WHITE PAPER FAST PROTEIN INTERACTION BINDING CURVES WITH INO S F-HS CONFOCAL MICROSCOPE Christian Tardif, Jean-Pierre Bouchard Pascal Gallant, Sebastien Roy, Ozzy Mermut September 2017
Introduction Protein-protein interactions are involved in all cellular processes, including gene expression and cell growth, proliferation, nutrient uptake, morphology, motility, intercellular communication, apoptosis, etc. For that reason, constructing protein binding curves is one of the most powerful methods for studying molecular interaction strength in order to develop drugs that interfere with cellular processes. The most robust and accurate technique for determining protein binding curves of live cells in vitro is Förster resonance energy transfer (FRET) measured through fluorescence lifetime imaging microscopy (FLIM). Until now, this technique was considered complex and prohibitively time consuming. INO, in collaboration with Prof. David Andrews at Sunnybrook Research Institute, has developed a new imaging system that allows fast, automated, live-cell analysis of large multi-well plate experiments with the capacity of generating an unprecedented 32 protein-protein binding curves per hour. This performance is achieved through the successful integration of a novel eight-channel parallel photon counting detection module, a high power broadband tunable excitation source module, an ultra-stable confocal scanner, and efficient and seamless control and batch data analysis software. Additionally, the system is equipped with a 62-channel spectroscopic detection module that can be used as the reference channel in FRET experiments and as a high resolution (7 nm) and high bandwidth (430 850 nm) hyperspectral imaging channel. Figure 1: F-HS confocal microscope schematic 2 Technology Confocal Scanner The heart of the INO F-HS system is a versatile, ultra-stable confocal scanner with dual output channels. A motorized pinhole array with nine positions covers from 0.5 Airy unit (AU) at 450 nm to 3.9 AU at 650 nm and one 15 AU. This capability offers the user the freedom to select the thickness of the acquisition slice and in so doing the amount of background rejection. This feature is particularly useful for accommodating interrogation of different types of samples, for example, allowing either bright or dim samples to be investigated under our microscope. The scanner offers the capability to scan the laser beam without distortion over a field of view (FOV) of 9 mm at the microscope port, which corresponds to 225 x 225 µm at 40x magnification. The scan rate can be as fast as four frames per second for a 600 x 600 pixel image. Two ultra-stable filter wheels make it possible to reconfigure the dichroic and detection spectral channel filters without any
need for alignment or signal optimization. Each dichroic filter wheel has six available positions, offering plenty of flexibility for programmable instrument reconfiguration in between images or even within an automated imaging experiment. The majority of the optical components are rigidly fixed and permanently aligned to ensure stable and robust measurement. Prior to final alignment and delivery, the entire confocal scanner is subjected to thermal cycling to relax any residual stress and ensure months of alignment-free, optimal operation. Figure 2: Confocal scanner/splitter submodule attached to the microscope body diagram Source Interleaver Module FRET-FLIM analysis requires the interrogation of two fluorophores. Because FRET interaction occurs at the sub-µm resolution scale, where Brownian motion is not negligible, it is critical to acquire the two channels quasi-simultaneously. Our source interleaver module achieves this in a wavelength-flexible manner by splitting the beam of a broadband pulsed supercontinuum laser into two arms, delaying one arm by half a period and recombining them at output. Independent motorized filter wheels in each arm allow for full flexibility in the choice of the two wavelengths between 430 nm and 750 nm. The module is permanently aligned for maintenance-free operation using assembly techniques borrowed from INO s industrial fiber laser technology. Figure 3: Source interleaver module Parallel Detection TCSPC Module Fluorescence lifetime extraction of dim and sensitive samples like live-cell cultures is best achieved with time-correlated single photon counting (TCSPC). TCSPC detects photons one at a time, and the circuit allowing the timing of each photon detection with respect to the excitation pulse is limited to one photon per excitation pulse. TCSPC acquisition speed is therefore ultimately limited by the laser repetition rate. Statistical biases, such as the photon pile-up effect, further limit the maximum photon acquisition rate to about 5% of the laser repetition rate. The 3
INO F-HS system pushes the boundary of photon acquisition speed with a multiplexed detection solution, dividing the output beam evenly between an array of eight independent SPAD detectors. Even splitting of the intensities is crucial to ensuring that all detectors reach saturation at the same time. A cascade of D-shaped mirrors is used to split the output with a minimum of optical surfaces. The net result is a system that can image eight times faster with a negligible increase in the total excitation energy used per pixel. Figure 4: TCSPC detector module Hyperspectral Detection Module The capability to collect cellular context information along with binding curves can provide even more insight into the biochemical processes being studied. The INO F-HS system s second channel is a hyperspectral detector capable of unmixing multiple fluorophores for multiple-label experiments. The hyperspectral detector module can therefore be used to simultaneously detect the reference channel for a FRET fluorophore pair as well as other spectrally separated molecular labels. Hyperspectral detectors compatible with laser scanning confocal microscopes are generally based on Hamamatsu s H7260 32-channel linear PMT array. Systems based on this detector usually trade resolution for bandwidth or vice versa due to the limited channel count of the detector. The hyperspectral detector module splits the incoming light on two of these PMT arrays using a rooftop mirror. A small spectral overlap ensures a seamless transition between the two detectors. The result is a spectrally resolved detector with 62 effective channels covering the 430 850 nm bandwidth with a resolution of 7 nm. Figure 5: Hyperspectral detection module 4
Microscope Module The INO F-HS system is coupled to a Nikon Eclipse Ti-E inverted microscopy platform. This microscopy platform provides the additional features, such as focus tracking and motorized stages, required for automated microscopy experiments. An OkoLab incubator controls temperature, humidity, and CO 2 levels. Both the Nikon element software and its HCS automation module can take control of INO F-HS system configuration to provide programmable control instrument parameters such as filter selection, excitation power, and acquisition. F-HS Software Interface The F-HS system is controlled by a user-friendly and flexible software interface. Specific configurations can be saved and recalled at a later time either manually, through the F-HS software interface, or in an automated fashion, through Nikon Element. The software interface also includes data visualization and analysis functionalities. Multiple regions of interest can be selected manually or automatically and analyzed for lifetime and spectrum. Below is a non-exhaustive list of the F-HS software interface functionalities: Instrument control - Frame rate - Image dimension - Pixel size (zoom) - Frame averaging Automatic image saving Visualization of the transmission spectra of the selected filter configuration Real-time photon count rate display for laser power adjustment Automatic ROI binning Single and double lifetime fitting Documented software API for user-developed automatic ROI selection and fitting routines Interface with NIS Element Batch analysis of large imaging datasets Figure 6: Acquisition App window 5
Figure 7: Analysis View window Figure 8: Automatic ROI selection and lifetime histogram and spectrogram of ROI ID 1 6 Experimental Results The following experiment demonstrates the capabilities of the INO F-HS confocal HCS microscope. In cancer cells, anti-apoptotic transmembrane mitochondria protein B-cell lymphomaextra large (BCL-XL) and its associated pro-apoptotic partners, such as BCl-2-associated death promoter (Bad) are particularly viable. Drugs that prevent these linkages can promote cancer cell death, reduce resistance to traditional chemotherapy, and limit tumor initiation. Prof. Andrews used the INO F-HS confocal HCS microscope to investigate drugs that disrupt this molecular partnership through FRET measurement between mcerulean3-bcl-xl and Venus-Bad. Results are shown in Figure 9. In this experiment, mcerulean3 was a donor fluorophore and was detected with the TCSPC channel for lifetime determination. Venus was the acceptor fluorophore and was detected by the hyperspectral detector module. For each well, a 600 x 600 pixel image corresponding to a 200 µm square field of view was captured. The imaging process was repeated for each well of the 384- well plate. Imaging the entire plate took less than 8 hours. For the analysis, the automatic ROI selection algorithm was used to extract the mitochondria pixels where mcerulean3-bcl-xl was most likely to be present. The analysis software adds all ROI pixels to extract lifetime curves of the donor and intensity of both channels. Each ROI was fitted for fluorescence lifetime using the
software s batch processing capabilities. Lifetime FRET efficiency was plotted against the intensity ratio between the acceptor and the donor channels to obtain the binding curves. This analysis was performed over all 384 wells where different conditions were distributed. This experiment was conducted with Bad, Bad2A (a mutant that does not bind with BCL-XL) and the drug ABT263. Figure 9: HCS BC experiment using INO F-HS confocal microscope. Measurement of interaction between mcerulean3-bcl-xl proteins with their partner Venus-Bad. FRET efficiency is weaker when ABT263 is present, suggesting that the interaction is less probable. Kd map represents the screen realized with the system The results clearly show that ABT263 interferes with the binding of BAD to BCL-XL and is therefore an interesting drug candidate for cancer treatment. Conclusion A thousand things can go wrong in a complex live-cell study preparation. INO s F-HS system allows life scientists to focus on their biological experiments, set the plate onto the microscope, and go home for a good night s sleep, confident that their imaging results will be there for review the next day. The combination of high imaging throughput, permanent alignment, automated instrument control, and easy and powerful batch data analysis will unlock the analytical power of the FRET-FLIM protein-protein interaction characterization technique for high content screening drug development and other cellular process research. 7
General Specifications INO 2740, rue Einstein Québec, Qc www.ino.ca info@ino.ca Min Max Typ. Unit Source interleaver Wavelength 400 680 - nm Number of filters - - 12 - Delay line - - 12.5 ns Output polarization - - Circular - Pulse width (FWHM) @ 441 nm 150 250 195 ps @ 500 nm 72 112 92 ps Confocal scanner/splitter Number of exc./em. dichroic - - 6 - Image size @ microscope port 0.24 23 9 mm Image size @ sample with 40x obj. 6 575 225 µm Frame rate @ 600 x 600 0.005 4 1 fps Number of pinholes - 9 - Number of output channels 1 2 2 - Number of dichroic channels - - 6 - Number of filters by channel - - 6 - TCSPC detection Detector type - - SPAD - Photon detection efficiency @ 400 nm 2-5 % @ 650 nm 50-65 % @ 830 nm 35-45 % Photon detection rate 0 100 16 Mcps Time bin 1 - ps Sync rate - 150 40 MHz Full-scale range histogram mode 65 2.19 x 10 9 65 ns Buffer depth - - 2097152 - Hyperspectral detection Detector type - - 32 PMT array - Spectral range 438 854 nm Spectral resolution 6 10 7 nm Number of channels - - 62 - Microscope Model - - Nikon Ti-E inverted - Software Operating system - - Windows 7 64 bit - 8