High-flexibility combinatorial peptide synthesis with laser-based transfer of monomers in solid matrix material

Laser writing is used to structure surfaces in many different ways in materials and life sciences. However, combinatorial patterning applications are still limited. Here we present a method for cost-efficient combinatorial synthesis of very-high-density peptide arrays with natural and synthetic monomers. A laser automatically transfers nanometre-thin solid material spots from different donor slides to an acceptor. Each donor bears a thin polymer film, embedding one type of monomer. Coupling occurs in a separate heating step, where the matrix becomes viscous and building blocks diffuse and couple to the acceptor surface. Furthermore, we can consecutively deposit two material layers of activation reagents and amino acids. Subsequent heat-induced mixing facilitates an in situ activation and coupling of the monomers. This allows us to incorporate building blocks with click chemistry compatibility or a large variety of commercially available non-activated, for example, posttranslationally modified building blocks into the array's peptides with >17,000 spots per cm2.


Transfer process over longer distances
We statistically analyzed a larger area of the transfer (shown in Supplementary Figure 2) with increasing distance up to 100 µm (about 75 spots for each distance). The analysis in Supplementary Figure 3 shows an increasing spot width median and a drastically increasing variation with increasing distance. However, transfer is generally conducted without a distance between donor and acceptor slide, which significantly decreases the amount of outliers.

Multiple usage of donor slides
We have analyzed the multiple use of one donor slide by irradiating the exact same donor slide positions up to 20 times. Therefore, we generated different patterns corresponding to the number of repetitive irradiations ( Supplementary Fig. 6).
We prepared one donor slide, covered with an Fmoc-Glycine-OPfp polymer material layer, by conducting the laser transfer according to the layout in Supplementary Fig. 6. Afterwards, an acceptor slide was patterned once with this donor slide, according to the complete green pattern in Supplementary Fig. 6. The transferred amino acid pattern was coupled and the surface washed, capped, and stained with a rhodamine (NHS)-ester. No significant decrease in fluorescent signal intensity compared to unused areas of the donor slide could be observed ( Supplementary Fig. 7). Thus, we can reuse the donor slides at least up to 20 times.
The average standard deviation of spot fluorescence intensity is 11.3 %. During peptide synthesis, we repeat the transfer step once, which renders even lower standard deviations.

Optimum laser transfer conditions
To obtain the optimum process parameters (absorbed power and duration of laser irradiation), we varied these parameters systematically in transfer experiments and thoroughly analyzed the results. Supplementary Figure 8 shows the fluorescence image of a parameter variation pattern. For higher irradiation energies, we obtain larger spots. Our transfer process is limited by two factors: Too strong lasing destroys the laser absorption layer by burning the polyimide and creates a ring of molten transfer material (see Supplementary Fig. 4). With too weak lasing energies, no transfer occurs. We used the formula: to model this boundary condition (dashed line in Supplementary Fig. 8), and found a good match to experimentally obtained data ( Supplementary Fig. 8).
Hence, we assume that a successful transfer is possible when the transfer material layer of the donor is heated by laser irradiation above a certain threshold temperature , which should be close to the glass transition temperature of about 70 °C of the matrix material 25 . • . Supplementary Figure 8 shows that this functional relationship is indeed experimentally observed, by comparing a fluorescence image of a parameter variation pattern to such a curve.

Spot fluorescence statistics
The fluorescent staining of the 3-, 6-, and 9mer peptide arrays (Fig. 4a) with a fluorescently labeled streptavidin resulted in a rather low intensity standard deviation (about 10 %). We analyzed for each peptide type 1600 spots (Supplementary Fig. 9).
The synthesis of the Flag and HA peptides (see Fig. 4b-d) and 62 variants resulted in the expected results. Supplementary Fig. 10 shows the detail of the synthesis, where the Flag and HA peptides and 62 variants were synthesized in 4 replicas of 8 x 16 spots (each sequence synthesized as 8 spot replicates). The median standard deviation of a spot type was 9.6 % for the anti-Flag antibody staining and 6.6 % for the anti-HA staining (also see Supplementary Table 4).
The monoclonal anti-Flag antibody mainly recognizes the Asp-Tyr-Lys (DYK, amino acid positions 2-4) sequence 1 , the monoclonal anti-HA antibody mainly recognizes the sequence Asp-Tyr-Ala (DYA, amino acid positions 7-9) 4 . The following tables conclude the binding motives and the more important amino acids for binding. However, because in some variants more than one amino acid is altered, the binding is more complex than described in the following tables (Supplementary Tables 1 & 2), which only denote a general tendency (list of all 64 average raw signals and standard deviation in Supplementary Table 4).
Considering the whole array, shown in Fig. 4b, we calculated an average deviation of 24.4 % for the HA peptide staining signal and 24.8 % for the Flag peptide staining (Supplementary Table 3).

Repetitive coupling yield and mass spectrometry
To determine the repetitive coupling yield, we synthesized the HA peptide (sequence: YPYDVPDYA). Therefore, we processed 5 surfaces with 10/90 PEGMA/PMMA coating for the synthesis of the HA peptide. Four surfaces were derivatized with a Rink amide linker (Iris Biotech GmbH, Germany), which allows for cleaving the fully synthesized peptides from the surface after the synthesis. The fifth slide was used for a control synthesis, without the Rink amide linker. For a planar and very dense synthesis, we chose a pattern of >215,000 spots with 100 µm pitch and maximum laser intensity, assuming to cover the whole slide with an overlapping spot pattern. Thus, we did not use reference markers to position the spot patterns of the consecutive amino acid layers. However, after the HA peptide synthesis, a control staining revealed that the spots did not overlap (Supplementary Fig. 13) and only about 56 -80 % of the slide was covered uniformly with the amino acid polymer mixture in each layer. This explains why we only achieved an average coupling yield per amino acid of 72.3 ± 11.0 % ( Supplementary Fig. 11), in comparison to about 90 % in Stadler et al. 1 : the surface was not covered uniformly and without reference markers, the overlap of the patterns of sequential layers was not sufficient. Furthermore, the HA peptide is known to have a difficult amino acid sequence (e.g. contains two prolines), which, according to Stadler Patterning, coupling, and the short washing step were repeated once for a better yield.
Acetylation ("capping") was performed as described in the methods section. However, we performed the capping step overnight, to ensure that no residual amino groups might interfere with the measurement.
To measure the amount of cleaved Fmoc from the surface with UV spectrometry, surfaces  Fig. 12).
For the control slide without the Rink amide linker, the side chain protecting groups were removed according to the protocol in the Materials section. The incubation of this control slide with monoclonal mouse anti-HA antibodies conjugated with Cy3 fluorescent dye ( Supplementary Fig. 13) was performed as described in the methods section.
A similar experiment was described in the supplementary materials section of Stadler et al. 1 .
We rely on the same chemistry, including coupling steps of 90°C for 60 min. However, a surface with a much higher amino group loading (100 PEGMA coating) was used in Stadler et al. 1 . There, a laser printer is used to print the same type of amino acid derivatives and matrix material in form of solid particles onto the same type of acceptor slides. In the supporting information of this publication, it was shown that the average amino acid coupling efficiency ranges between 90 -93 % and no racemization could be observed.
The distinct difference to our approach is the laser-based transfer and the amount of material. We have shown in Maerkle et al. 2 that we can fuse the aforementioned amino acid particles with strong laser irradiation. There, we showed that irradiation and the resulting heat do not harm the activated amino acid monomers in the matrix material. Thus, our laserbased transfer approach does not seem to change the chemistry in comparison to previously published approaches.

Click reaction
Fmoc-Pra-OPfp was prepared from commercially available Fmoc-Pra-OH via a reaction with pentafluorophenyl tetrafluoroacetate according to Green et al. 5 .  To label the patterned Fmoc-Pra-OPfp with a styrylpyridinium dye 6 , we used the click reaction. The dye was functionalized with an azide group in a diazo transfer reaction 7 with imidazole-1-sulfonyl azide hydrochloride 5 to yield the corresponding azides (shown above).