1. Basic idea of microarray copying
Microarrays are versatile tools for high throughput screening. Nevertheless they are severely limited. Either the molecules are synthesized in-situ directly on the surface or in-vitro or in-vivo produced externally and then transferred onto the surface. In-situ synthesis shows low yield in terms of purity and restricts therefore the biomolecule probe length to ~50 bp for light-synthesized DNA (Affymetrics) or ~200 bp print-synthesis (Agilent), but allows up to millions of spots per array. Ex-array synthesis on the other hand provides high-purity molecules, but the (bio)synthesis and purification of these molecules is tedious, time consuming and expansive. Also the printing process takes time. Even if 1 spot can be made per second 100,000 spots will take more than a day.
Therefore the idea arose, why not copy microarrays? Why not make DNA, RNA and protein microarrays as high quality copies of a high quality original? It worked fine for text books and images. So why not apply it for DNA? Why not build a biomolecule copying machine? A biomolecule xeroxer?
Interestingly copying is a standard procedure in biochemistry, but never really seen under the aspect of “make a copy”. DNA is replicated by DNA polymerase, RNA is a copy of DNA made by RNA-polymerase and proteins are copies from RNA made by ribosomes. Each of this copy processes is commercially available as enzymatic mix and partially used since more than 50 years! Therefore nature provides all “inks” to enable a copying process.
The only question is, which matrix should be taken for copying. DNA- is the answer. Preferably DNA from a whole genome, best the DNA from a sequencing chip. As any DNA which ever will be topic of analysis has to be sequences the best choice to as copy master and original will be a sequencing chip. Such a sequencing chip will provide all DNA needed and ever requested for analysis.
With this idea we set up the first Microarray Copying approach in 2010. Our copy master is a structured plate with small wells – initially a 454 sequencing chip. Each well contains a different DNA. The copy approach is now very simplistic. Fill in the DNA, either in solution, immobilized on the wall of the wells or on a bead put into the well. How the DNA is located in the well does not matter, it only has to be ensured that it remains there for the copy process. Then the wells are filled with enzymatic mix producing the intended biomolecules (DNA polymerase = DNA, RNA polymerase = RNA and so called cell-free or in-vitro transcription and translation mix = protein). Directly after filling in the copy mix, the wells are sealed with a lid, which provides binding properties. These binding properties are in accordance to the copy process and ensure that the copied molecules bind to the lid (primer coating for DNA and RNA copy, halo-tag-ligand or Nickel-NTA for protein copy). After the incubation time the lid is removed and the copied microarray can be applied for measurements.
In 2012 Jochen Hoffmann realized within his PhD under the supervision of Günter Roth first a “DNA copy paper” (Hoffmann, J.; Hin, S.; Stetten, F. v.; Zengerle, R.; Roth, G. (2012): Universal protocol for grafting PCR primers onto various lab-on-a-chip substrates for solid-phase PCR. In: RSC Advances 2 (9), S. 3885–3889.) which enabled him to run a solid phase PCR on common materials used in microfluidics like PDMS, PMMA, Glas etc.
With acquired knowledge and experience we could miniaturize the same PCR reaction down to the size of a well of a 454 NGS chip – 20 pL. Therefore the first DNA microarray copy was realized in a sequencing chip geometry. In the beginning he copied DNA from beads deposited in the wells.
But with time and increasing experience combined with optimized PCR protocols we could realize a digital PCR, which allowed us to amplify even a single DNA strand into a whole spot on the array. The copy source still was a 454 sequencing chip, but instead of beads only DNA solved in PCR mix was needed. He could realize a copy in more than 100,000 cavities with DNA length ranging from 100 bp to 1,500 bp.
The according publication (Hoffmann, J.; Trotter, M.; Stetten, F. v.; Zengerle, R.; Roth, G. (2012): Solid-phase PCR in a picowell array for immobilizing and arraying 100000 PCR products to a microscope slide. In: Lab on a Chip 12 (17), S. 3049–3054.) describes details of such a picowell microarray copying approach. Hs efforts have been honoured by receiving the Inside cover of Lab on a Chip.
In the following year Christin Rath realized further improvements of the picowell copying approach. She generated a pool of copyable aptamers (single strand DNA, which binds a target just like an antibody). This pool was then copied as digital PCR to generate a DNA microarray yielding ~ 20,000 aptamer spots. On this array she performed first a hybridization experiment allowing her to identify by sequence binding aptamers (red & green stained) and non-binders (blue). After the hybridization she removed the hybridization probes and performed a binding experiment against the target (thrombin) followed by staining with a primary and secondary antibody. This was the first fully functional and assayed DNA microarray copy. Each single dot can now be valued for its binding functionality.
But also first steps with protein copying were done in 2012. Jürgen Burger realized in his PhD thesis the first microfluidic setup enabling to copy a DNA pattern from a PDMS chip onto a glass slide.
The according microfluidic setup and the biochemistry was constantly improved by groups members (microfluidics by Suleman Shakil, Nessim Ben Amar, Jürgen Burger, Johannes Wöhrle and Philipp Meier, biochemistry by Normann Kilb, Tobias Herz, Joke Lambrecht, Stefanie Rösch) and allows now the first steps to make several protein copies from a DNA template.
3. Detection technology: iRIf
To detect and analyse binding between molecules typically a labelling with a fluorophore or enzyme is applied. But such labelling is an additional effort and also a severe source of artefacts as it changes the labelled molecules properties or can simply be a steric hindrance blocking an important interaction site.
Therefore we looked for a label-free real-time detection technology and made a find in the so called iRIf-detection technique (imaging Reflectometric Interference) initially invented by the group of Prof. Günter Gauglitz, Tübingen and commercialized by Biametrics. We build up very fast a close cooperation to Biametrics which lead to several joint diploma thesis works resulting in several prototypes of microfluidic microarray reading iRIF-devices (PhD thesis of Jürgen Burger, Diploma thesis David Lämmle and Linda Rudmann). An iRIF-based microarray scanner is now commercially available under the product name bScreen LB 991 .
iRIf is a known phenomenon, well known by every child. The wonderful colour play of a soap bubble is induced by interference of white light reflected on the different interfaces between air-liquid-air. If the molecular thickness of this layer system changes the colour changes. Therefore the colour holds detailed information about how many molecules are there. Seeing a soap bubble means to see an image of Reflectometric Interferometry of its optical properties.
In a more scientific manner the iRIf signal can be analysed to get mass information and as such real-time kinetics of molecular binding events. The iRIf signal can be either generated under white light (just like the soap bubble) and the shift of the minimum/maximum wavelength is measured or at a single wavelength and the change in intensity is measured. The one wavelength monitoring is much faster, as it is only needing a CCD camera instead of a spectrometer (like for with light imaging). In any case a special glass slide with a buried high-refractive layer is recommended to obtain a high signal-to-noise iRIf signal. Such glasses are still transparent, can be used for fluorescence measurements and are typically manufactured in the same way like solar cells.
4. Results: Label-free detection and analysis
On the prototypes built by Jürgen Burger we now can analyse whole microarrays for molecular binding interactions. Several microarray generation systems have been realized already with spotted, microcontact printed and also light-synthesized microarrays.Spotting was mainly performed in collaboration with Rühe group at IMTEK.
But as iRIf is a real time detection system the most impressive part of this technology is to simply watch the videos:
From each pixel of an iRIF-movie a binding curve can be derived. The actual pixel resolution is 8.6 µm. For statistical reasons and reduced background noise we average typically 200 pixel (~0.15 mm^2) to generate a binding curve and analyse them quite similar to a Biacore measurement. Derived from on- and off-kinetics Kd-values can be determined.
Recently we were able to copy from PDMS chips directly DNA aptamers onto iRIf glasses. Each spot was generated from one single DNA strand.
From each spot a binding curve was obtained allowing us to identify good, weak and non-binders.
And also each spot could be sequence identified via sequence specific hybridization. We seen then that not all green spots show a similar binding. E.g. the green spot characterized to the lower right shows only a weak binding.
This allowed us to identify the binders as well as non-binders directly label-free by their iRIf signal. It is quite interesting that we found some spots (~2% of all spots), which showed a binding behaviour different to their staining behaviour. We aim to re-gather these spots from the surfaces and will analyse them. We intend now to re-gather the DNA via photo-cleavage together with BIOSS and the Life Imaging Center.
Nevertheless, we showed now on the DNA level, copying of DNA microarrays in combination with direct analysis of binding kinetics is possible. The actual layout realizes ~4,000 spots.
Meanwhile we proposed to show „immunizations“ from blood samples by copying potential vaccine candidates onto surfaces. First experiments look quite promising. As on 30.07.2015 we had a “political visit” by Edith Sitzman and Cem Özdemir we also prepared a very special experiment. In the last years of last millennia rabbits have been immunized against different molecules. The blood was taken after the immune reaction and frozen in 1998. By serendipity and thanks to BIOSS we got some of these rabbit blood samples and also the information about the potential used molecules. Therefore we prepared a 10 pixel protein microarray with according potential vaccine candidates.
The results can be also envisioned as video.
We applied subsequently after blocking the serum of rabbit #1 (video time 00:004 till 00:008) and then serum of rabbit #2. Clear to see we had no immunogen or vaccine for rabbit #1 on our chip, but also clear to see rabbit #2 was shows strong reaction for GFP-constructs.
A detailed analysis shows that not all constructs had the same reaction intensity, but all GFP-constructs are clearly positive.
Requesting the experimental details to this helpful animal donating blood 17 years ago revealed – yes, this rabbit was immunized with GFP. So that proofs us right with our approach and the nice thing is, we need only some microliters of serum for our experiment. So one drop of blood may be enough to find all vaccine candidates, if our protein microarray provides enough antigens. So we are working now on “making more proteins for finding all vaccine candidates” – simply increase the 10 pixel protein microarray to the same size as our DNA microarray copies already run 10,000 to 100,000 pixels. More than enough to cover all proteins from any pathogens.
… more to come…