High-content imaging continues to be a theme, and will Hi-C become a popular research tool?
Yesterday Akoya Biosciences showed a system that could do 50 fluorescent markers for high-content imaging of tissue samples, with a separate stage the mounts on an existing fluorescent microscope. Nanostring also celebrated the GeoMx launch complete with an evening launch party with speeches by Joe Beechem and OHSU’s Gordon Mills, in addition to Nanostring CEO Brad Gray. For additional background about the GeoMx system I wrote this up a month ago at #AGBT19. And for those of you at #AACR19 in Atlanta and did not come to this event hosted at the Georgia Aquarium, well it was a remarkable venue complete with a ballroom right across from an amazing display.
Miltenyi Biotec MACSima Imaging Platform
Also yesterday Miltenyi Biotec unveiled a system currently in early access that can measure more than 100 protein biomarkers on a sample (a fixed tissue slice or cells in suspension) in a self-contained, completely automated and integrated system, containing both integrated fluidics and optics. The photo below, with a neat transparent LCD mounted on a huge glass frame that changed the images depending on what instrument it was in front of, shows the MACSima Imaging Platform.
While Akoya uses fluor-labeled oligos as the intermediate interrogator (hybridizing only the existing bound antibodies on any given round and then removing the oligo probes), Miltenyi takes a different approach. By using a cleavable fluor, after staining with antibody and interrogating several colors through a combination of six excitation LEDs and five different emission filters, they use older ‘classic’ fluors like FITC where they can either ‘erase’ the fluor by cutting off the fluorescent dye, or they can gently photobleach. I say gently because the wavelength and intensity to bleach out the dye has to be optimized and carefully controlled so as not to damage the underlying tissue.
With an epifluorescent microscope resolution down to 100nm, it is sub-cellular resolution, and the multiple LEDs and filters enable a relatively high level of multiplexed markers per imaging round. The upshot of all this is that the system can interrogate ‘potentially hundreds’ of markers per sample, a remarkable feat even if say 5 markers are used per imaging cycle.
In addition, Miltenyi was smart to include several ways areas of interest can be interrogated from standard slides with special slide imaging frames. In addition you have a choice of validated methods including formaling fixated, paraffin-embedded, paraformaldehyde (PFA), and acetone. Three different tissue imaging frames, a 24-well imaging plate, and a unique ‘Micro Slide’ with 1.5 million hexagonal cavities 20 um in diameter for cell suspensions.
Miltenyi also has a collection of validated antibodies, as part of their ‘REAfinity’ recombinant and ‘REAlease’ releasable antibody line. The list of antibodies available for the system are not available at this time, but was assured that this list would become available soon. Another important component was the software, which they were not able to demonstrate but their brochure (available online here as a PDF).
Their launch video was available, and for convenience I’ve embedded it for you below.
The company webpage for the MACSima Imaging Platform is available here.
Arima Genomics for easy and content-rich Hi-C library creation
In genome-wide association studies (the NGHRI-EBI website has a cool on-the-fly map generator here) about 90% of all ‘hits’ where SNPs are associated with disease are in gene deserts. You read that correctly: a population variant or collection of variants that code for mathematical ability, for body mass index (and on and on) do not code for anything, they are variants in what is colloquially called ‘junk DNA’.
It is likely these variants affect the expression of genes very far away – up to a million bases away or more. And the mechanism to determine what is associated with what is a method called Hi-C. Originally 3C (chromosome conformation capture) as a one-to-one association between two specific regions, 4C was then developed as a one-to-all association, then 5C as a many-to-many association, and Hi-C is an all-to-all association. Hi-C is a method to determine what regions of the genome are close to each other in its native conformation, through a protocol that cross-links adjacent regions, and uses sequencing to determine which region is close to which.
Formaldehyde is used to cross-link the DNA in a native sample, the sample is then fragmented, purified and ligated, then a digestion with restriction enzymes and PCR amplified. The resulting library is then sequenced deeply.
Hi-C is an increasingly popular research method; here is the original 2009 paper describing the method and the description of ‘two genome-wide compartments’ and something called a fractal globule. This method gives 3D information about genomic organization, a third-order level of organization that you may remember from introductory Biology, where the first-order organization is the linear order of bases, the second-order organization is wrapping of linear DNA around histones, and the third-order being these stacks of histones. An active area of 3D genomics and its impact on gene expression is the topologically-associate domain or TAD. There are cases where a disrupted TAD leads to disease due to disrupted gene regulation, and you can imagine the effect of 3D organization of the genome with clinically relevant cancer conditions such as HRR and the effect of PARPi.
There are two existing companies that offer Hi-C kits and services; Phase Genomics (leveraging the Hi-C concept also to the investigation of metagenomes, through cross-linking of bacterial DNA in-situ) and Dovetail Genomics (primarily through a service offering).
A third company, Arima Genomics, has told me they have data from customers showing superiority along several dimensions with these two other offerings, including less noise (as measured by percentage of contact maps less than 15 kb in distance), much lower input requirements (nominally 2 to 3 million cells but they can go as low as 5,000 cells depending on the biological question being asked), and a simplified kit protocol.