraft Inc., Chicago, IL, USA), which connected to the controller via a six-conductor cable.

The controller was based on an Arduino Nano Every (Fig. 1D; Arduino LLC, Wilmington, DE, USA), housed in a plastic relay box (Bud Industries, Willoughby, OH, USA). The Arduino was programmed to read the pressure, temperature and humidity outputs of the BME 280 via an I²C bus, which also controlled an OLED display in the controller box (code is provided in Supplementary Materials and Methods).

The Arduino was programmed to maintain pressure between 1015 and 1017 hPa, slightly above the ambient barometric pressure (985–1015 hPa). A 5 V output was triggered when pressure dropped below a pre-set value (ON threshold 1015 hPa) and was turned off when pressure reached a second pre-set value (OFF threshold, 1017 hPa). Thus, at the beginning of the experiment, the O₂ generator was activated to pressurize the chamber to 1017 hPa (Fig. 1E). When enough O₂ was consumed by the beetle to decrease pressure below the ON threshold, the Arduino provided current to the O₂ generator and pressure increased until it reached the OFF threshold, thus maintaining pressure and P_O2 within a 2 hPa range. Pressure was maintained at a slightly hyperbaric level (≤20 hPa above ambient) so that the same setpoints could be used across experiments, and to facilitate detection of leaks. Current output to the O₂ generator was gated by a toggle switch on the controller to prevent premature activation of the O₂ generator, and depletion of CuSO₄, during setup.

The current through the O₂ generator was monitored by one of the analog to digital converters (ADCs) built into the Arduino, across a 10 Ω sense resistor (R_sense). The value of R_sense determines the voltage signal to the ADC, so accurate determination of R_sense is critical for accurate measurement of current through the generator. Current through the CuSO₄ solution was determined by the voltage across the two electrodes and the sum of the resistance of the CuSO₄ solution (∼50 Ω) and R_sense. The combined resistance across the generator and sense resistor was constrained by the requirement for a minimum 2 V potential to sustain electrolysis.

The Arduino sent comma-delimited data (identity of the sensor, time in ms, chamber pressure, chamber temperature, chamber humidity, and current through the O₂ generator) to a serial port of the computer every 500 ms. The same information was displayed on an OLED display in the controller for quick reference. Input from multiple controllers was routed to the computer via a USB hub, and data were logged using puTTY (putty.org).

Most experiments were carried out under constant temperature conditions in a biological incubator (Percival 130VL, Perry, IA, USA). Fresh soda lime pellets were placed in a small, perforated centrifuge tube in the bottom of the respirometer chamber to absorb exhaled CO₂, T. molitor were weighed and added, and the O₂ generator was connected. Seals were thoroughly cleaned before each experiment, silicone grease (Dow Corning) was applied, and stoppers were inserted into the chambers. Chambers were placed into the incubator, allowed to equilibrate for 15 min with the thermometer port unsealed, and then sealed for another hour to allow the temperature and humidity in the chamber to stabilize (Fig. 1E). At this point, O₂ generators were switched on and the chambers were pressurized until they reached the OFF threshold, initiating the experiment. In a few cases, a chamber failed to pressurize because of a damaged seal, and was removed from the experiment. As described previously (Hoegh-Guldberg and Manahan, 1995), maintaining stable temperature was critical to the function of the apparatus, as even small variations in temperature (<0.15°C) resulted in detectable fluctuations in the pressure record (Fig. 1E). After 4–5 h of recording, O₂ generators were turned off, chambers were unsealed, and recording ceased.

To examine the effects of alternative configurations, larvae were tested in groups of five in a chamber consisting of a 50 ml, two-neck, round-bottom flask, with each neck having a 19/22 joint (total chamber volume ∼100 ml). For these chambers, the sensor was contained in a stopper similar to that used above, but with the incoming wires sealed into the stopper with epoxy. Pressure in the chamber was equilibrated using a stopcock in the second neck of the flask. For these experiments, temperature was controlled using a circulating water bath/chiller (Amersham Biosciences Multitemp III), with the flasks secured by ring stands and clamps in an acrylic trough (Glass Cages, Dickson, TN, USA). The tubes containing soda lime and the O₂ generator were the same as those used above.

The following values were used to calculate the expected pressure change from a single pulse from the O₂ generator using the ideal gas equation: n (number of moles of O₂ per pulse)=Q/4F; V (chamber volume)=0.0318 l, measured in multiple chambers by subtracting the mass of an empty chamber from one filled with water; T=298 K; and R (gas constant)=0.08205 l atm K⁻¹ mol⁻¹.

Data plots, regression analyses and statistics were performed in Graphpad Prism (Graphpad Software, San Diego, CA, USA). Regression lines were tested for significant difference from zero using F-tests. In all cases, data fitted the assumptions of parametric statistics using Bartlett's test, so differences between groups were tested using one-way ANOVA followed by Tukey's test for multiple comparisons. Data are reported in the text as means±s.e.m.

Once assembled, equilibrated and sealed, chambers were pressurized to 1017 hPa using the O₂ generators. Respiration of a single adult T. molitor at 25°C caused a steady, slow decrease in pressure until it reached the ON threshold (1015 hPa), when the O₂ generator was activated and pressurized the chamber to the OFF threshold (1017 hPa). Experiments lasted 4–5 h, and this cycle of slow decrease and rapid return produced a sawtooth pattern (Fig. 2A). Current through the generator ranged from 45 to 55 mA, depending on the relative positions of the electrodes, and each pulse lasted between 17 and 22 s. Because the generator produced pure O₂, and pressure remained above ambient at all times, chambers remained slightly hyperoxic for the entire experiment. Although there were small baseline fluctuations due to slight temperature variations, pressure in a control chamber that contained no beetle was stable for the duration of all experiments (Fig. 1E).

To assess the accuracy of coulometric measurement, we used the ideal gas law to calculate the expected pressure change due to hydrolysis of CuSO₄ (see Materials and Methods) and compared it with the measured change in pressure during the cycling of the respirometer. In one experiment using five chambers at 25°C, each current pulse produced 0.953±0.07 C (n=39 pulses), which should produce enough O₂ to increase pressure by 1.90×10⁻³±1.41×10⁻⁴ atm per pulse. The measured pressure change caused by each pulse was 2 hPa (=1017 hPa−1015 hPa), or 1.97×10⁻³ atm. Therefore, the value predicted from charge transfer was within 4% of the measured pressure change, showing that current across the O₂ generator accurately reflected O₂ production.

Fig. 2B shows an example of data recorded from four chambers, each containing one beetle, simultaneously. Because computers can accommodate many universal serial bus (USB) ports, the number of channels that can be recorded at any given time is limited only by the available hardware, such as chambers, controllers and physical USB connections. Fig. 2B also illustrates the variation between individuals, which results from differences in size (see below), activity level, age, feeding state, hydration state or a combination of these.

Average O₂ consumption rate (V̇_O2) for adult beetles at 25°C was 1.93±0.17 µl min⁻¹ (=0.077±0.007 µmol min⁻¹ or 48.4±3.4 µmol h⁻¹; Table S2). There was a modest but significant effect of body mass on V̇_O2 (r²=0.351; F_1,10=5.404, P<0.05; Fig. 2C). To facilitate the following comparisons with larval data and previous studies, data were normalized to body mass and are presented as mass-specific V̇_O2 in ml O₂ g⁻¹ h⁻¹.

At 25°C, V̇_O2 of adult T. molitor was 1.185±0.084 ml O₂ h⁻¹ g⁻¹ (n=12; Fig. 3A). At the same temperature, larval V̇_O2 was lower, whether recorded singly in the same apparatus as adults (0.693±0.086 ml O₂ g⁻¹ h⁻¹; n=13) or in groups of five using larger chambers in a water bath (0.531±0.081 ml O₂ g⁻¹ h⁻¹; n=5 groups). The difference between adults and both groups of larvae was significant (one-way ANOVA with Tukey post hoc test F_2,27=14.24, P<0.001), whereas the two groups of larvae did not differ significantly (P=0.553).

V̇_O2 recorded by the coulometric respirometer was within the range described previously using different methods. For example, using direct calorimetry, Acar et al. (2004) reported heat production of approximately 5.5 µJ mg⁻¹ s⁻¹ (estimated from data from Acar et al., 2004) for adult beetle Harmonia at 25°C. A standard conversion factor of 20.2 J ml O₂⁻¹ renders a value of ∼0.9 ml O₂ g⁻¹ h⁻¹, essentially indistinguishable from the value determined in the present study for adult T. molitor (1.185±0.084 ml O₂ g⁻¹ h⁻¹). Data from larvae in the present study (0.531–0.693 ml O₂ g⁻¹ h⁻¹) are similar to those reported by Bjøge et al. (2018) (∼0.4±0.2 ml O₂ g⁻¹ h⁻¹) for larval T. molitor using stop-flow respirometry. Therefore, the apparatus can measure V̇_O2 of individual T. molitor, the apparatus can distinguish the difference in V̇_O2 between two life stages, and the metabolic rates are consistent with those described previously.

As expected for ectotherms, V̇_O2 of T. molitor increased with temperature (Fig. 3B). There was a significant effect of temperature on V̇_O2 (one-way ANOVA, F_4,56=26.88, P<0.0001). Post hoc Tukey tests (Table S2) showed that V̇_O2 was significantly higher at 25 and 30°C than at 15 or 20°C (P<0.01), and V̇_O2 at 35°C was higher than that at 25 or 30°C (P<0.001). The temperature coefficient, Q₁₀, calculated from the slope of log-transformed V̇_O2 versus temperature, was 1.69 (Fig. 3B). This value falls within the broad range of acute thermal responses reported for arthropods (Seebacher et al., 2015), and demonstrates that the microrespirometer can measure changes in metabolism due to biologically meaningful manipulations.

In conclusion, we have fabricated and demonstrated the efficacy of a simple, inexpensive coulometric respirometer. With the exception of the custom glass chamber (which can be replaced by a commercially available flask), the entire apparatus can be assembled with simple tools from standard parts stocked by scientific supply companies and/or online retailers. In the present study, the lowest V̇_O2 of 1.1 µl min⁻¹ was easily detectable (Fig. 3B; Table S2), and sensitivity can be increased by using a smaller chamber or longer recording times.

The design is modular and versatile. V̇_O2 can be quantified solely by measuring current through the O₂ generator, so knowledge of the precise volume of the chamber is not needed. Therefore, the respirometer can be used with any airtight chamber that is suitable for the organism and has a standard glass joint that can accommodate the sensor assembly (Fig. 1B). Data from multiple controllers can be acquired at once, with each controller recording pressure, temperature, humidity and current simultaneously.

As shown here, temperature can be controlled by a biological incubator or water bath. With minor modifications, such as adding a secure digital (SD) memory card to the controller box to serve as a data logger, the whole system can be battery powered and left in the field for as long as adequate CuSO₄ remains in the O₂ generator. Based on the V̇_O2 of T. molitor beetles at 25°C (1.77×10⁻³ ml min⁻¹), 1 ml of saturated (1.27 mol l⁻¹) CuSO₄ will produce more than 28 ml O₂ which should last for at least 11 days. Once the chamber is pressurized, P_O2 remains slightly hyperoxic and varies by less than 0.002 atm during cycling of the O₂ generator. Therefore, experimental subjects the size of T. molitor can be maintained in the chamber for days at a time without concern about stress from hypoxia.

Given its sensitivity, versatility, ease of construction and the quality of data generated, the design presented here is a significant improvement upon previous designs and has the potential for broad application in the study of metabolism.

We thank the Universities at Shady Grove for the use of laboratory space and the environmental chamber, and students of the 2020–2021 USG Invertebrate Physiology Research Seminar for inspiration, help with the early stages of this work and some larval recordings.