To achieve our goal of site-directed delivery of fructan-degrading enzymes over a prolonged timespan the selected approach was three-fold.

  • Firstly, the Cloning team aimed to engineer L.plantarum using Golden Gate Assembly to express the enzymes endo-inulinase, levanase, and invertase for the degradation of fructans to glucose.
  • Secondly, the Scaffolds team intended to create a biocompatible, mucoadhesive scaffold from cellulose sulfate and Polydiallyldimethylammonium chloride (polyDADMAC) to encapsulate the transgenic bacteria and thus protecting it from extreme pH values and digestive enzymes.
  • Lastly, the Cell Culture team would test the cytotoxicity of the created capsules in vitro using Caco-2 cells.

On this page, we present the achievements of our three teams through the course of our laboratory work from July until October.
Details, on how these approaches were performed can be found in the Experiments and Protocol section as well as in our chronologically documented laboratory journals.


    To lay the groundwork for all Golden Gate Assembly steps, the sequences for the raw backbone 1 (=BB1), backbone 2 (=BB2), and backbone 3 (=BB3) with their specific fusion sites (=FS) were planned and cloned in silico via Benchling. The BB1 and BB2 vectors with their respective fusion sites were kindly obtained by our supervisors at BOKU Vienna, but for BB3 a novel vector was created by our Cloning team, using the pSIP403 vector as a basis. The pSIP403 vector however contains five illegal sequences forbidden for the Type II-S system. To circumvent this, the plasmid was split into five parts and five forward and five reverse primers for each part were designed. Using these primers, five DNA constructs (c1b, c2b, c3b, c4b and c5a) deficient of any illegal sites were amplified, which were then used to assemble a novel BB3 via Golden Gate Assembly.

    Plasmid of assembled raw BB3

    The sequencing results of the five constructs used to assemble the raw BB3 showed a few mutations upon alignment with the in silico cloned vector. However, these turned out to be of no significance as they did not impair the function of vital elements of the backbone.

    Raw BB3 sequencing result

    With the novel BB3 as a basis, four different versions containing different fusion sites for more flexibility in cloning were created, namely version A-B, A-C, A-D, and A-E. This allows to insert up to five constructs into BB3.

    To achieve this, our supervisors at BOKU kindly provided us with BB2 containing the aforementioned fusion sites as inserts, which were then used to assemble our four BB3 versions via Golden Gate Assembly. All four versions were first cloned in silico via Benchling, which were then used as templates to align our sequencing results to ultimately confirm that our assembly of all BB3 vector variations was correct. As mentioned above, the shown point mutations did not result in any disfunction of the vector and can be attributed to mutations that naturally occurred over time.

    Empty BB3 with fusion sites AB, AC, AD, and AE

    The first assembly step was to clone our promotors (p1 – p8), enzymes (endo-inulinase = Endi, invertase = Invi, levanase = Levi) with their respective secretion signal A and B, and the terminators (ttCAT and pepNTT) into their corresponding BB1. The promotor p8 (=pGAP) in BB1 was already provided by our supervisors. To align our sequencing results, first, every BB1 with the respective gBlock was cloned in silico via Benchling. All promoters were cloned into BB1 with fusion site 1-2, all enzymes into BB1 fusion site 2-3, and all terminators into BB1 fusion site 3-4.

    Below, see the sequencing results for all BB1 are depicted, each with the in silico cloned template on top and with the corresponding sequencing result on the bottom, proofing the correctness of each BB1 construct.

    Promotors in BB1
    Enzymes in BB1
    Terminators in BB1

    The red fluorescent protein mCherry was also cloned into BB1 to be used as a reporter gene to quantify our promotors. This is discussed further below.

    Fluorescent protein mCherry in BB1

    Having all promotors, enzymes, and terminators successfully cloned into BB1, the second step was to assemble them into BB2, creating a number of different expression cassettes with varying promotor-enzyme-terminator-combinations. Through sequencing, it became apparent that clones suspected to be positive all either had frameshift mutations, deletions, insertions, or even the whole enzyme was missing (examples see below), rendering most BB2 expression cassettes unusable for further cloning steps

    For instance, in the picture below the sequencing result for a BB2 expression cassette (BB2_FSBC_p3_Levi_pepNTT) can be seen. Here, with clones 6.1 and 6.2 a shared deletion of the first base of the secretion signal can be observed. Clones 5.1 and 5.2 have a shared mutation in the promotor, that results in a translational change (TAT to CTA results in S à *).

    Example of a faulty BB2 sequence of promotor p3 with enzyme Levi and terminator pepNTT

    As our initially approach seemed to be too overwhelming for the E.coli to express correctly, we reconsidered our original approach and changed course.

    First, we used the weaker promotors to try and circumvent overwhelming the E.coli with our strong promotors as we suspected that the metabolic burden was too high for E.coli. However, this approach still did not result in any positive clones.

    Example of enzyme Endi AF with secretion signal B and terminator ttCAT with the weak promotor p4

    Next, we attempted to simply amplify the PCR products after the second Golden Gate Assembly step into BB2, therefore making our way from BB2 to BB3 without transformation in E.coli. This however proofed to be ill-fated, as our amplified PCR-products were either of too little concentration for further assembly steps or they were unusable after too many amplification-rounds. Below an example of this can be seen on a preparative 1 % Agarose gel: The bands were smeared and not possible to cut out, therefore they could not be purified for further use.

    Smeared prep. gel of PCR amplifications

    As we were unable to properly clone our intended promotor-enzyme-terminator-combinations into BB3, we made the decision, on one hand, to try another promotor to prove that our enzymes can be correctly expressed, and on the other hand by choosing a simpler protein to be expressed to proof our promotors were functioning. For the latter, we chose the red fluorescent protein mCherry. The results for this promotor quantification can be seen in our wiki page Contributions.

    To get to the bottom of the issue regarding our incorrectly expressed enzymes, we proceeded to clone our enzymes together with another promotor, namely the inducible orfXP promotor. With this promotor, we were able to clone our three enzymes into BB3 correctly into E.coli. However, as to not encounter any limitation of a too high metabolic burden with our initial approach of cloning all three enzymes into one expression cassettes in BB3, we instead cloned each enzyme separately into BB3. As a next step, the three enzymes could be cloned into one expression cassette in order to observe the effects of the metabolic burden on Lactobacillus. Due to time constraints, we will not conduct these experiments.

    With this approach, our enzymes Endi KM A, Invi A, and Invi B with the terminator ttCAT and Levi with the terminator pepNTT were cloned correctly into BB3. Furthermore, we were also able to transform the expression cassettes containing these constructs into L.plantarum, as seen in the sequencing results below.

    Promotor orfXP in BB1
    BB2 transformed in E.coli
    BB3 transformed in E.coli
    BB3 transformed in L.plantarum

    However, we were not able to obtain promising clones of Endi KM B in BB3. Nevertheless, we could identify promising E.coli clones carrying the expression cassette of the inducible promoter with the enzymes Endi AF A and Endi AF B. To this date, these clones still need to be sequenced and transformed into L.plantarum.
    The next step with our transformed L.plantarum was to prove that the enzymes are indeed expressed and secreted after induction.
    We aimed to prove the secretion of our enzymes by performing a His-tag purification of the supernatant and afterwards perform an SDS-PAGE stained with Coomassie blue to detect the proteins. Unfortunately, we could not detect any proteins with this method. Therefore, we aimed for a more sensitive method. This allows us to analyse the supernatant of previously induced liquid cultures of the clones in a straightforward manner. Subsequently, the membrane was stained using an Anti-His HRP conjugate antibody. This provided us an easy and quick proof that the enzymes are secreted into the supernatant.

    Slot Blot of the supernatants of the liquid cultures that were previously induced. WT refers to the supernatant of the wildtype, while NC refers to not inoculated MRS medium handled the same way as the samples.Purified enzymes refer to the supernatant purified by a His-Tag purification, while PC-P is a supernatant spiked with a known His-tagged protein. PC-SB refers to the same protein applied to the Slot Blot without previous purification.

    As seen in the picture, we achieved to detect His-tagged proteins in almost all supernatants of our clones. Differences in strength of the signal are most likely due to concentration differences as a result of different amounts of applied samples as the membrane unfortunaletly clogged while loading. We also analysed the His-tagged purified samples and could not detect any protein there. Since we could detect the enzymes in the supernatant and additionally also the positive controls shows a signal, we assume that the protein concentration after purification was too low to be detected.
    Hence, we successfully showed the transformation of L.plantarum and additionally demonstrated the secretion of our enzymes.


    At the time of the Wiki-freeze, we achieved to clone the enzymes endo-inulinase with secretion signal A and invertase with secretion signal A&B with ttCAT as terminator as well as the enzyme levanase with terminator pepNTT with the inducible promotor orfXP into one Backbone 3 with the fusion site AB. Additionally, we successfully transformed each construct into L.plantarum and showed that our enzymes are secreted into the supernatant. In a final step we aim to prove that our enzymes are able to degrade fructans, but this remains subject of future experiments.

    The Caco-2 cells were thawed, and the media was regularly changed according to protocol.

    The cells were made sure to be at 80 – 90 % confluency before passaging them.

    Caco-2 cells, 2 days after the 3rd passage:

    3rd passage

    Caco-2 cells at 90% confluency:

    Cells at 90 % confluency before passaging

    We had our Caco-2 cells in active culture for six passages. Unfortunately, due to limited time in our Cell Culture Lab and the slower than anticipated progress in our Cloning and Scaffold labs, we were unable to finally test the cytotoxicity of our capsules.

    However, a protocol of how we would have undergone the Resazurin assay for testing the toxicity of the created capsules in vitro using Caco-2 cells can be found in the “Protocols” section.

    Capsule synthesis

    At the beginning we focused on synthesizing our own cellulose sulfate to use as capsules. There were two options that we found promising: the sulfation of cellulose in ionic liquids or in Dimethylamine (DMA)/LiCl, both using SO3/Py as sulfating reagent. For the first experiments with DMA/LiCl microcrystalline cellulose was used but because of its light molecular weight, it did not form capsules when dripping it into the PolyDADMAC solution. Thus, cellulose of higher molecular weight was needed. Our candidates were spruce sulfite pulp (SSP) and cotton linters (CL), they were used for the following synthesis. After analyzing the samples, it was shown that the SSP-synthesis provided the best results for building capsules, and the sulfating rate was the highest using DMA/LiCl instead of the ionic liquid because of its lower viscosity. The respective protocol can be found in the "Protocol" section.

    The best result was provided with 4 mol equivalents of SO3/Py, which you can not only see in the capsule building process but also in the FTIR spectrum.

    FTIR of cellulose sulfate and native cellulose

    Our synthesized cellulose sulfate (“SSP_IV”, in dark blue) was compared to native cellulose (“Reference Avicel PH-101", in orange. The peak at 811.6cm-1 shows C-O-S bonds and the signal at 1217cm-1 is characteristic for S=O bonds, both are clear evidence for sulfate groups.

    Furthermore, the bath solution had also a significant influence on the capsules. In high molecular PolyDADMAC the capsules were more stable than in medium molecular PolyDADMAC. Also, a detergent (Tween 80) had to be added to lower the surface tension of PolyDADMAC so that the capsules could be built more easily. (In experiments with E.coli we noticed that Tween 80 was too aggressive for living cells.)

    The best outcomes of sulfated SSP capsules are listed in the tables below:

    Sample name SO3/Py equivalent PolyDADMAC high 4% (diluted in 0.9% NaCl)
    SSP IV 3 +
    SSP 16.2 6 +
    SSP VI 4 ~

    In the first table, you can see 3 different samples in the same solution. Although our best sample (SSP VI, 4mol equivalents) was not very stable compared to the other capsules in PolyDADMAC high 4% (diluted in 0.9% NaCl), it showed better results in other bath solutions. You can see the effect of the bath solution on the capsule stability in the second table.

    Sample name SO3/Py equivalent PolyDADMAC high 4% (diluted in 0.9% NaCl), Tween 80, thiolated chitosan PolyDADMAC high 4% (diluted in destilled water), Tween 80, thiolated Chitosan Chitosan, 0.9% NaCl 1:4 PolyDADMAC high 4% (diluted in 0.9% NaCl) PolyDADMAC high 3% (diluted in 0.9% NaCl), chitosan
    SSP VI 4 + ~ - ~ +

    See also SSP capsule in PolyDADMAC Protocol in the "Protocol" section

    In our journey to synthesize scaffold forming cellulose sulfate, we were confronted with many challenges. More details are described in the engineering cycle.

    We achieved the best results with spruce sulfit pulp and 4 mol equivalents of SO3/Py. To test their stability, we put the empty scaffolds in distilled water and shook them over several days. They were still stable, but due to time constraints further tests and steps for optimization could not be conducted. Instead, we used the cellulose sulfate solution of the Cell-in-a-box kit from Austria Nova for the pore size test, which our system is based of.

    Pore size & Cell escape

    As a key part of our Friendzyme delivery platform, we chose to encapsulate our living cells into a scaffold functioning as a biocontainer. The main requirement to the scaffold itself is thus, that the bacteria remain inside while the enzymes and nutrients are able to freely diffuse in and out. The only way to achieve this is by finding an optimal pore size that allows both of these properties.
    We decided to demonstrate the functionality by assessing both the retention of cells (“Cell escape assay”) and the diffusion of small molecules (“Enzyme diffusion assay”).
    At the time we started the tests, we did not have our enzyme expressing lactobacilli ready. However, we need cells with an antibiotic resistance because some of our components were not sterile. Thus we decided to encapsulate some of our E.coli cells instead. They are slightly smaller in size than lactobacilli but if we were able to demonstrate that they were retained by the scaffold, automatically the same would be true for larger cells. (see "Protocol" section) The enzyme escape was tested separately to get better and clearer measurement results. (see "Protocol" section) For encapsulation, we used the Cell-in-a-Box kit from Austrianova as well as our own polyDADMAC solution with thiolated chitosan.

    Cell escape assay

    In the cell escape assay, solution 1 (kit) was first mixed with E.coli as described in the protocol, then dripped into the diluted solution 2 (kit) to encapsulate the bacteria. The resulting capsules were then placed in a PBS buffer. At certain time points, samples of capsules and of the supernatant were taken and plated out. After incubating the plates for 24 hours we got clear results: no cells grew on the plates of the supernatant. This indicates that the pores of the scaffolds were small enough so that the E.coli could not escape.

    Supernatant after 0 hours, 3 hours, and 24 hours

    We also performed tests to see if the scaffold would be harmful to the cells. For this, we mechanically broke down the scaffold, after down after 2 and 24h and plated out the insides. After incubation for 48 h at 37 °C, we saw that colonies were growing on the plates. Thus proving that the encapsulation was not harmful to the cells.

    Scaffold after 0 hours, 3 hours, and 24 hours

    The same test was performed with our own prepared solution (see "Protocol" section), where we mixed PolyDADMAC with thiolated Chitosan. Same as with the kit solution, no cells could be found in the supernatant. Most of the plates of the capsules were overgrown, but some of them were also empty. As soon as there was a little air bubble in the scaffold the solution could enter the scaffold. One reason for the instability could be the shape, which was more a drop instead of a sphere. Another reason is the molecular weight of solution 1, which did not match that well with our self-prepared solution.

    Enzyme diffusion essay

    We tested the enzyme escape by measuring the absorbance at 280nm, the absorption maximum of proteins, with a NanoDrop spectrophotometer.
    Instead of encapsulating enzymes we chose to try to encapsulate Bovine Serum Albumin (BSA) for cost and convenience reasons. BSA is larger than our enzymes, but if it was able to exit the scaffoldy our expressed enzymes would be as well.
    At first, we mixed the protein with solution 1 (containing cellulose sulfate) and dropped it into the diluted solution 2 (containing polyDADMAC) to form a capsule. It was then placed in PBS buffer. To measure the absorbance, we took samples of the supernatant of the capsules after 0 min, 30 min, 60 min, 120 min and 24 h. This was done in triplicates.

    Calibration curve of absorption of BSA dissolved in PBS at 280 nm
    Measured absorption of the supernatant of scaffolds without chitosan at 0, 30, 60, 120 min and 24 h (triplicates)

    We first created a calibration curve using BSA standards dissolved in PBS. When measuring our samples, the absorbance in the beginning was always very low but noticeably increased over time. We thus were able to show that proteins diffused out of the capsule. However, the results indicate quite clearly, that diffusion of BSA out of the capsule in the beginning occurs only very slowly. Almost all of the BSA is released after several hours of incubation. Thus, the timeframe over which diffusion occurs was too long for our envisioned application. This shows that further adjustments of the pore size will be necessary in the future.

    We concluded that the pore size is small enough to keep the bacteria from escaping, but not large enough to allow all of the proteins to escape. It is possible that the diffusion of our enzymes might occur faster than that of the larger BSA, however this needs further investigation using enzymes of the same size as the ones we want to secrete.


    We proved that the pore size of our scaffold was small enough to keep the bacteria inside and for some protein to diffuse out. However, the flux out of the scaffold was not really satisfactory. Due to the time restraint, we unfortunately had no room for further experiments, so some more pore size fine-tuning would be necessary in the future and also our actual secreted proteins should need to be used in order to assess the actual efficiency of protein diffusion.

    For the scaffold formed out of our own solution, additional experiments like the optimization of the concentration of the thiolated chitosan, are needed to achieve better results.

    Thiolation results

    One of the remarkable and innovative characteristics of our scaffold is its mucoadhesive property. It sticks to the mucosal surface of the human intestine thus prolonging the time it can remain there and release its therapeutic enzymes. This is achieved using a method called thiolation, where thiol groups are added which form disulfide-bridges with cysteine residues of the mucosa.
    This method was taught and tested in the laboratory of a leading expert on the field of thiolated polymers (“thiomers”), Prof. Andreas Bernkop-Schnürch of the University of Innsbruck. With the help and expertise of Ass.- Prof. Dr. Gergely Kali, we thiolated cellulose sulfate and chitosan with a microwave method using thiourea. The reaction only needed half a day and at the end no thiourea residue was found in the NMR.
    To test if our thiolated chitosan is harmful to our cells, we planned to perform cell toxicity tests on Caco2 cells. But, considering thiolated chitosan remains insoluble in water but needing it water-soluble for the testings thereafter, water-soluble and thiolated chitosan hydrochloride was produced. Sadly, due to the time limitiation, we were not be able to perform the toxicity test. More information regarding the testing of the toxicity of the created capsules in vitro using Caco-2 cells can be found in the "Protocol" section.
    As the cellulose sulfate was not commercially available, it was synthesized beforehand from microcrystalline cellulose. The plan was then to thiolate our synthesized cellulose sulfate.
    To test if our thiolation was successful, we used the Ellman’s test and disulfide bridge test with an UV/VIS with absorbance = 450 nm. The respective protocols can be found in the "Protocol" section.

    Calibration Curve - Thiolated cellulose sulfate Ellman's test

    Thiolated cellulose sulfate Ellman’s test:

    Sample absorbance Thiol concentration [mM]
    Average 0.07578 0.01529411
    Calibration Curve - Thiolated cellulose sulfate disulfide-bridge test

    'Thiolated cellulose sulfate' disulfid-bridge test:

    Sample absorbance Thiol concentration [mM]
    Average 0.053213 0.013965

    As we can see, the sample was slightly thiolated. Unfortunately, we noticed that the cellulose sulfate was desulfated, apparently due to of the microwave irradiation during thiolation. This rendered the cellulose water insoluble and the negative charge of the sulfate, which is needed for the forming of a polyelectrolyte complex, was lost. Even though the thiolation was successful, we could not use this cellulose.
    As an alternative, we chose low viscosity chitosan, which is a naturally abundant polysaccharide. It is biocompatible, positively charged and could be thiolated with an adjusted method. The respective protocols can be found in the "Protocol" section.
    After thiolation, it would be added to the also positively charged polyDADMAC solution, which forms the outer layer of our polyelectrolyte complex and thus can directly interact with the mucosal surface.
    Back in our BOKU laboratory, we used the newly learned method to thiolate chitosan with different parameters to optimize the thiolation. By using the Ellman’s test and disulfide bridge test, we evaluated if the reaction worked.

    Calibration Curve - Thiolated chitosan disulfide-bridge test

    Thiolated chitosan disulfid-bridge test:

    Sample absorbance Thiol concentration [mM]
    Thio_chitosan_1 0.051 0
    Thio_chitosan_2 0.0494 0
    Thio_chitosan_3 0.1538 0.374766
    Thio_chitosan_4 0.0643 0.039686
    Thio_chitosan_5 0.05507 0.005129
    Calibration Curve - Thiolated chitosan Ellman's test

    Thiolated chitosan Ellman's test:

    Sample absorbance Thiol concentration [mM]
    Thio_chitosan_1 0.0984 0.063191
    Thio_chitosan_2 0.07187 0.026507
    Thio_chitosan_3 0.06957 0.023327
    Thio_chitosan_4 0.0706 0.247511
    Thio_chitosan_5 0.0657 0.017976

    The most promising product was thio_chitosan_3 since disulfid-bridge test showed us, that thio_chitosan_3 had the highest concentration of disulfid-brides even as the Ellmann’s test showed us there were a a few thiol groups on every sample. There were more disulfide-bridges than thiol groups, which leads to the conclusion that intermolecular disulfid-bridges were formed during the thiolation of thio_chitosan_3. This of course, was highly undesired as we wanted our thiol groups to later engage with cysteine residues on the mucosa and not be already saturated by bonds with thiolated chitosan molecules.
    Thus, the unwanted bonds were cleaved in the next step using the Cleland reagent upon help of Ass.- Prof. Dr. Gergely Kali. To cleave the disulfide bridges, the sample was suspended in DMF, the Cleland reagent was added, and it was shaken over night at 30 °C. Afterwards it was precipitated in diethylether and washed with ethanol to remove the Cleland reagent. The finished sample was freeze dried.
    To test the most promising sample thio_chitosan_3, we then wanted to demonstrate mucoadhesion on a real mucosal surface. We did so by examining the viscosity of a thiolated chitosan with porcine mucosa using a rheometer and tensile studies with porcine intestine with a texture analyser.

    Texture analyzer - Tensile studies

    The texture analyser allowed us to show whether our thiolated chitosan had mucoadhesive properties. The freeze dried product and native chitosan was compressed into flat-faced disks. The disks were pressed against the mucosa layer of porcine intestine with a specific force for 90 seconds to form disulfid-bridges. Then the detachment force was measured.
    The respective protocol can be found in the "Protocol" section.

    Native chitosan from crab shells:

    Triplicate N N*sec
    1 0.036 0.890
    2 0.061 0.118
    3 0.071 0.605
    Avg. 0.056 0.538
    SD. 0.018 0.390

    Maximal detachment force (MDF)= 56.00 mN
    Total work of adhesion (TWA) = 53.77 mJ

    Native chitosan from crab shells:

    Triplicate N N*sec
    1 0.152 1.875
    2 0.160 0.949
    3 0.165 1.462
    Avg. 0.159 1.429
    SD. 0.007 0.464

    Maximal detachment force (MDF)= 159.00 mN
    Total work of adhesion (TWA) = 142.9 mJ

    As the tests showed, the adhesion work and detachment force of the thiolated sample is much higher than the native chitosan. We can thus conclude that the thiolation increased the mucoadhesive property of the chitosan.

    Rheological evaluation

    The rheometer tests the viscosity of our samples. We added suspended thiolated chitosan and native chitosan to porcine mucosa. They were incubated for 0 h and 3 h at 37 °C. When disulfide bridges are formed, the viscosity increases. This effect was indeed observed with our thiolated chitosan, thus showing that disulfide bridges are formed and mucoadhesivity is achieved.The respective protocol can be found in the "Protocol" section.

    Native chitosan from crab shells:

    0 h (Pas) 3 h (Pas)
    1 22.40 27.41
    2 21.03 64.75
    3 12.68 53.38
    Avg. 18.85 48.51
    SD. 5.22 19.14

    Modified chitosan:

    0 h (Pas) 3 h (Pas)
    1 62.94 170.70
    2 76.90 161.70
    3 81.67 236.40
    Avg. 73,84 189,60
    SD. 9,73 40,78

    As the tests showed, the viscosity of the thiolated sample increased over time and is much higher than the native chitosan. More disulfid-bridges were formed, this showed that the thiolation worked.

    NMR evaluation

    ssNMR of thiolated chitosan

    Since the thiolation of chitosan worked and through the solid-state NMR, we can see that there is no peak at exactly 180 ppm. A sharp peak there would indicate that the presence of thiourea residue and/or the intermediate product isothiouronium ion. Thus, we can see the synthesis of pure thiolated chitosan was successful.

    We thank our sponsors: