Action potentials induce biomagnetic fields in Venus flytrap plants

Upon stimulation, plants elicit electrical signals that can travel within a cellular network analogous to the animal nervous system. It is well-known that in the human brain, voltage changes in certain regions result from concerted electrical activity which, in the form of action potentials (APs), travels within nerve-cell arrays. Electrophysiological techniques like electroencephalography, magnetoencephalography, and magnetic resonance imaging are used to record this activity and to diagnose disorders. In the plant kingdom, two types of electrical signals are observed: all-or-nothing APs of similar amplitudes to those seen in humans and animals, and slow-wave potentials of smaller amplitudes. Sharp APs appear restricted to unique plant species like the"sensitive plant", Mimosa pudica, and the carnivorous Venus flytrap, Dionaea muscipula. Here we ask the question, is electrical activity in the Venus flytrap accompanied by distinct magnetic signals? Using atomic optically pumped magnetometers, biomagnetism in AP-firing traps of the carnivorous plant was recorded. APs were induced by heat stimulation, and the thermal properties of ion channels underlying the AP were studied. The measured magnetic signals exhibit similar temporal behavior and shape to the fast de- and repolarization AP phases. Our findings pave the way to understanding the molecular basis of biomagnetism, which might be used to improve magnetometer-based noninvasive diagnostics of plant stress and disease.

wounding 12 , and chemicals 13 . In contrast to the three-dimensional complex electrical network of the human brain, the circuitry of a plant leaf is two-dimensional only. The bilobed trap of the Dionaea plant ( Fig. 1a,b), formed by the modified upper part of the leaf, snaps closed within a fraction of a second when touched. Three trigger hairs that serve as mechanosensors are equally spaced on each lobe. When a prey insect touches a trigger hair, an AP (Fig. 1c) is generated and travels along both trap lobes. If a second touch-induced AP is fired within 30 s, the viscoelastic energy stored in the open trap is released and the capture organ closes 14 , imprisoning the animal food stock for digestion of a nutrient-rich meal. The leaf base, or petiole, is not excitable and is electrically insulated from the trap.
Because of this, the trap can be isolated functionally intact from the plant by a cut through the petiole.
On the isolated trap, mechanical stimuli trigger APs and closure just as on the intact plant. For the comfort of electrophysiological studies, one of the trap lobes can be fixed to a support while the other is removed, without affecting the features of AP firing. It has been shown that this simplified experimental flytrap system is well-suited to study the AP under highly reproducible conditions 15 .
Other than by touch or wounding (mechanical energy), traps can be stimulated by salt loads (osmotic energy) 16 and temperature changes (thermal energy).
Since touch activation of APs can cause unwanted mechanical noise in electric and magnetic recordings, we use thermal stimulation in our experiments. The interdisciplinary work presented here encompasses two complementary sets of experiments: the temperature dependence of flytrap electrical activity was studied in a plant-physiology laboratory, while magnetometer measurements of heatstimulated traps were conducted in a magnetically shielded room. low-conductivity water pipes (xylem) and living conductive phloem. Here the vasculature was imaged by staining for the dead vascular tissue. c, Intracellular AP lasting 2 s is subdivided into six phases (numbers), as explained in the text. The depolarization peak is indicated by an asterisk; the dotted line represents 0 mV. Inset, Zoom-in on the AP, resolving the first five phases of the AP.

Heat-induced action potentials
When we heated up the support to which excised open traps were fixed, APs were elicited and the traps closed (Extended Data Fig. 1). To study the temperature dependence of heat-induced AP initiation, on one of the trap lobes we mounted a clamp equipped with a Peltier device and surfacevoltage electrode (Fig. 2a). From a resting temperature of 20°C, the trap temperature was increased monotonically to 45°C at a rate of 4°C/s (Extended Data Fig. 2). Below 30°C, no APs were observed; above 30°C, the probability of AP firing increased and was maximal (100%) above 40°C. In 60 independent experiments using 10 different traps, we recorded the temperature at which an AP was 4 first induced. When these data were plotted as temperature-dependent AP-firing probability (Fig. 2b), the curve could be well-fitted by a single Boltzmann equation characterized by a 50% AP-firing probability at 33.8°C. This behavior indicates that heat activation of the AP is based on a two-state process. The ion channels that carry the classical animal-type AP also occupy two major states: closed and open. In contrast to the animal sodium-based AP, the plant AP depolarization is operated by a calcium-activated anion channel 5 . Thus, we conclude that the temperature "switch" of the Dionaea AP is based on a calcium-dependent process. Following Ca 2+ binding, the anion-channel gates open. Our experiments indicate that at temperatures of T ≲ 34°C the cellular Ca 2+ level remains below threshold, but at T ≳ 34°C there is enough chemical energy to open a critical number of anion channels, driving the fast depolarization phase of the AP. Increasing the thermal energy input changed not only the probability for an AP to be fired, but also led to an increased AP amplitude and decreased half-depolarization time (Supplementary Information).
These facts indicate that heat-sensitive ion channels trigger and shape the AP: at higher temperatures, thermal energy input causes more closed Ca 2+ -activated anion channels to open and depolarize the membrane potential. Compared to depolarization, fast repolarization (mediated by K + channels) and transient hyperpolarization (caused by depolarization activation of outward-directed protein pumps) 5 were much less affected by temperature. The recovery time to reach the resting membrane potential was essentially insensitive to temperature changes.
Besides lowering the AP firing threshold and changing certain features of the AP, prolonged heat stimulation can induce trap lobes to enter an autonomous AP firing mode (Extended Data Fig. 1).
When increasing the bottom surface temperature of the recording-chamber base from 20 to 46°C, AP spiking activity sets in after a couple of seconds, reaching a steady AP firing frequency of 3.8 per minute at a stable 46°C surface temperature. Induction of autonomous APs has also been obtained using flytraps treated with NaCl salt (osmotic energy) 17 .

Biomagnetism
Having established heat stimulation as a reliable noninvasive technique for inducing flytrap APs, we searched for the magnetic field associated with this electrical excitability. Magnetometry experiments were carried out at Physikalisch-Technische Bundesanstalt (PTB) Berlin in the Berlin Magnetically Shielded Room 2 (BMSR-2) facility 18 , using four QuSpin Zero-Field Magnetometers (QZFM). These commercial optically pumped magnetometers (OPMs) employ a glass cell containing alkali vapor to sense changes in the local magnetic-field environment 8 . A magnetically shielded environment is required for operation of the magnetometers, and use of a walk-in shielded room allowed for the constant presence of an experimenter to prepare plant samples and carry out measurements. As shown in Fig. 3, an isolated trap lobe was attached to the housing of the primary sensor (denoted A), such that the distance between the plant sample and the center of the atomic sensing volume was approximately 7 mm. Two secondary sensors (B and C) were placed nearby the primary sensor to measure signal falloff, and an additional background sensor (D) was used to monitor the magnetic environment in the shielded room. Each magnetometer is sensitive to signals along two orthogonal axes. Sensor electronics were connected to a data-acquisition system in the PTB control room outside the magnetically shielded room. To monitor heat-induced APs, we used two silver-tipped copper surface electrodes, inserted in either end of the plant sample 19 . These data, together with other auxiliary trigger signals, were sampled simultaneously with the OPMs using the same data-acquisition system. Resistive heaters in the magnetometer housing, which are used to increase the atomic density and improve sensitivity, also served to induce autonomous AP firing via surface heat transfer. Electric and magnetic signals were recorded simultaneously from traps heated to a surface temperature of 41°C.
Prior to the measurements, we performed tests to ensure that no spurious magnetic fields were generated by the electrode system (Supplementary Information). To better distinguish the observed magnetic signals from background noise, we triggered on the electric signals and averaged the magnetic data in a time window around those trigger points. Examples of averaged magnetic data are shown in Fig. 4a,b. A clear magnetic signal with a time scale corresponding to that of the averaged electric signal is visible in the primary-sensor data. For comparison, data from several different 7 experiments were plotted (Fig. 5). To minimize common background noise, we subtracted the magnetic data of sensor D to create a gradiometer with a 48-mm baseline. Signals of up to 0.5 pT are visible in the y-axis gradiometric data, normal to the sample surface. The signal magnitude obtained was comparable to what one observes in surface measurements of nerve impulses in animals 20 .
To quantify the significance of the signals, signal-to-noise ratios (SNRs) were calculated from the average y-axis gradiometric time traces in Fig. 5   obtained using the same procedure as in a. The data from the secondary sensors, B and C, do not show a signal.
The data from the background sensor D can be used to remove noise common to all sensors (see Fig. 4). along cellular pathways. For the Venus flytrap system, it is known from electrode measurements that APs propagate through the trap at speeds of around 10 m/s 6 . A proposed pathway of long-distance signal propagation between plant cells in the trap is the electrically conductive phloem in the vasculature (Fig. 1b). Given that the typical resistance between two points on a trap is R ≈ 1 MΩ 23 , we can perform a basic calculation to confirm that the magnitude of the magnetic fields we measure is reasonable. We estimate the expected magnetic-field magnitude at the center of the sensing volume to be ≈ 2π , 1 where I = V/R ≈ 10 nA is the current passing through the trap between the electrodes, and r ≈ 7 mm is the perpendicular distance from the trap surface. Using these values, we find B ≈ 0.3 pT, a magnitude which corresponds well with the y-axis experimental results of sensor A. Although the precise distribution and directionality of current flow in the trap is unknown, we can use the geometry of the trap (Fig. 1a,b) and magnetometry setup (Fig. 3) to further interpret our results. If the x-oriented parallel-cable structure of the vasculature is the primary conduction pathway, magnetic field along the y-direction is expected at the primary sensor A, but not at the secondary sensors B and C. The symmetry of the trap about the x-direction could explain the relative lack of z-axis magnetic signal in our measurements. (Trap curvature and misalignment with respect to the sensor housing may give rise to z-axis signals in some experiments.) Thus, our magnetometry results agree with a hypothesis that the vasculature serves as a network for long-distance electromagnetic signaling within the trap.

Discussion
Although human and animal biomagnetism are well-developed areas of research 2,20,21,24,25,26 , very little analogous work has been conducted in the plant kingdom 9,12,22,27 , largely because biomagnetic signals are typically much smaller in amplitude and frequency than their animal counterparts. Previously reported detection of plant biomagnetism, which established the existence of measurable magnetic activity in the plant kingdom, was carried out using superconducting-quantum-interference-device (SQUID) magnetometers 9,12,22 . Atomic magnetometers are arguably more attractive for biological applications, since, unlike SQUIDs 28,29 , they are non-cryogenic and can be miniaturized to optimize spatial resolution of measured biological features 20,26,30 . Our study of plant biomagnetism using atomic magnetometers documents: (i) existence and features of biomagnetic signals in the Venus flytrap, (ii) magnetic detection of APs in a multicellular plant system generally, and (iii) electric and magnetic detection of heat-induced APs in the Venus flytrap. In the future, the SNR of magnetic measurements in plants will benefit from optimizing the low-frequency stability and sensitivity of atomic 10 magnetometers. Just as noninvasive magnetic techniques have become essential tools for medical diagnostics of the human brain and body, this noninvasive technique could also be useful in the future for crop-plant diagnostics-by measuring the electromagnetic response of plants facing such challenges as sudden temperature change, herbivore attack, and chemical exposure.