NeurostimML: a machine learning model for predicting neurostimulation-induced tissue damage.

Objective. The safe delivery of electrical current to neural tissue depends on many factors, yet previous methods for predicting tissue damage rely on only a few stimulation parameters. Here, we report the development of a machine learning approach that could lead to a more reliable method for predicting electrical stimulation-induced tissue damage by incorporating additional stimulation parameters.Approach. A literature search was conducted to build an initial database of tissue response information after electrical stimulation, categorized as either damaging or non-damaging. Subsequently, we used ordinal encoding and random forest for feature selection, and investigated four machine learning models for classification: Logistic Regression, K-nearest Neighbor, Random Forest, and Multilayer Perceptron. Finally, we compared the results of these models against the accuracy of the Shannon equation.Main Results. We compiled a database with 387 unique stimulation parameter combinations collected from 58 independent studies conducted over a period of 47 years, with 195 (51%) categorized as non-damaging and 190 (49%) categorized as damaging. The features selected for building our model with a Random Forest algorithm were: waveform shape, geometric surface area, pulse width, frequency, pulse amplitude, charge per phase, charge density, current density, duty cycle, daily stimulation duration, daily number of pulses delivered, and daily accumulated charge. The Shannon equation yielded an accuracy of 63.9% using akvalue of 1.79. In contrast, the Random Forest algorithm was able to robustly predict whether a set of stimulation parameters was classified as damaging or non-damaging with an accuracy of 88.3%.Significance. This novel Random Forest model can facilitate more informed decision making in the selection of neuromodulation parameters for both research studies and clinical practice. This study represents the first approach to use machine learning in the prediction of stimulation-induced neural tissue damage, and lays the groundwork for neurostimulation driven by machine learning models.

NeurostimML: A machine learning model for predicting neurostimulation-induced tissue damage.

Objective: The safe delivery of electrical current to neural tissue depends on many factors, yet previous methods for predicting tissue damage rely on only a few stimulation parameters. Here, we report the development of a machine learning approach that could lead to a more reliable method for predicting electrical stimulation-induced tissue damage by incorporating additional stimulation parameters. Approach: A literature search was conducted to build an initial database of tissue response information after electrical stimulation, categorized as either damaging or non-damaging. Subsequently, we used ordinal encoding and random forest for feature selection, and investigated four machine learning models for classification: Logistic Regression, K-nearest Neighbor, Random Forest, and Multilayer Perceptron. Finally, we compared the results of these models against the accuracy of the Shannon equation. Main Results: We compiled a database with 387 unique stimulation parameter combinations collected from 58 independent studies conducted over a period of 47 years, with 195 (51%) categorized as non-damaging and 190 (49%) categorized as damaging. The features selected for building our model with a Random Forest algorithm were: waveform shape, geometric surface area, pulse width, frequency, pulse amplitude, charge per phase, charge density, current density, duty cycle, daily stimulation duration, daily number of pulses delivered, and daily accumulated charge. The Shannon equation yielded an accuracy of 63.9% using a k value of 1.79. In contrast, the Random Forest algorithm was able to robustly predict whether a set of stimulation parameters was classified as damaging or non-damaging with an accuracy of 88.3%. Significance: This novel Random Forest model can facilitate more informed decision making in the selection of neuromodulation parameters for both research studies and clinical practice. This study represents the first approach to use machine learning in the prediction of stimulation-induced neural tissue damage, and lays the groundwork for neurostimulation driven by machine learning models.

Advanced automated treatment planning for total body irradiation: Implementation and effects on standardization.

PURPOSE: This work describes the automation of our volumetric modulated arc therapy (VMAT) total body irradiation (TBI) treatment planning. It also aims to determine if plan standardization is impacted by automation. METHODS: We introduced automated beam placement for TBI in March 2021. For manual beam placement pre-2021, Python-modified DICOM files were imported to pre-set cumulative meterset weights, with other parameters selected by dosimetrists. Our automated planning script automates these processes and sets gantry stop angles and isocentre placement. To determine the impact of automation on plan standardization, we performed a retrospective review of a matched cohort of 168 patients. Plan parameters were compared with an external standard, and passing rates compared between patient cohorts. The dosimetric impact was investigated by comparing a Body-5 mm homogeneity index (HI = D2%/D98%) and mean lung dose (MLD) between cohorts. RESULTS: Results are listed for manual and automated groups respectively. Median (range) passing rates were 97.7% (96.1-100) and 99.2% (98.3-100). Automated plans had a significantly higher passing rate (p â‰ª 0.05) and smaller variance (p = 0.001). Most failures were attributed to human error. Automated plans also had more consistent parameter identifiers. After considering dimensional outliers, median (range) Body-5 mm HI were 1.18 (1.14-1.23) and 1.18 (1.15-1.26), and mean ± standard deviation MLD were 103.8 ± 1.3% and 104.1 ± 0.9%. Variances were not significantly different between Body-5 mm HI (p = 0.092) but were for MLD (p = 0.013). CONCLUSIONS: Implementation of automated planning in TBI resulted in significantly improved plan standardization. The decrease in variance of the MLD for the automated planning group points towards a potential dosimetric benefit of automation.

Chronic stability of activated iridium oxide film voltage transients from wireless floating microelectrode arrays.

Successful monitoring of the condition of stimulation electrodes is critical for maintaining chronic device performance for neural stimulation. As part of pre-clinical safety testing in preparation for a visual prostheses clinical trial, we evaluated the stability of the implantable devices and stimulation electrodes using a combination of current pulsing in saline and in canine visual cortex. Specifically, in saline we monitored the stability and performance of 3000 µm2 geometric surface area activated iridium oxide film (AIROF) electrodes within a wireless floating microelectrode array (WFMA) by measuring the voltage transient (VT) response through reverse telemetry. Eight WFMAs were assessed in vitro for 24 days, where n = 4 were pulsed continuously at 80 µA (16 nC/phase) and n = 4 remained in solution with no applied stimulation. Subsequently, twelve different WFMAs were implanted in visual cortex in n = 3 canine subjects (4 WFMAs each). After a 2-week recovery period, half of the electrodes in each of the twelve devices were pulsed continuously for 24 h at either 20, 40, 63, or 80 µA (200 µs pulse width, 100 Hz). VTs were recorded to track changes in the electrodes at set time intervals in both the saline and in vivo study. The VT response of AIROF electrodes remained stable during pulsing in saline over 24 days. Electrode polarization and driving voltage changed by less than 200 mV on average. The AIROF electrodes also maintained consistent performance, overall, during 24 h of pulsing in vivo. Four of the in vivo WFMA devices showed a change in polarization, access voltage, or driving voltage over time. However, no VT recordings indicated electrode failure, and the same trend was typically seen in both pulsed and unpulsed electrodes within the same device. Overall, accelerated stimulation testing in saline and in vivo indicated that AIROF electrodes in the WFMA were able to consistently deliver up to 16 nC per pulse and would be suitable for chronic clinical use.

Orientation dependence of inhomogeneous magnetization transfer and dipolar order relaxation rate in phospholipid bilayers.

Inhomogeneous magnetization transfer (ihMT) is a novel MRI technique used to measure white matter myelination in the brain and spinal cord. In the brain, ihMT has a strong orientation dependence which is likely to arise from the anisotropy of dipolar couplings between protons on oriented lipids in the myelin bilayers. We measured the orientation dependence of the second moment (M2) of the lineshape, dipolar order relaxation rate (R1D), and ihMT ratio (ihMTR) in an oriented phospholipid bilayer at 9.4 T. We found a strong orientation dependence in all three parameters. ihMTR and R1D were maximized when the bilayers were aligned perpendicular to B0 and minimized near the magic angle (∼54.7°). M2 followed an orientation dependence given by the second Legendre polynomial squared as predicted by the form of the secular dipolar Hamiltonian. These results were used to calculate the orientation dependence of R1D and ihMTR in a diffusionless myelin sheath model, which showed ihMTR was maximised for fibers perpendicular to B0 and minimised at 45°, similar to ex-vivo spinal cord with a larger prepulse frequency offset, but in contrast to in vivo brain findings. Adding fiber dispersion to this model smoothed the orientation dependence curve as expected. Our results suggest the importance of the effects of lipid diffusion and prepulse offset frequency on ihMTR.

Wireless transmission of voltage transients from a chronically implanted neural stimulation device.

Objective.Consistent transmission of data from wireless neural devices is critical for monitoring the condition and performance of stimulation electrodes. To date, no wireless neural interface has demonstrated the ability to monitor the integrity of chronically implanted electrodes through wireless data transmission.Approach.In this work, we present a method for wirelessly recording the voltage transient (VT) response to constant-current, cathodic-first asymmetric pulsing from a microelectrode array. We implanted six wireless devices in rat sciatic nerve and wirelessly recorded VT measurements throughout a 38 week implantation period.Main results.Electrode maximum cathodic potential excursion (Emc), access voltage, and driving voltage (extracted from each VT) remained stable throughout the 38 week study period. Average Emc(from an applied +0.6 V interpulse bias) in response to 4.7µA, 200.2µs pulses was 267 ± 107 mV at week 1 post-implantation and 282 ± 52 mV at week 38 post-implantation. Access voltage for the same 4.7µA pulsing amplitude was 239 ± 65 mV at week 1 post-implantation and 268 ± 139 mV at week 38 post-implantation.Significance.The VT response recorded via reverse telemetry from the wireless microelectrode array did not significantly change over a 38 week implantation period and was similar to previously reported VTs from wired microelectrodes with the same geometry. Additionally, the VT response recorded wirelessly in phosphate buffered saline before and after device implantation appeared as expected, showing significantly less electrode polarization and smaller access voltage than the VT responsein vivo.

Effects of Varying Pulse Width and Frequency of Wireless Stimulation in Rat Sciatic Nerve.

Peripheral nerve stimulation is a commonly used method for assisting movements after spinal cord injury, stroke, traumatic brain injury, and other types of neurological damage or dysfunction. There are many different patterns of electrical stimulation used to accomplish movement. And so, our study investigated stimulation with a wireless floating microelectrode array (WFMA) in comparison to previously reported data on functional electrical stimulation. To determine the effect on hindlimb movement, we tested a range of frequencies and pulse widths using WFMAs that were implanted in the rat sciatic nerve for 38 weeks. Frequencies between 1 and 50 Hz did not change the minimum current amplitude required to elicit movement in the hindlimb. Increasing pulse width from 57.2 to 400.4 µs decreased the minimum current required but had an associated increase in total charge applied per pulse. Overall, the WFMA provides a stable wireless peripheral nerve interface suitable for functional electrical stimulation.Clinical Relevance- This work establishes the efficacy of various stimulation parameters for controlling movement with a wireless peripheral nerve stimulator.

The Effect of Implanted Functional Electrical Stimulation on Gait Performance in Stroke Survivors: A Systematic Review.

The emerging literature suggests that implantable functional electrical stimulation may improve gait performance in stroke survivors. However, there is no review providing the possible therapeutic effects of implanted functional electrical stimulation on gait performance in stroke survivors. We performed a web-based, systematic paper search using PubMed, the Cochrane Library, and EMBASE. We limited the search results to human subjects and papers published in peer-reviewed journals in English. We did not restrict demographic or clinical characteristics. We included 10 papers in the current systematic review. Across all included studies, we found preliminary evidence of the potential therapeutic effects of functional electrical stimulation on walking endurance, walking speed, ankle mobility, and push-off force in stroke survivors. However, due to the heterogeneity between the included studies, small sample size, and lack of randomized controlled trials, more studies are critically needed to confirm whether implanted functional electrical stimulation can improve gait performance in stroke survivors.

Wireless microelectrode arrays for selective and chronically stable peripheral nerve stimulation for hindlimb movement.

Objective. Maximizing the stability of implanted neural interfaces will be critical to developing effective treatments for neurological and neuromuscular disorders. Our research aims to develop a stable neural interface using wireless communication and intrafascicular microelectrodes to provide highly selective stimulation of neural tissue.Approach. We implanted a wireless floating microelectrode array into the left sciatic nerve of six rats. Over a 38 week implantation period, we recorded stimulation thresholds and movements evoked at each implanted electrode. We also tracked each animal's response to sensory stimuli and performance on two different walking tasks.Main results. Presence of the microelectrode array inside the sciatic nerve did not cause any obvious motor or sensory deficits in the hindlimb. Visible movement in the hindlimb was evoked by stimulating the sciatic nerve with currents as low as 4.1µA. Thresholds for most of the 96 electrodes we implanted were below 20µA, and predictable recruitment of plantar flexion and dorsiflexion was achieved by stimulating rat sciatic nerve with the intrafascicular microelectrode array. Further, motor recruitment patterns for each electrode did not change significantly throughout the study.Significance. Incorporating wireless communication and a low-profile neural interface facilitated highly stable motor recruitment thresholds and fine motor control in the hindlimb throughout an extensive 9.5 month assessment in rodent peripheral nerve. Results of this study indicate that use of the wireless device tested here could be extended to other applications requiring selective neural stimulation and chronic implantation.

Selective Wireless Stimulation of Rat Sciatic Nerve.

Chronic stability of functional performance is a significant challenge to the success of implantable devices for neural stimulation and recording. Integrating wireless technology with typical microelectrode array designs is one approach that may reduce instances of mechanical failure and improve the long-term performance of neural devices. We have investigated the long-term stability of Wireless Floating Microelectrode Arrays (WMFAs) implanted in rat sciatic nerve, and their ability to selectively recruit muscles in the hind limb via neural stimulation. Thresholds as low as 4.1 µA were able to generate visible motion of the rear paw. Each implanted device (n=6) was able to selectively recruit plantar flexion and dorsiflexion of the rear paw, and selective stimulation of both movements was achieved throughout the study period. The evoked limb motion was electrode specific and was dependent on location within the fascicular structure of the nerve. Motor thresholds and movement patterns remained stable for more than 8 weeks after device implantation. No major changes in limb function were observed between the implanted and contralateral limb, or between implanted animals and control group animals. The results of this study show that WFMAs with intrafascicular electrodes implanted in a healthy peripheral nerve can provide stable and selective motor recruitment, without altering overall limb function.

Evaluating Microelectrode Arrays in Peripheral Nerve Using Micro Computed Tomography.

Many advances have been made with imaging of implanted neural devices; however, the ability to image whole nerve samples remains limited. Further, few imaging modalities are well suited for visualizing both whole devices in vivo and individual microelectrodes within a nerve. In this study, we used micro-computed tomography (micro-CT) to evaluate Wireless Floating Microelectrode Arrays (WMFAs) implanted in rat sciatic nerve at the level of whole devices and individual electrodes. WFMAs were also used to track selective recruitment of plantar flexion and dorsiflexion of the rear paw, which was achieved by each implanted device (n=6) during chronic implantation. Evoked limb motion was correlated to end-of-study assessments using micro-CT to visualize electrode locations within the fascicular structure of the sciatic nerve. Results of this study show that micro-CT imaging can provide valuable assessments of microelectrode arrays implanted in peripheral nerves for both whole devices visualized in vivo and individual electrodes visualized in whole nerve tissue samples.Clinical relevance- This work informs the use of micro-computed tomography as a tool for correlating neural device performance with physical attributes of the implant location.

Activated iridium oxide film (AIROF) electrodes for neural tissue stimulation.

OBJECTIVE: Iridium oxide films are commonly used as a high charge-injection electrode material in neural devices. Yet, few studies have performed in-depth assessments of material performance versus film thickness, especially for films grown on three-dimensional (instead of planar) metal surfaces in neutral pH electrolyte solutions. Further, few studies have investigated the driving voltage requirements for constant-current stimulation using activated iridium oxide (AIROF) electrodes, which will be a key constraint for future use in wirelessly powered neural devices. APPROACH: In this study, iridium microwire probes were activated by repeated potential pulsing in room temperature phosphate buffered saline (pH 7.1-7.3). Electrochemical measurements were recorded in three different electrolyte conditions for probes with different geometric surface areas (GSAs) as the AIROF thickness was increased. MAIN RESULTS: Maintaining an anodic potential bias during the inter-pulse interval was required for AIROF electrodes to deliver charge levels considered necessary for neural stimulation. Potential pulsing for 100-200 cycles was sufficient to achieve charge injection levels of 2.5 mC cm-2 (50 nC/phase in a biphasic pulse) in PBS with 2000 µm2 iridium probes. Increasing the electrode surface area to 3000 µm2 and 4000 µm2 significantly increased charge-injection capacity, reduced the driving voltage required to deliver a fixed amount of charge, and reduced polarization of the electrodes during constant-current pulsing. SIGNIFICANCE: This study establishes methods for choosing an activation protocol and a desired GSA for three-dimensional iridium electrodes suitable for neural tissue insertion and stimulation, and provides guidelines for evaluating electrochemical performance of AIROF using model saline solutions.

Standardized flattening filter free volumetric modulated arc therapy plans based on anteroposterior width for total body irradiation.

In this work, the feasibility of using flattening filter free (FFF) beams in volumetric modulated arc therapy (VMAT) total body irradiation (TBI) treatment planning to decrease protracted beam-on times for these treatments was investigated. In addition, a methodology was developed to generate standardized VMAT TBI treatment plans based on patient physical dimensions to eliminate plan optimization time. A planning study cohort of 47 TBI patients previously treated with optimized VMAT ARC 6 MV beams was retrospectively examined. These patients were sorted into six categories depending on height and anteroposterior (AP) width at the umbilicus. Using Varian Eclipse, clinical 40 cm × 10 cm open field arcs were substituted with 6 MV FFF. Mid-plane lateral dose profiles in conjunction with relative arc output factors (RAOF) yielded how far a given multileaf collimator (MLC) leaf must move in order to achieve a mid-plane 100% isodose for a specific control point. Linear interpolation gave the dynamic MLC aperture for the entire arc for each patient AP width category, which was subsequently applied through Python scripting. All FFF VMAT TBI plans were then evaluated by two radiation oncologists and deemed clinically acceptable. The FFF and clinical VMAT TBI plans had similar Body-5 mm D98% distributions, but overall the FFF plans had statistically significantly increased or broader Body-5 mm D2% and mean lung dose distributions. These differences are not considered clinically significant. Median beam-on times for the FFF and clinical VMAT TBI plans were 11.07 and 18.06 min, respectively, and planning time for the FFF VMAT TBI plans was reduced by 34.1 min. In conclusion, use of FFF beams in VMAT TBI treatment planning resulted in dose homogeneity similar to our current VMAT TBI technique. Clinical dosimetric criteria were achieved for a majority of patients while planning and calculated beam-on times were reduced, offering the possibility of improved patient experience.

Extended SSD VMAT treatment for total body irradiation.

In this work, we develop a total body irradiation technique that utilizes arc delivery, a buildup spoiler, and inverse optimized multileaf collimator (MLC) motion to shield organs at risk. The current treatment beam model is verified to confirm its applicability at extended source-to-surface distance (SSD). The delivery involves 7-8 volumetric modulated arc therapy arcs delivered to the patient in the supine and prone positions. The patient is positioned at a 90° couch angle on a custom bed with a 1 cm acrylic spoiler to increase surface dose. Single-step optimization using a patient CT scan provides enhanced dose homogeneity and limits organ at risk dose. Dosimetric data of 109 TBI patients treated with this technique is presented along with the clinical workflow. Treatment planning system (TPS) verification measurements were performed at an extended SSD of 175 cm. Measurements included: a 4-point absolute depth-dose curve, profiles at 1.5, 5, and 10 cm depth, absolute point-dose measurements of an treatment field, 2D Gafchromic® films at four locations, and measurements of surface dose at multiple locations of a Alderson phantom. The results of the patient DVH parameters were: Body-5 mm D98 95.3 ± 1.5%, Body-5 mm D2 114.0 ± 3.6%, MLD 102.8 ± 2.1%. Differences between measured and calculated absolute depth-dose values were all <2%. Profiles at extended SSD had a maximum point difference of 1.3%. Gamma pass rates of 2D films were greater than 90% at 5%/1 mm. Surface dose measurements with film confirmed surface dose values of >90% of the prescription dose. In conclusion, the inverse optimized delivery method presented in the paper has been used to deliver homogenous dose to over 100 patients. The method provides superior patient comfort utilizing a commercial TPS. In addition, the ability to easily shield organs at risk is available through the use of MLCs.

Drosophila eggshell production: identification of new genes and coordination by Pxt.

Drosophila ovarian follicles complete development using a spatially and temporally controlled maturation process in which they resume meiosis and secrete a multi-layered, protective eggshell before undergoing arrest and/or ovulation. Microarray analysis revealed more than 150 genes that are expressed in a stage-specific manner during the last 24 hours of follicle development. These include all 30 previously known eggshell genes, as well as 19 new candidate chorion genes and 100 other genes likely to participate in maturation. Mutations in pxt, encoding a putative Drosophila cyclooxygenase, cause many transcripts to begin expression prematurely, and are associated with eggshell defects. Somatic activity of Pxt is required, as RNAi knockdown of pxt in the follicle cells recapitulates both the temporal expression and eggshell defects. One of the temporally regulated genes, cyp18a1, which encodes a cytochromome P450 protein mediating ecdysone turnover, is downregulated in pxt mutant follicles, and cyp18a1 mutation itself alters eggshell gene expression. These studies further define the molecular program of Drosophila follicle maturation and support the idea that it is coordinated by lipid and steroid hormonal signals.

Multiple pathways influence mitochondrial inheritance in budding yeast.

Yeast mitochondria form a branched tubular network. Mitochondrial inheritance is tightly coupled with bud emergence, ensuring that daughter cells receive mitochondria from mother cells during division. Proteins reported to influence mitochondrial inheritance include the mitochondrial rho (Miro) GTPase Gem1p, Mmr1p, and Ypt11p. A synthetic genetic array (SGA) screen revealed interactions between gem1Delta and deletions of genes that affect mitochondrial function or inheritance, including mmr1Delta. Synthetic sickness of gem1Delta mmr1Delta double mutants correlated with defective mitochondrial inheritance by large buds. Additional studies demonstrated that GEM1, MMR1, and YPT11 each contribute to mitochondrial inheritance. Mitochondrial accumulation in buds caused by overexpression of either Mmr1p or Ypt11p did not depend on Gem1p, indicating these three proteins function independently. Physical linkage of mitochondria with the endoplasmic reticulum (ER) has led to speculation that distribution of these two organelles is coordinated. We show that yeast mitochondrial inheritance is not required for inheritance or spreading of cortical ER in the bud. Moreover, Ypt11p overexpression, but not Mmr1p overexpression, caused ER accumulation in the bud, revealing a potential role for Ypt11p in ER distribution. This study demonstrates that multiple pathways influence mitochondrial inheritance in yeast and that Miro GTPases have conserved roles in mitochondrial distribution.

Moving mitochondria: establishing distribution of an essential organelle.

Mitochondria form a dynamic network responsible for energy production, calcium homeostasis and cell signaling. Appropriate distribution of the mitochondrial network contributes to organelle function and is essential for cell survival. Highly polarized cells, including neurons and budding yeast, are particularly sensitive to defects in mitochondrial movement and have emerged as model systems for studying mechanisms that regulate organelle distribution. Mitochondria in multicellular eukaryotes move along microtubule tracks. Actin, the primary cytoskeletal component used for transport in yeast, has more subtle functions in other organisms. Kinesin, dynein and myosin isoforms drive motor-based movement along cytoskeletal tracks. Milton and syntabulin have recently been identified as potential organelle-specific adaptor molecules for microtubule-based motors. Miro, a conserved GTPase, may function with Milton to regulate transport. In yeast, Mmr1p and Ypt11p, a Rab GTPase, are implicated in myosin V-based mitochondrial movement. These potential adaptors could regulate motor activity and therefore determine individual organelle movements. Anchoring of stationary mitochondria also contributes to organelle retention at specific sites in the cell. Together, movement and anchoring ultimately determine mitochondrial distribution throughout the cell.

Yeast Miro GTPase, Gem1p, regulates mitochondrial morphology via a novel pathway.

Cell signaling events elicit changes in mitochondrial shape and activity. However, few mitochondrial proteins that interact with signaling pathways have been identified. Candidates include the conserved mitochondrial Rho (Miro) family of proteins, which contain two GTPase domains flanking a pair of calcium-binding EF-hand motifs. We show that Gem1p (yeast Miro; encoded by YAL048C) is a tail-anchored outer mitochondrial membrane protein. Cells lacking Gem1p contain collapsed, globular, or grape-like mitochondria. We demonstrate that Gem1p is not an essential component of characterized pathways that regulate mitochondrial dynamics. Genetic studies indicate both GTPase domains and EF-hand motifs, which are exposed to the cytoplasm, are required for Gem1p function. Although overexpression of a mutant human Miro protein caused increased apoptotic activity in cultured cells (Fransson et al., 2003. J. Biol. Chem. 278:6495-6502), Gem1p is not required for pheromone-induced yeast cell death. Thus, Gem1p defines a novel mitochondrial morphology pathway which may integrate cell signaling events with mitochondrial dynamics.

BYPASS1 negatively regulates a root-derived signal that controls plant architecture.

Plant architecture is regulated by endogenous developmental programs, but it can also be strongly influenced by cues derived from the environment. For example, rhizosphere conditions such as water and nutrient availability affect shoot and root architecture; this implicates the root as a source of signals that can override endogenous developmental programs. Cytokinin, abscisic acid, and carotenoid derivatives have all been implicated as long-distance signals that can be derived from the root. However, little is known about how root-derived signaling pathways are regulated. Here, we show that BYPASS1 (BPS1), an Arabidopsis gene of unknown function, is required to prevent constitutive production of a root-derived graft-transmissible signal that is sufficient to inhibit leaf initiation, leaf expansion, and shoot apical meristem activity. We show that this root-derived signal is likely to be a novel carotenoid-derived molecule that can modulate both root and shoot architecture.