Measuring Glucose Uptake in Drosophila Models of TDP-43 Proteinopathy

Summary

Glucose uptake is increased in Drosophila motor neurons affected by TAR DNA binding protein (TDP-43) proteinopathy, as indicated by a FRET-based, genetically encoded glucose sensor.

Abstract

Amyotrophic lateral sclerosis is a neurodegenerative disorder causing progressive muscle weakness and death within 2-5 years following diagnosis. Clinical manifestations include weight loss, dyslipidemia, and hypermetabolism; however, it remains unclear how these relate to motor neuron degeneration. Using a Drosophila model of TDP-43 proteinopathy that recapitulates several features of ALS including cytoplasmic inclusions, locomotor dysfunction, and reduced lifespan, we recently identified broad ranging metabolic deficits. Among these, glycolysis was found to be upregulated and genetic interaction experiments provided evidence for a compensatory neuroprotective mechanism. Indeed, despite upregulation of phosphofructokinase, the rate limiting enzyme in glycolysis, an increase in glycolysis using dietary and genetic manipulations was shown to mitigate locomotor dysfunction and increased lifespan in fly models of TDP-43 proteinopathy. To further investigate the effect on TDP-43 proteinopathy on glycolytic flux in motor neurons, a previously reported genetically encoded, FRET-based sensor, FLII12Pglu-700µδ6, was used. This sensor is comprised of a bacterial glucose-sensing domain and cyan and yellow fluorescent proteins as the FRET pair. Upon glucose binding, the sensor undergoes a conformational change allowing FRET to occur. Using FLII12Pglu-700µδ6, glucose uptake was found to be significantly increased in motor neurons expressing TDP-43^G298S, an ALS causing variant. Here, we show how to measure glucose uptake, ex vivo, in larval ventral nerve cord preparations expressing the glucose sensor FLII12Pglu-700µδ6 in the context of TDP-43 proteinopathy. This approach can be used to measure glucose uptake and assess glycolytic flux in different cell types or in the context of various mutations causing ALS and related neurodegenerative disorders.

Introduction

Amyotrophic Lateral Sclerosis (ALS) is a progressive neurodegenerative disorder that is currently incurable. ALS affects upper and lower motor neurons leading to loss of motor coordination, irreversible paralysis, respiratory failure, and eventual death within 2-5 years of diagnosis¹. ALS is associated with metabolic defects such as weight loss, dyslipidemia, and hypermetabolism (reviewed in²); however, it remains unclear how these alterations in metabolism relate to motor neuron degeneration. A common denominator in ALS and related neurodegenerative diseases is TDP-43, a nucleic acid binding protein involved in several steps of RNA processing^3,4,5. Although mutations in TDP-43 only affect 3%-5% of patients, wild-type TDP-43 protein is found within cytoplasmic aggregates in >97% of ALS cases (reviewed in⁶). This pathology was modeled in Drosophila by overexpression of human wildtype or mutant TDP-43 (G298S) in motor neurons, which recapitulates multiple aspects of ALS, including cytoplasmic inclusions, locomotor dysfunction and reduced lifespan^7,8. Using these models, it was recently reported that TDP-43 proteinopathy causes a significant increase in pyruvate levels and phosphofructokinase (PFK) mRNA, the rate limiting enzyme of glycolysis⁹. Similar increases in PFK transcripts were found in patient-derived motor neurons and spinal cords, suggesting that glycolysis is upregulated in the context of TDP-43 proteinopathy. Interestingly, further increase in glycolysis using dietary and genetic manipulations mitigated several ALS phenotypes such as locomotor dysfunction and increased lifespan in fly models of TDP-43 proteinopathy, consistent with a compensatory, neuroprotective mechanism in degenerating motor neurons.

To further probe changes in glycolysis and measure glucose uptake in Drosophila models of TDP-43 proteinopathy, a previously reported genetically encoded FRET-based sensor FLII12Pglu-700µδ6¹⁰ was expressed in motor neurons specifically using the UAS-GAL4 expression system. The FLII12Pglu-700µδ6 glucose sensor uses resonance energy transfer between two variants of green fluorescent protein, cyan and yellow fluorescent proteins (CFP and YFP) to detect glucose at the cellular level. It consists of a bacterial glucose binding domain from the E. coli MglB gene fused to CFP and YFP at opposite ends of the molecule. When bound to a glucose molecule, the sensor undergoes a conformational change bringing CFP and YFP closer together and allowing FRET to occur, which can then be used to quantify intracellular glucose levels^10,11,12 (Figure 1). Here, we show how the FLII12Pglu-700µδ6 sensor can be used to determine changes in glucose uptake caused by TDP-43 proteinopathy in motor neurons. The experiments described here show that overexpression of an ALS-associated mutant, TDP-43^G298S, in motor neurons causes a significant increase in glucose uptake compared to controls. This approach can be used in other types of ALS (e.g., SOD1, C9orf72, etc.) and/or other cell types (e.g., glia, muscles) to determine changes in glucose uptake associated with neurodegeneration.

Protocol

The UAS FLII12Pglu-700µδ6 transgenic flies were reported in Volkenhoff et al.¹⁰ and kindly provided by Dr. S. Schirmeier. The UAS TDP-43^G298S transgenic lines were kindly provided by Dr. T. Iwatsubo¹³. Recombinant Drosophila lines harboring both UAS FLII12Pglu-700µδ6 and UAS TDP-43 transgenes were generated in the Zarnescu laboratory using standard genetic approaches and reported in Manzo et al.⁹. D42 GAL4 was used to drive the expression of the glucose sensor alone or together with TDP-43^G298S in motor neurons. All fly lines are maintained on molasses cornmeal media, at 25 °C in 12 h dark:light cycle.

1. Drosophila Ventral Nerve Cord (VNC) dissections

1. Make silicone elastomer dissection dishes.
   1. Pour silicone elastomer components into a 50 mL tube as indicated in the manufacturer's protocol and stir well.
   2. Fill 35 mm tissue culture dishes with approximately 2 mL of the elastomer mixture. Use a syringe to remove any bubbles. Cover the dishes and place them on a level surface in a 60 °C oven overnight to cure.
2. Dissect Drosophila larvae to expose the VNC while maintaining tissue viability.
   1. Collect a wandering third instar larva, rinse with ddH₂O, and then place it in a drop of HL3-buffer (70 mM NaCl, 5 mM KCl, 20 mM MgCl₂, 10 mM NaHCO₃, 115 mM sucrose, 5 mM trehalose, 5 mM HEPES; pH 7.1) on an elastomer lined dish.
   2. Using a pair of forceps (#5 or 55 forceps) under a dissecting microscope, pin the anterior and posterior ends of the larva dorsal side up, carefully stretching the larva lengthwise with insect pins (Minutein Pins).
   3. Make an incision just above the posterior pin using a pair of angled iris scissors. Make a vertical cut starting from the incision toward the anterior end of the larva.
   4. Add a few drops of HL-3 buffer if needed. Remove the trachea and the rest of the floating organs without disturbing the CNS. Pin the flaps stretching the body wall to expose the CNS while keeping the neuromuscular system (VNCs, axons, and neuromuscular junctions) intact.

2. Image acquisition

1. Optimize the imaging parameters.
   1. Turn on the microscope and lasers before starting dissections. Images must be acquired using an upright confocal microscope with a 40x water immersion lens. Use a 405 nm laser to excite CFP and acquire the images in both the CFP (465-499 nm) and the FRET (535-695 nm) detection channels.
   2. Optimize acquisition parameters such as scan speed (6), average (2), objective (40x), zoom (1x), pinhole size (1 AU), and spatial resolution (512 x 512). Adjust the gain such that the signal is in the optimal dynamic range. The same parameters must be used for all genotypes.
2. Image motor neurons within the VNC.
   1. Place the silicone dish with the dissected sample under the lens. Lower the lens so that it comes in contact with trehalose-sucrose HL3 buffer (see 1.2.1). It is critical to ensure that the lens is completely submerged in the buffer. Add more buffer if required.
   2. Use the CFP and FRET channels to manually select an optical section that consists of at least 6 motor neurons in focus, located along the VNC midline. The motor neurons are identified based on the expression of the glucose sensor and position along the anterior-posterior axis.
   3. Acquire images every 10 s for 10 min. These images represent the baseline.
3. Stimulate with glucose supplemented HL-3.
   1. Remove the trehalose-sucrose containing HL3-buffer using a Pasteur pipette and replace it with 5 mM glucose supplemented HL-3 (70 mM NaCl, 5 mM KCl, 20 mM MgCl₂, 10 mM NaHCO₃, 115 mM sucrose, 5 mM glucose, 5 mM HEPES, pH 7.1). This step needs to be performed with great care, to avoid the movement of the VNC as much as possible.
   2. Acquire images every 10 s for another 10 min. These images represent the stimulation phase.
4. Save the images as .czi files or any file type supported by the imaging software with a file name including date, genetic background, experimental condition (baseline or stimulation), and channels used.
5. Reuse the parameters to image all the genotypes (Figure 2A).

3. Image processing and ROI selection

1. Open the czi files in Fiji/ImageJ and set Color Mode: Grayscale, choose View Stack with: Hyperstack, and select Split Channels into Separate Windows.
2. Process images using the drift correction plugins: turboreg and stackreg. Each channel is corrected separately by selecting Stackreg under Plugins, and then choose Transformation: Translation.
3. Manually select the Regions of interest (ROIs) to extract the mean gray value of each channel.
   1. Choose the ROIs by tracing all motor neurons that remain in focus over the entire time series. Use the ROI manager to save each motor neuron outline.
   2. Use the Multi-measure function to measure the mean gray value in each ROI at each time point. Copy the ROIs traced from the first channel to the second channel for the pre-stimulation image.
   3. Repeat this process for the post-stimulation images in each channel using the same cells/ROIs selected for the pre-stimulation image.

4. Data analysis

1. Calculate the FRET/CFP ratio, using the mean gray values for the FRET and CFP channels. Changes in the FRET/CFP ratios reflect alterations in the intracellular glucose concentration.
   1. Plot FRET/CFP ratios for each ROI and time point versus time. Within every VNC some of the ROIs will exhibit a drop in FRET/CFP ratios following stimulation.
      NOTE: This is interpreted as an inability of some neurons to uptake glucose, possibly due to stress caused by dissections and are therefore eliminated from the analysis. The remaining FRET/CFP ratios for each ROI and time point are averaged to generate a single FRET/CFP ratio per time point per genotype (glucose sensor alone and glucose sensor with TDP-43^G298S). These single FRET/CFP ratios are used for all subsequent normalizations. 31 ROIs were analyzed for Glucose Sensor controls and 24 ROIs were analyzed for TDP-43^G298S.
2. Baseline normalization: Calculate the average of FRET/CFP ratios for the first 10 min (baseline), and then normalize individual FRET/CFP ratios for each time point of the entire experiment to the 10 min baseline average (see Figure 2B).
3. Timepoint normalization: Normalize TDP-43^G298S FRET/CFP ratios to FRET/CFP ratios from glucose sensor alone at each time point (see Figure 3A).
4. Bar graphs: For baseline versus stimulation comparisons in the form of bar graphs, use FRET/CFP ratios from 5-10 min (baseline) and 15-20 min (stimulation) for each genotype. Normalize stimulation ratio to the baseline ratio (see Figure 3B).

5. Statistical analyses

1. Evaluate the normalized data sets using Mann-Whitney correction (for baseline and timepoint normalizations) or Kruskall-Walis test (for bar graph representation of stimulation versus baseline FRET/CFP ratios).

Results

Image acquisition of the glucose sensor in the ventral nerve cord (VNC), ex vivo
To determine differences in glucose uptake in a Drosophila model of ALS based on TDP-43, a genetically encoded FRET-based glucose sensor was used. The sensor comprised CFP and YFP fused to the glucose binding domain from the E. coli MglB gene. Glucose binding elicits a conformational change, which can be detected by fluorescence resonance energy transfer (FRET) between the CF...

Discussion

The technique described in detail here can be applied to measure glucose uptake in a specific cell type of interest in live Drosophila using FLII12Pglu-700µδ6, a FRET based sensor which can detect changes in glucose levels to a millimolar range^10,11,12. This sensor has been previously used in conjunction with the UAS-GAL4 system to target its expression to specific cell types, including neurons

Materials

Name	Company	Catalog Number	Comments
35 mm tissue culture dishes	Sigma Aldrich	CLS430165
40X water immersion lens	Zeiss	440090	dippable, N.A. 0.8
dissection scissors	Roboz	RS-5618
Dumont #5 forceps	VWR	100189-236
Dumont #55 forceps	VWR	100189-244
Minutien pins	Fine Science tools	26002-10	used for dissections
SYLGARD 184 Silicone Elastomer Kit	Dow	1317318
Zeiss LSM880 NLO upright multiphoton/confocal microscope	Zeiss	N/A