Mammals regulate their brain tissue P_O2 tightly, and only small changes in brain P_O2 are required to elicit compensatory ventilation. However, unlike the flow-through cardiovascular system of vertebrates, insect tissues exchange gases through blind-ended tracheoles that may involve a more prominent role for diffusive gas exchange. We tested the effect of progressive hypoxia on ventilation and the P_O2 of the metathoracic ganglion (neural site of control of ventilation) using micro-electrodes in the American locust, Schistocerca americana. In normal air (21 kPa), P_O2 of the metathoracic ganglion was 12 kPa. The P_O2 of the ganglion dropped as air P_O2 dropped, with ventilatory responses occurring when ganglion P_O2 reached 3 kPa. Unlike vertebrates, insects tolerate relatively high resting tissue P_O2 levels, and allow tissue P_O2 to drop during hypoxia, activity and discontinuous gas exchange before activating convective or spiracular gas exchange. Tracheated animals, and possibly pancrustaceans in general, seem likely to generally experience wide spatial and temporal variation in tissue P_O2s compared to vertebrates, with important implications for physiological function and the evolution of oxygen-utilizing proteins.

Animals

We used seven female American locusts, Schistocerca americana, derived from a colony maintained in the lab for ten years by JFH at Arizona State University. Animals were reared on lettuce and wheat flour, at 22°C, with access to a light bulb to allow for thermoregulation. Adults were 3-6 weeks past the final molt, and weighed 1.5 - 2 g.

Chamber design, animal preparation, and experimental protocols

Animals were glued ventral-side up using hot glue into the chamber, which was a 1 x 2.5 x 7.5 cm space within a solid block of clear plastic. The chamber had fittings for inflow and outflow of air, and a thermistor probe (SEMI 833 ET, Hygrosens Instruments, Löffingen, Germany) to measure temperature.  The top of the chamber was formed by two plastic plates, that were held down on top of a rubber washer and the chamber with two rubber bands; these had predrilled holes for the electrodes (Fig. 1A). 

            The P_O2 electrode was inserted at a site on the ventral thorax directly over the metathoracic ganglion. The location of the metathoracic ganglion was determined by dissections on practice animals. Holes in the cuticle were drilled 10-30 min before electrode insertion with a 0.5 mm O.D. drill bit and a micro-electric drill. The cuticle puncturing was observed using a dissecting microscope through the lateral chamber wall, to ensure that the drill bit did not pass into the tissue below the cuticle. Holes for the reference electrode were made in a similar manner, slightly posterior to the hole for the P_O2 electrode.

            The reference electrode and P_O2 electrode were then inserted through the holes in the ventral thorax. Insertion of the electrodes was observed from above and the side with two Zeiss dissecting scopes. The reference electrode was inserted just enough to make electrical contact (less than 1 mm). The P_O2 electrode was inserted 2.25 mm through the cuticular hole using a vernier-driven micro-manipulator. Preliminary studies found no difference in P_O2 values measured between 1 and 3 mm insertion, which spanned the diameter of the metathoracic ganglion. After the P_O2 electrode was inserted into the ganglion, we waited 30 - 60 min to allow partial recovery from the surgery. Unlike reports for the flight muscle (13), tissue P_O2 of the locusts at any given air P_O2 was quite stable (± 1 kPa), perhaps because these locusts were ventilating continuously. Then the air P_O2 was lowered to 15, 10, 5, 2.5, 1 and 0 kPa, with ten min allowed at each P_O2. We recorded the metathoracic P_O2 when it had stabilized at a given P_O2, and then recorded ventilation rate over one min. Ventilatory movements were recorded with an infrared reflection sensor (SFH 900, Siemens Semiconductor, Neubiberg, Germany) mounted over the abdomen of the grasshopper (Fig. 1A). The sensor sent infrared light of 940 nm to the abdomen which was reflected and detected by an infrared diode on the same device. The attenuation of the light due to the changing distance of the pumping abdomen from the sensor was used to record ventilation rate. Gas mixes were creating by using two MKS 147 mass flow controllers (MKS, Matthuen, MA, United States), and compressed tanks of nitrogen and oxygen. Flows through the chamber were maintained at 200 ml min^-1.  We confirmed that P_O2 of gas mixes were within 0.1% of those desired by checking the outflow from the animal chamber with a Ametek S3A oxygen analyzer. All experiment were conducted at 22°C.

Oxygen electrodes

Single-barrel, recessed cathode, polarographic oxygen microelectrodes were constructed (Waser and Heisler, 2005). Electrode tips ranged from 10-40 μm in diameter, with the cathode recessed 50-80 μm. A nanoamp-amplifier headstage was constructed using a OPA 128 electrometer operational amplifier (Burr Brown, Tucson, AZ, United States). Using the negative input of the OP-amp and a 1 GOhm feedback resistor a voltage of 1V nA^-1 was achieved. A polarisation voltage of -800 mV was provided by a stable voltage reference and a precision buffer operational amplifier (OPA177, Burr Brown, Tucson, AZ, United States) at the positive input of the OPA 128. The oxygen-signal, the voltage difference between polarisation- and OPA 128 output voltage (usually some 100 millivolts at normoxia, corresponding to an oxygen consumption in the range of 10 pmol s^-1), was fed into a differential instrumentation amplifier (INA 114, Burr Brown, Tucson, AZ, United States). The oxygen electrodes were calibrated in a simple insect saline (100 mmol l^-1 NaCl, 5 mmol l^-1 KCl, 20 mmol l^-1 NaH₂PO₄, pH 7.1 at 22°C, bubbled either with room air (21 kPa) or pure nitrogen. Offset and amplification of the signal from the headstage was done with a custom made lab amplifier. Voltage data were digitized at a rate of 5 s^-1 and recorded on a PC equipped with an A/D card (Data Translation DT 2821) using Turbolab Software. 

Statistics

The data did not satisfy all assumptions of parametric tests, so we used a repeated measures Friedman’s test , using Prism 8, to test for significant differences among treatment groups.