>476 Million Cells Every 24 Hours
Automated Ultra High Throughput High Content Imaging
Araceli Biosciences, Tigard, Oregon, USA
Unlock the next level of ultra-high throughput screening with the Araceli Endeavor®, a high-content imaging system designed to automate and accelerate your workflows. Utilizing automated plate feeding, in just 24 hours the Endeavor imaged 165 full 1536-well plates, covering 253,440 wells, with >476 million cells across 5 channels at submicron resolution, enabling unprecedented scalability. Delivering high-resolution images of four fluorescent (FL) channels and brightfield across most of the usable well area (1.2 x 1.2mm) at a remarkable pace, the Endeavor offers unparalleled speed without compromise.
The Endeavor achieves exceptional data acquisition throughput empowering high-content imaging labs to achieve more in less time. 45TB of data were written over the Endeavor’s 25Gbps fiber connection to a network-attached storage (NAS) device. Writing directly to NAS, images can be analyzed by Araceli Clairvoyance™ immediately after plate acquisition, allowing for immediate feedback. With this capability, >100,000 compounds can now be screened in a day, in duplicate with controls (>3,700 cells/condition), in 5 channels.
The Endeavor excels in ultra high throughtput Z-dimensional imaging when integrated with an automation solution. We were able to image 3 z slices 1 micron apart and autogenerate a maximum intensity projection image (MIP) for all 87 1536wp in 4 fluorescent channels in <15 minutes/plate, generating 2,138,112 images of 251,166,070 cells in <24 hours. Z-dimensional imaging is crucial for resolving objects across multiple focal planes, ensuring optimal focus for complex structures like actin and mitochondria, even in challenging polymer-bottom HCS plates. Data were binned, with a 0.54µm pixel size instead of the default 0.27µm. Binning is a boon for large screen data storage: un-binned, these data would be 76 terabytes, but 2×2 bin reduces this to 19TB. If individual Z planes aren’t needed, the Endeavor protocol can specify saving just the MIP, further reducing data size to 4.5TB.
High throughput is only meaningful when paired with reliable, high-quality imaging—a standard that the Araceli Endeavor consistently meets. When dealing with over a million images/ day (or even thousands), it’s vital to be certain that your data is worth analyzing. Using the Araceli Clairvoyance™’ image QC tool, <0.1% of images were identified as out of focus (OOF), with 42.5% plates having no out of focus images. Using two Fourier ring-based metrics, Clairvoyance flags focal failures and sample aberration, taking 30-40sec/binned 1536wp and 2-2.5min for full resolution with cell counts only adding an additional 0.5-1 minute. With unparalleled confidence in data quality, automation unlocks the next stage in ultra-high throughput screening, making Araceli Endeavor a no comprise solution to scaling up high content imaging to previously inconceivable levels.
Methods: Plate load and imaging automation using GreenButtonGo (GBG) from BioSero with a Brooks PF400 robotic arm and barcode scanner feeding from 2 x 25 plate hotels; plate load/unload takes ~2 minutes/imaging round (not optimized). All imaging with Araceli Endeavor 2.1 in automation mode using a SiLA2 interface with predefined protocols specified in a worklist by barcode in GBG, with ‘loop worklist forever’ checked, manually stopped after 24 hours. For single plane: default focal plane and full resolution (0.27µm pixel) used with 10/30/40/50/10ms B/G/R/FR/TL exposures; Z stack data: 2×2 bin (0.54 µm pixel), focus from -1 to +1µm, save slice and MIP, exposure times 2/10/10/10ms. Due to the number of 1536wp required for this study, plates were reimaged multiple times throughout the 24-hour period. Plates were loaded and unloaded to the hotel between imaging runs, with barcode scanning at the beginning of the run. 1536 well microscopy-grade COC bottom plates used, containing 80-90% confluent epithelial cells. Focal quality, similarity and cell counts calculated with Araceli Clairvoyance™. Briefly, focus assessed on channel 1 (blue excitation, nuclear staining) using both Fourier ring correlation (OOF<0.99) and a patented Fourier-based similarity metric using a known in-focus image (OOF<0.8). Cell counting used template matching on nuclear stain.
69,120 Wells in an 8.5-Hour Workday
Ultra High Throughput High Content Imaging and Analysis
How productive can you be in a workday? Here we image and analyze 45 1536-well plates in a single 8.5 hour workday, yielding quantitative object-level data for over 34 million cells in 4 fluorescent channels with sub-micron resolution. Utilizing the Araceli Endeavor® imager and Clairvoyance™ analysis, this technical note shows true end-to-end ultra-high throughput high content analysis (HCA). With imaging starting at 9:00am, plates were imaged in an average of 10 minutes 9 seconds with analysis in 9 minutes 14 seconds per plate, yielding 46 plates imaged and 45 analyzed by 5:30pm.
The Araceli Endeavor® High Content Imaging system partnered with the Clairvoyance™ analysis software creates a platform that delivers seamless imaging and analysis. This powerful combination allows the high-resolution images to be written directly to a network attached storage device (NAS) where they are immediately analyzed after image acquisition. Utilizing 25GBs fiber connections, latency between imaging and analysis is eliminated, allowing near immediate feedback and iteration of imaging results, whether you are working with a single plate, or screening an entire compound library. With 69,120 wells delivered end-to-end, this means tens of thousands of compounds may be screened in duplicate, with controls, in a single workday, with dose response curves or a list of hits in hand before you leave the lab. At the end of the day, batch analysis was set up in Clairvoyance for further analysis overnight: cell morphology and quality control data were available for these plates the next morning.
Quality does not need to be sacrificed for speed. Araceli Endeavor® images 96- 384 and 1536- well plates at ultra-high throughput speeds with maximized well coverage and submicron resolution in 4 fluorescent channels plus brightfield in 10 minutes or less. With maximized well coverage representing 1.44 mm2/well for a 1536 well plate, a majority of the well is imaged with a 0.27 µm/pixel digital resolution, collecting an average of 763,122 cells per plate in four fluorescent channels at medium confluency. Of the 276,480 images acquired, only 0.37% were out of focus (excluding plate edges), as measured with Clairvoyance’s quality control algorithm and confirmed with cell counts. The analysis here delivers subcellular information at the level of the single nucleus with quantitative fluorescence, including nuclear:cellular intensity ratios, in <10 minutes/plate with the ability to generate actionable answers within the software itself. While these 46 plates were imaged manually by load and unloading plates by hand, throughput can be maximized through automation with Endeavor’s SiLA2 integration allowing 24/7 sample acquisition when paired with a robot arm. Overall, the workflow described here can deliver analyzed subcellular data at unparalleled speed, marrying ultra-high throughput screening to high content imaging without compromising coverage or quality.
Experiment details: Phenovista Biosciences generated 1536-well plates (Grenier 783892) with fluorescently stained (standard Cell Paint) human epithelial cells. Plates were imaged with Endeavor 1.3 using with a manual load/unload with a standard imaging protocol (default focal plane, 1 4400 x 4400 FOV/well, with blue/green/red/far red exposure times of 10/30/60/50ms) and written directly to a NAS. Clairvoyance 1.0 analysis read directly from the NAS when a run finished, using template match to detect nuclei with watershed-based segmentation yielding results on a single object basis. Cells were segmented by thresholding the far-red channel, using nuclei as a seed, intensity measured in 3 channels, and a ratio of nuclear:cellular intensity was generated. Batch analysis was run after initial imaging, generating ImageQC scores using an untreated control well as reference and morphology data.
