Investigations on the Secondary Structure of Polypeptide Chains in Polyelectrolyte Multilayers and their Effect on the Adhesion and Spreading of Osteoblasts
- 1.4k Downloads
Inspired by the composition of the native extracellular matrix, biomimetic polyelectrolyte multilayers were assembled from polypeptides and the glycosaminoglycan chondroitin sulfate (CS). To investigate whether peptide conformation imposes an effect on the cell biological functions of osteoblasts, the secondary structure was analyzed by in situ infra-red and circular dichroism spectroscopy. Multilayers composed of polypeptides and CS reveal a predominantly random coiled conformation and impede osteoblast spreading. On the contrary, polypeptide chains in assemblies of poly-l-lysine and poly-l-glutamic acid (PGA) primarily adopt an intermolecular β sheet structure and reveal an increased area of spread, which consequently supports the proliferation of osteoblasts. When CS is replaced by PGA in mixed multilayers, we observe a structural rearrangement from random coils to β sheets with a concomitant improved cell response. We conclude that polypeptide conformation in biomimetic multilayer assemblies affects osteoblast response by altering the stiffness of the multilayer.
Polyelectrolyte multilayers (PEM) that are constructed through Layer by Layer deposition of oppositely charged polymers have become a powerful tool for tailoring biointerfaces. Over the past 20 years a significant amount of research has been dedicated to the characterization of PEM thin films and their application in life sciences (recently reviewed in [1, 2]). One of the main advantages of Layer by Layer assembly is the straightforward technology of immersing substrates into solutions of the selected polyelectrolytes. A wide variety of polyelectrolytes ranging from designed synthetic polymers to naturally derived biopolymers may be employed for film build-up. Biopolymers that can be assembled into multilayer architectures include the following: polypeptides, proteins, nucleic acids and polysaccharides. These biomolecules have been used to construct biomimetic coatings [3, 4, 5, 6]. The resulting biomimetic PEM films are of special interest for coating dental and orthopedic implants, particularly when they contain components of the native extracellular matrix of bone. Glycosaminoglycan chondroitin sulfate (CS) is one of the most ubiquitous polysaccharides of the extra cellular matrix (ECM). CS macromolecules are covalently attached to proteoglycan core proteins in a high density and play a key role in tissue metabolism. CS supports the biomineralization process [7, 8, 9], and sulfated glycosaminoglycans in general possess binding affinity to growth factors and synergistically support their bioactivity . Enhanced bioactivity can be imposed onto biomimetic PEM films by incorporating sensitive biomolecules, e.g. growth factors, from their aqueous solutions due to the mild coating conditions of the LbL technique [11, 12]. Despite the apparent potential of CS, only a limited number of studies have been dedicated to the characterization and application of PEM assemblies containing CS (just to mention a few [13, 14, 15, 16, 17, 18, 19, 20, 21]). This study is the first report on the conformation of the cationic polypeptide poly-l-lysine (PLL) complexed with CS in multilayers. Under fully hydrated conditions, in situ infra-red spectroscopy in the SBSR mode (single beam sample reference) and circular dichroism spectroscopy were employed to characterize the internal structure of (PLL-CS)n and (PLL-PGA)n films (PGA: poly-l-glutamic acid). Multilayers constructed from two polyanions were characterized with respect to their relative secondary structure composition. This study describes the preparation and analysis of multilayers that display conformational differences in a two dimensionally patterned fashion. We investigated whether the cell morphology and cell density of osteoblasts on PEM films vary with different conformational compositions.
2.1 Multilayer Preparation
Chemicals and reagents were purchased from Sigma Aldrich (Taufkirchen, Germany) and used without further purification, unless stated otherwise. PEM films were assembled on germanium crystals (Komlas, Berlin, Germany) and/or borosilicate glass discs (diameter 15 mm, B33 glass, Schott, Jena, Germany). The substrates were cleaned with detergents and deionized water prior to film assembly. Glass discs were etched with concentrated nitric acid immediately before the coating procedure. Germanium surfaces were plasma etched for 5 min in order to generate hydroxyl groups and the PEM coating procedure was initiated immediately. PEM films were assembled from poly-l-lysine (PLL, 30–70 kDa), poly-l-glutamic acid (PGA, 50–100 kDa) and chondroitin sulfate (CS, ~63 kDa, from shark cartilage). The polyelectrolytes were dissolved in HEPES/NaCl buffer (25 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid, 137 mM NaCl, pH 7.4) at a concentration of 1 mg/ml.
Film construction was performed either manually or automatically employing a dipping robot (DR3, Riegler&Kirstein, Germany) as described previously . Briefly, the cleaned substrates were first soaked into the polycation solution (PLL) and left to adsorb for 5 min. Subsequently the samples were soaked into three HEPES/NaCl wash solutions to rinse the surface from unbound polyelectrolytes. The polyanions (PGA or CS) were adsorbed by an incubation for 5 min followed by three rinsing steps. The film construction was performed by repeating these cycles until reaching the desired film architecture. All samples were rinsed in deionized water and air dried in a gentle stream of pressurized air.
2.2 FT-IR Spectroscopy
2.2.1 Spectra Acquisition in SBSR Mode
Employing a flow-through cuvette made of Delrin (OPTISPEC, Neerach, Switzerland) the effective length one of the MIRE is limited to 39.7 mm and the resulting number of active reflections is ~19 (for a MIRE thickness d = 2 mm). The Ge MIRE was coated with PEM by sequentially injecting the polycations and polyanions at 0.3 ml/min by means of a peristaltic pump and leaving them to adsorb without flow for 10 min. Subsequently unbound polymers were rinsed off the surface by pumping HEPES/NaCl buffer at flow rate of 0.3 ml/min for 5 min. The reference chambers were charged with HEPES/NaCl buffer (0.3 ml/min) throughout the entire experiment. SBSR FTIR spectra were recorded after the rinsing step without applying flow. Fifty scans were accumulated in a spectral range from 4,000 to 800 cm−1 with a resolution of 2 cm−1 (zero filling factor two). Experiments were performed at least three times.
2.2.2 Data Processing
The integration of the water bands and the vector normalization of the raw spectra for comparative investigations concerning the secondary structure of the polypeptide chains were performed with the OPUS software package (Bruker, Karlsruhe, Germany). In order to quantify the changes in secondary structure during the build-up of different PEM architectures a partial least square analysis (PLS) based on the principal component analysis (PCA) was performed employing the QUANT 2 package (Bruker, Karlsruhe, Germany, in depth information on multivariate data analysis is described elsewhere ). The QUANT 2 model was calibrated with representative spectra for the films (PLL-PGA)10 and (PLL-CS)10. Automatic data processing by the QUANT 2 algorithm comprised vector normalization, first deviation and smoothing (17 points) in the spectral range of 1,599 to 1,702 cm−1 (amide I vibration).
2.2.3 FTIR-ATR Microscopy
FTIR-ATR microscopy was performed on patterned PEM films. (PLL-CS)n films were prepared on glass substrates and finally incubated with a droplet (~0.1 μl) of PGA solution for 5 min. Subsequently the films were rinsed with buffer and deionized water and dried in a stream of pressurized air. Dried films were analyzed with FITR-ATR microscopy employing a Ge ATR objective (Hyperion, Bruker Optics, Germany). In a scan area of 1.4 × 1.5 mm2, 72 spectra were recorded. Spectra were analyzed with the Opus 3D software package by integration over defined wave number ranges (specified in the results section).
2.3 CD Spectroscopy
Suprasil quartz substrates (50 × 8.3 × 0.3 mm3) were cleaned and plasma etched prior to PEM coating. The substrates were dip coated manually as described above. The PEM coated quartz samples were analyzed in phosphate buffer (5 mM, pH 7.4) in a custom made QS cuvette (Hellma, Germany). The rectangular cuvette contains vertical grooves and can hold up to five quartz substrates (optical path 10 mm) . Spectra were recorded in a range from 180 to 260 nm with a resolution of 0.1 nm employing a Jobin–Yvon CD6-Dichrograph (Horiba—Jobin–Yvon, Paris, France).
2.4 Colloidal Force Spectroscopy
Samples with patterned PEM films as described in 2.2.3 were mounted in a fluidic cell and allowed to equilibrate in buffer for 30 min. With the CellHesion AFM (JPK, Berlin, Germany) force maps of 100 × 100 μm2 were recorded employing cantilevers that were equipped with glass spheres (diameter 11 μm, spring constant was determined to be 63.4 mN/m). For obtaining force distance curves the piezo movement was set to 200 nm (corresponding to an indentation of 30 to 70 nm depending on the stiffness of the coating) after reaching a set point of 1.9 nN. The speed was set to 8 μm/s and after each approach the cantilever was retracted from the polymer film to z = 4 μm above to sample surface before starting a new measurement. The slope of the force distance curves was obtained through a curve fitting procedure and plotted as a stiffness map of the sample surface (all employing the JPK software). Finally, the obtained stiffness values were transformed into Young’s moduli employing the modified Hertz sphere model, as described in a previous work .
2.5 Cell Biology
All PEM coatings for cell biological analyses were prepared on glass discs and sterilized by UV irradiation. Saos-2 osteoblasts were seeded onto film at a density of 2 × 104 cells/cm2 and cultured in McCoy’s 5A medium supplemented with 15 % fetal bovine serum and 1 % penicillin/streptomycin lasting up to 7 days (37 °C, 5 % CO2, 80 % humidity). After an incubation period of 7 days, cells were enzymatically detached from the surface, stained with trypan blue and counted in a haemocytometer. Cell vitality was assessed as the percentage of viable cells. Complete trypsinization was verified by phase contrast microscopy.
Fluorescence Microscopy: the culture medium was gently aspirated and the cells were washed with PBS buffer. Fixation was performed by incubation with 4 % paraformaldehyde for 30 min with subsequent rinsing of the fixed cells with PBS buffer. The cytoskeleton was stained with AlexaFluor®594 phalloidin conjugates (Invitrogen, 45 min) and nuclei were counterstained with Hoechst 33258 (Invitrogen, 1 μg/ml in PBS, 10 min). Confocal fluorescence microscopy images were recorded with a LSM 710 (Zeiss, Germany) with appropriate excitation and emission settings.
Scanning Electron Microscopy: samples were fixed with glutaraldehyde, dried by incubating in solutions with increasing alcohol content with subsequent supercritical drying and sputtered with a thin film of gold. SEM analyses were performed on an Evo LS 10 (Zeiss NTS, Oberkochen, Germany).
3 Results and Discussion
3.1 Film Construction
In an earlier study the growth regimes of both PEM systems were investigated by using reflectometric interference spectroscopy . Both revealed a non-linear increase in the optical film thickness with a higher growth rate for (PLL-PGA)n films. The ATR-FTIR spectra underline the non-linear growth of both films and support the steeper increase in film growth of the (PLL-PGA)n multilayer. According to the “in and out diffusion model” for exponential film growth this difference may be attributed to the different degrees of diffusibility of PLL chains in PEMs when complexed with PGA or CS .
In some cases the PLL specific absorbance decreases upon CS adsorption for films composed of more than eight bilayers. This finding is indicative of a reduced PEM stability for (PLL-CS)n with partial film erosion, also referred to as overshooting phenomena [27, 28].
3.2 Polypeptide Conformation
The distinct shift of the amide I vibration from 1,630 to 1,610 cm−1 corresponds to the presence of intermolecular β sheets as opposed to an intramolecularly paired β structure . Thus, building PEM films from PLL and PGA leads to the formation of inter-molecular β sheets (IMβS), which is in agreement with results of previous reports [31, 32].
Circular dichroism spectroscopy was performed to validate the secondary structure evaluations that were obtained by means of FTIR spectroscopy. CD spectra of native (not dried) PEM films on quartz substrates were recorded and compared (Fig. 4, panel B). The data for (PLL-CS)n films reveal a negative ellipiticity that converges to the X-axis at ~215 nm that is characteristic for polypeptide chains in a random coiled conformation. Only minor contributions of other secondary structure elements like α and β conformations are expected to account for the recorded CD spectra. The CD spectra of (PLL-PGA)n films in buffer reveal similar negative π–π* transitions at ~220 nm and positive n-π* transitions at 202 nm like they were observed for PLL-PGA complexes in solution . We attributed the dichroitic maxima and minima to the high content of β structure. Whether the β sheets in (PLL-PGA)n films are formed between single or multiple polypeptide chains (intra- and intermolecular, respectively) cannot be evaluated from CD spectroscopy. The CD spectra provide further evidence supporting the hypothesis that the polypeptide chains adopt a predominantly random coiled structure in (PLL-CS)n films and a predominantly β-structural conformation in (PLL-PGA)n films.
Both PLL and PGA chains are known to adopt random coiled secondary structures in a physiological milieu (pH 7.4, isotonic salt concentrations) . It is well known from the literature that it is possible to induce conformational changes for polypeptides like PLL and PGA in solution. By varying the pH, temperature and salt ions of the solution a transition of the polypeptide conformation from random coiled to α helices, β sheets and intermolecular β sheets was observed . When these folded polypeptides are adsorbed onto material surfaces their secondary structure is usually retained [31, 34, 35]. The results obtained in this study for PLL chains in (PLL-CS)n assemblies, i.e. that they retain their predominantly random coiled conformation, are consistent with recent reports on the conformation of PLL chains in (PLL-PAA)n assemblies  and PGA chains in (PAH-PGA)n assemblies  (PAA: poly-l-aspartic acid, PAH: poly-allylamine hydrochloride). Currently, IMβS formation in multilayers was reported for the following polypeptide pairs: PLL/PGA, poly-d-lysine/PGA and poly-l-ornithine/PGA . The self assembly of (PLL-PAA)n and (PLL-PGA)n into completely different conformations despite sharing similar chemical features is striking. A possible explanation for the formation of IMβS structure for (PLL-PGA)n assemblies could be due to the longer side chain of PGA as compared to PAA, indicating that hydrophobic interactions between the side chains of the polypeptide may contribute to the stability of the conformation. However, although ornithine’s (an amino acid not involved in protein synthesis) side chain is one methylene group shorter than lysine its polymer readily forms IMβS structures with PGA. Alternatively, the lower pKa value (PAA: pKa = 5.4, PGA: pKa = 4.5, ) and consequently the higher degree of ionization may affect the IMβS formation and stability indicating that salt bridges between PGA and PLL side chains are critically important. In this case, the shorter length of the PAA side chain may sterically hinder the formation of ionic bonds with the corresponding amides of the polycation. PLL and poly-l-ornithine possess similar pK values of 10.4 and 10.1, respectively .
After investigating the structure of PEM films composed of one polyanion the effects of changing the polyanion during film construction was analyzed. When CS was adsorbed onto a (PLL-PGA)5-PLL film an effect on the secondary structure was observed (Fig. 5). With each deposited bilayer of (PLL-CS) the content of conformation B decreases, and consequently the number of polypeptide chains in random coiled conformations increases. After an addition of 5 bilayers (PLL-CS) onto a (PLL-PGA)5 basis approximately 25 % of the polypeptide chains are in conformation A and 75 % in conformation B. The intensity of the IMβS structure does not decrease while the signal for random coiled conformation rises (as judged from difference spectra after PE adsorption), this suggests that a bizonal film structure results due to the change in the polyanion from PGA to CS during PEM assembly. The first zone (in proximity to the substrate surface) is predominantly composed of intermolecular β structure while the second zone (towards the film–buffer interface) is characterized by chains in random coiled conformation.
An even more pronounced effect on the film structure was observed when the polyanion was switched from CS to PGA. Upon the adsorption of the first PGA layer onto (PLL-CS)5-PLL a significant structural reorganisation occurred (Fig. 5). PGA deposition caused a rise in the content of conformation B from 0 % to more than 80 %. After the deposition of five bilayers of PLL-PGA on top of a (PLL-CS)5 film a secondary structure that closely resembles the conformation of a pure (PLL-PGA)10 assembly (i.e. 100 % conformation B) was obtained. As observed for the construction of pure (PLL-PGA)n films, the adsorption of PGA causes a dramatic increase in β sheet structure when compared to the adsorption of PLL.
The difference spectrum after PGA adsorption reveals a certain loss of CS chains either due to desorption of soluble PLL-CS complexes or through the substitution of CS chains by PGA. The fact that CS chains are not completely removed from the multilayer may be explained by the structural model that Abdelkebir et al.  recently proposed for (PLL-CS)n multilayers. This type of PEM film is proposed to consist of a loosely assembled surface zone and a densely packed (ionically cross-linked) zone in proximity to the substrate interface. The extent of hydration of the loose surface zone is sufficiently high, that the CS chains may be more prone to a replacement with PGA chains and/or film erosion via the formation of soluble polyelectrolyte complexes (PECs). The CS chains in the deeper zone, which is proposed to display a dense network of ionic cross-links, may be less susceptible to substitution via detachment processes. The CS chains that remain tightly bound to the PLL molecules in the dense zone are postulated to account for the sugar and sulfate vibrations in the IR spectrum of the (PLL-CS)5-(PLL-PGA)5 mixed multilayers.
Only few studies have been dedicated to the influence of using multiple polyanions (or polycations) for film construction on the resulting secondary structure of the assembly [36, 37, 40]. Debreczeny et al.  assembled films from PLL complexed with different mixtures of PGA and poly-l-aspartic acid (PAA), while in the present study the effect of changing the polyanion was investigated. While (PLL-PAA)n assemblies were characterized by a mainly random coiled structure, the addition of PGA to the PAA solution during build-up induced the formation of IMβS in the PEM film. The content of IMβS strongly increased when the concentration of PGA was raised. In a similar experiment, the same group deposited PGA onto (PLL-PAA)n films and revealed a substitution of the PAA chains by PGA with a concomitant formation of IMβS . We suggest that the mechanisms by which PGA displaces PAA in (PLL-PAA)n assemblies  and CS in (PLL-CS)n films (results of this study) are similar. Apparently, PGA breaks the bonds between PLL and CS (or PAA) and compels the PLL backbone into a β pleated sheet. The formation of the intermolecular β sheet seems to be an energetically favourable process and the resulting structure is extremely stable. This remarkable stability towards pH changes (pH range 4 to 10.5 ), temperature and drying-rehydration cycles [41, 42] is likely due to the multitude of hydrogen bonds between adjacent backbone carbonyl and amide, ion pairings of the charged side chains and potentially hydrophobic interactions between the methylene moieties of the PLL and PGA side chains. Only the exposure to extremely acidic buffers induces conformational changes and yields the formation of a porous film structure . Albeit, the (PLL-PGA)n assemblies are susceptible to chain substitution of PGA by stronger polyelectrolytes like poly(styrene sulfonate) that also breaks the IMβS structure .
3.3 Cell Biological Evaluation
Summary of the nomenclature, estimated secondary structure composition and architecture for the samples that were characterized in cell biological experiments
Estimated content of conformation B (%)
CS 70 %
CS 84 %
CS 90 %
PGA 84 %
PGA 90 %
PGA 100 %
The cell morphological analyses provided corresponding results (Fig. 6a). While the osteoblasts do not spread on films with a content of conformation B less than 70 %, the area of cell spread increased significantly when the content of conformation B was raised. In accordance with the cell count data the osteoblasts that were cultured on mixed multilayers with PGA as the top layer displayed a more extended morphology than cells that were seeded onto CS terminated films. While cells on PGA layers are predominantly well spread, the actin cytoskeleton is poorly developed when the osteoblasts are seeded onto CS top layer films, especially when they contain a high content of random coiled structure. As described in a previous study and by other groups there is a strong correlation between proliferation capacity and the spreading area of substrate dependent cells like osteoblasts (Fig. 6c) [17, 43, 44].
Given that the cell biological effect cannot solely be due to the physico-chemical differences between CS and PGA, the results of the cell culture experiments indicate that osteoblasts preferentially adhere and proliferate on polypeptide films with an intermolecular β structure. In contrast conformation A, a predominantly random coiled composition of polypeptide chains, poorly supports cell spreading and proliferation. In a previous study, we demonstrated that the surface charge (zeta potential) and the surface roughness (by means of AFM) are comparable for (PLL-CS)10 and (PLL-PGA)10 films. A prominent dissimilarity was detected for the stiffness of these multilayers . The mechanical properties of PEM assemblies may well be correlated to the conformation of their polypeptide chains. Random coiled chains are usually loosely assembled with a high degree of flexibility resulting in a low mechanical stiffness. The low stiffness is a result of the high degree of swelling caused by the CS chains. On the contrary, the backbones of polypeptide chains that self-assemble into intermolecular β sheets are tightly packed and their conformation is stabilized by a multitude of hydrogen bonds (peptide backbone) and salt bridges (side chains). This tightly packed polymer network consequently displays an elevated stiffness. Additionally, the hydration with water molecules may be sterically hindered and reduce the swelling of the films.
The cell biological investigations on mixed multilayers further reveal that the PEM film conformation is not sensed as a property of a homogeneous 3D-volume phase by the cells, since a difference in cell density was observed on PGA compared to CS terminated films. These results support the hypothesis that (PLL-PGA)n–(PLL-CS)x multilayers display a bizonal structure with distinct polypeptide conformations. When the number of CS bilayers x increases the superficial, random coiled zones covers the tightly packed underlying (PLL-PGA)n. Consequently, both the content of conformation B in the film volume and the conformation of the surface exposed layers affect the cell response. When the top layers of the mixed multilayers with similar content of conformation B are composed of random coiled chains, their inherent mechanical softness negatively interferes with cell spreading and proliferation.
3.4 Patterned Multilayers
FTIR microscopic analyses were performed on dried multilayers with a germanium ATR objective and revealed tremendous conformational rearrangements in the PGA treated areas of the coating. The recorded spectra were integrated with respect to conformation A (1,635–1,700 cm−1), conformation B (1,600 to 1,635 cm−1) and the entire amide I vibration (1,600 to 1,700 cm−1). The maps for the integrated conformation B band display an increase of IMβS from the untreated to the PGA treated area (Fig. 7, panel A, white to grey, respectively).
The IMβS formation is accompanied by a concomitant decrease in conformation A (data not shown). No differences in the total area under the amide I vibration were observed for the scanned area (data not shown). The structural rearrangement into conformation B is accompanied by an impressive increase in film stiffness as assessed with colloidal force spectroscopy in mapping mode (Fig. 7, panel B, black corresponds to soft, white corresponds to more rigid). In relation to the Young’s modulus of untreated (PLL-CS)n films, which was determined to be approximately ~9 kPa , the elasticity of the PGA treated area was significantly increased to 369 ± 143 kPa (corresponding to a difference in film stiffness of ~12 pN/nm, p < 0.0001). This stiffness is lower compared to the stiffness of (PLL-PGA)n films that was determined to be ~700 kPa . In a recent study by Schoeler et al.  the effects of the conformation of PE on the multilayers mechanical properties were observed, as well. Films composed of polysaccharides (carrageenan) in an ordered conformation displayed increased stiffness as compared to their respective counterparts that displayed a random conformation.
Whether the CS chains are “forced to diffuse” out of the PGA treated film volume and complex “free” PLL chains at the air/buffer/film interface or whether entire soluble PLL-CS complexes are released from the film and accumulated to form the rim, remains speculative. Since we found evidence for a limited diffusion of the PGA chains within the (PLL-CS)20 films and the film thickness decreased in the treated area, a mechanism based on the erosion of soluble PLL-CS complexes seems to be the more plausible explanation. Nevertheless, to understand this phenomenon in more detail further research has to be focused on the patterned films.
The layer-by-layer assembly of polypeptides offers the opportunity to control conformational arrangements. Compared to the secondary structure in solution the conformation of the polypeptide chains can either be maintained (as in the case of PLL-CS) or rearranged (as in the case of PLL-PGA) depending on the choice of the PE pairs. PGA was revealed to be a potent inducer of an intra-molecular β pleated structure in both PLL-PGA and in PLL-CS assemblies. The results of this study suggest that there is a correlation between the conformation of polypeptides in the film and the mechanical properties of the coating. PEM films composed of polypeptide chains with a random coiled structure result in rather soft coatings, while similar films that display a high degree of intermolecular β pleated sheets reveal an increased Young’s modulus. Especially for cells that obey to the rules of durotaxis, this increase in the film’s β structure content strongly supports cell spreading and an increased cell density after 7 days was observed. In addition to the effects on the film’s mechanical properties the secondary structure of the PEM may also affect the conformation of proteins that adsorb to its surface . It was recently discussed in a study by Ball and coworkers whether the polypeptide conformation may influence the nucleation kinetics of calcium phosphates on PEM . An example of how the formation of IMβS may be harnessed in (nano)biotechnological applications was recently demonstrated by Higashi et al.  who utilized PGA conjugation to nanoparticles to assist a well-ordered assembly of the nanoparticles on PEM templates. The capability of PGA to induce β pleated sheets in predefined areas can be exploited to create biomimetic coatings with patterned biological functions which would be useful for future applications in the fields of chip based co-culture approaches, biotechnology and life sciences.
The authors thank Martina Lackner and Christina Biermann for technical assistance. Financial support for parts of this study was provided by the Deutsche Forschungsgemeinschaft (DFG) and the German Israeli Foundation (GIF), and is gratefully acknowledged.
This article is published under license to BioMed Central Ltd. Open Access This article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution and reproduction in any medium, provided the original author(s) and source are credited.