Introduction

The support for our project stands not on the shoulders of giants but rests on the four pillars that we have defined as follows; Microbiome analysis, Nanobody secretion, Nanobody efficacy, and Biocontainment.

Microbiome analysis

DNA Extraction

To test whether lung samples from Dutch poultry actually contained the Limosilactobacillus reuteri strains, we conducted a lung microbiome analysis using lung animal-by-products from Dutch poultry. At first, we used a DNA extraction protocol which relied on the homogenization of our lung tissue before extraction. After numerous troubleshooting steps and optimization, we concluded that the chicken DNA present in the sample was present in significantly different higher amounts, which would block efficient primer binding in our quality control PCR resulting in no bands. Subsequently, we hypothesized that diluting the sample would make the concentration of microbial DNA so low, that the PCR would also not work.

We continued with an alternative protocol, such that we would still be able to gain insights regarding the microbes that live in the poultry lung microbiome. The alternative protocol we developed relied on growing our lung samples in LB medium overnight, meaning we would still be able to identify which microbes are present in the chicken lungs, but we would not be able to say anything about composition ratios.

For the purposes of our research (identifying presence, rather than total composition) these results would still be valuable. We conducted our second DNA extraction protocol (growth based) for 9 lung samples and one negative control (LB medium), and followed up with a quality control of our samples as described in our optimized quality control PCR for microbiome analysis found on our experiments and protocols page under microbiome analysis protocols.

Figure 1 below shows clear bands for all samples around the 450bp mark as anticipated. We used primers for the hypervariable V3/V4 region of microbial 16S rRNA, indicating that our DNA extraction samples contained DNA that could be amplified using these primers. This indicates that our microbiome DNA extraction was successful, and we have extracted microbial DNA which could be applied for downstream sequencing. However, since our negative control also showed a band at the same size, we suspected that the LB medium we had used was contaminated with microbial DNA. To exclude that our samples might have been contaminated during DNA extraction, we grew the remaining, theoretically sterile, LB medium at 37 ℃ overnight. After noticing growth we concluded that our LB was indeed contaminated. This could suggest that the bands seen for each sample is a result of contamination. Even though contamination was observed, we still considered sequencing the samples due to time constraints.

Figure 1. Quality control for microbial DNA extracts from the chicken lung samples acquired through growing the lungs in LB medium overnight. 1,6% (m/v) agarose gel showing the V3/V4 hypervariable 16S rRNA region (~450bp) amplified PCR products of our DNA extraction samples.

Figure 2 below shows the nanodrop results we acquired from the aforementioned microbial DNA extracts, showing widely varying purities and final DNA concentrations. Our negative control appears to have very low concentrations of DNA present, indicating that there must have been other bacteria in our lung samples resulting in the increased concentrations. Pressed for time, we decided to continue to Next Generation Sequencing analysis using the microbial DNA extracts, as these results could still confirm the presence of L. reuteri in our samples. As seen in figure 2, we had obtained a high enough concentration (>5 ng/μL per sample), enough DNA in our total sample (>100ng), and had very clear bands around the 450 bp mark on our 1,6% (m/v) agarose gel (figure 1), we concluded that these samples would be sufficient to send for downstream 16S rRNA Next Generation Sequencing to identify whether we could identify L. reuteri in our lung samples.

Figure 2. Quality control for microbial DNA extracts from the chicken lung samples acquired through growing the lungs in LB medium overnight. Results show the concentration of DNA extracted in the nine lung samples + negative control. Also the 260/280 and 260/230 absorbance ratios are provided as indication of sample purity.

Next Generation Sequencing:

We sent our DNA samples for 16S rRNA Next Generation Sequencing (Illumina NovaSeq6000) to a third party (Biomarker Technologies (BMK) GmbH, Münster, Germany). As of now, we are still waiting for the results which we should get any day now. We are looking forward to presenting them at the Grand Jamboree!

Nanobody secretion

Within the nanobody secretion pillar we focused on the ability of L. reuteri to produce and secrete nanobodies. With this pillar we will show one element of our proof of concept, which is to have L. reuteri produce and secrete nanobodies.

Growing L. reuteri under anaerobic conditions

When we received L. reuteri DSM 20016 we decided to make initial growth curves at 37 °C in a 96 wells plate, such that we can more easily predict growth over time for future experiments. However, when growing L. reuteri under aerobic conditions little to no growth was observed. This meant that, since L. reuteri is a facultative anaerobic organism, we first had to develop a method to reliably grow L. reuteri in our lab. This was necessary because we did not have an anaerobic chamber.

We developed a method using mineral oil on top of our L. reuteri cell cultures in order to slow down the diffusion of oxygen into the MRS medium to levels the cells find acceptable. When growing the cells with mineral oil on top, as shown in figure 3, more standard growth curves were observed. These growth curves show that we are now able to reliably grow L. reuteri in our lab, using a simple and inexpensive method. From the growth curves it can be observed that growing from an overnight culture the cells reach the exponential growth phase relatively faster (after ~3-7 hours) compared to growing L. reuteri from a glycerol stock (after ~13 hours).

Figure 3. Growth curves of L. reuteri DSM 20016 at 37 °C in anaerobic conditions by adding mineral oil on top of the cell culture. L. reuteri cells were grown overnight at 37 °C in MRS medium. From the overnight culture cells were diluted 10 times (yellow), 50 times (red), 100 times (brown), 200 times (green), and 20 times from a glycerol stock (blue). The diluted cells were placed in a 96 wells plate, to which mineral oil was added on top. The 96 wells plate was kept in the plate reader at 37 °C, where it was read every 5 minutes at 600 nm.

Transforming L. reuteri through electroporation

Our next challenge was to transform L. reuteri DSM 20016 by means of electroporation. Being able to insert a plasmid into L. reuteri is a foundational step in our proof of concept regarding the nanobody secretion pillar. Literature has shown DSM 20016 is able to be transformed, which prompted us to reproduce the protocols. However, these initial attempts showed no colonies, bringing us back to the bread and butter of science; optimization. The most important alteration to our electroporation protocol was to include 2% glycine to the cell growth medium. Glycine is a widely used cell wall weakening agent for the transformation of gram positive bacteria.

The results in figure 4 show a band at D1 at the expected size ~1200 bp. Here, the pYTK001 vector containing GFP was used for transformation. The cells were electroporated at 2,0 kV with all other variables as described in our protocol. The vector itself was plasmid purified from E. coli JH101 in order to combat DNA methylation incompatibilities between E. coli and L. reuteri, as hypothesized by the iGEM Oxford 2019 team. Due to time constraints we were not able to investigate further, although the presence of a band at the expected size gives a strong indication that we were able to transform L. reuteri DSM 20016, albeit with a low transformation efficiency.

Figure 4. Colony PCR of L. reuteri transformed under varying electroporation conditions. Cells were grown at 37 °C to OD600 ≈ 0,7 in MRS medium supplemented with 2% glycine and 0,5 M sucrose. Cell cultures were concentrated 100 times in 30% PEG 600 as electroporation medium. To the concentrated cells 250-500 ng of plasmid DNA was added, after which the cells were electroporated at either 2,0 or 2,5 kV. The electroporated cells were immediately incubated in MRS medium for 4 hours at 37 °C. After incubation the cells were plated with 10 μg/mL of the appropriate antibiotic. Resulting colonies were subjected to colony PCR, where samples A1,A2, and C1-D3 were transformed with the pYTK001 vector containing GFP (~1200 bp expected size) and B1-B3 were transformed with the pTRKh2 vector containing the R1A-A5 nanobody (~1900 expected size).

Cells were grown at 37 °C to OD600 ≈ 0,7 in MRS medium supplemented with 2% glycine and 0,5 M sucrose. Cell cultures were concentrated 100 times in 30% PEG 600 as electroporation medium. To the concentrated cells 250-500 ng of plasmid DNA was added, after which the cells were electroporated at either 2,0 or 2,5 kV. The electroporated cells were immediately incubated in MRS medium for 4 hours at 37 °C. After incubation the cells were plated with 10 μg/mL of the appropriate antibiotic.

Resulting colonies were subjected to colony PCR, where samples A1,A2, and C1-D3 were transformed with the pYTK001 vector containing GFP (~1200 bp expected size) and B1-B3 were transformed with the pTRKh2 vector containing the R1A-A5 nanobody (~1900 expected size).

Nanobody efficacy

Within the nanobody efficacy pillar we focused mainly on producing and purifying the anti-avian influenza nanobodies as described in literature [1]. We constructed several monovalent and bivalent nanobodies, which we then cloned into the pTRKh2 expression vector and transformed into E. coli BL21 cells.

Production and purification of anti-avian influenza nanobodies

In total, we managed to construct four different nanobodies, which we could produce in either E. coli DH5⍺ or BL21. The monovalent R1A-A5 and R1A-B6 were successfully cloned into E. coli BL21, while the bivalent R1A-B6 and monovalent R1A-B6 linked to monovalent R1A-A5 were cloned into E. coli DH5⍺; due to time constraints the two bivalent nanobodies were not (successfully) transformed into E. coli BL21.

The transformed cells were grown at 37 °C in 1 L LB medium to an OD600 = 0,5 ± 0,2, after which the cultures were supplemented with 1 mM IPTG as protein expression inducer. The induced cultures were incubated at 30 °C overnight, followed by sonication and His-tag purification with Ni-NTA agarose. The His-tag purification elution fractions were used for Western blot analysis to prove if we can produce the nanobodies of interest. The results in figure 5 show a clear presence of his-tagged proteins at the expected molecular weights in a broad range of elution fractions for all four nanobodies; this is with the exception of the bivalent R1A-B6 nanobody (figure 5D), where we expect a 30,79 kDa band, yet observe a band at approximately half of that. In the wash out fraction it can be seen that we lose some amount of nanobodies during the purification, thus lowering our overall nanobody production yield.

Figure 5. Western blot analysis of our nanobodies. (A) R1A-B6 monovalent nanobody with an expected molecular weight of 15,46 kDa. (B) R1A-A5 monovalent nanobody with an expected molecular weight of 15,13 kDa. (C) Bivalent nanobody consisting of R1A-B6 linked (head-to-tail) to R1A-A5 using a (gly-gly-gly-gly-ser)6 linker; expected molecular weight of 30,13 kDa. (D) Bivalent nanobody consisting of two monovalent R1A-B6 nanobodies linked (head-to-tail) together by a (gly-gly-gly-gly-ser)6 linker; expected molecular weight of 30,79 kDa. To detect the nanobodies we used an anti-6xhis antibody with an HRP conjugate. For all four blots a 70 kDa polyhis-tagged protein was used as a positive control.

Antigen-specific reactivity of anti-avian influenza nanobodies to hemagglutinin with ELISA

The purification of our nanobodies from the cell free extracts by means of his-tag purification proved to be a challenge. Because our nanobody yield is relatively low there is binding competition to the Ni-NTA agarose, significantly reducing the purity of our nanobodies. Since we were unable to successfully purify the nanobodies we decided to pool the elution fractions and use this for our nanobody efficacy assays.

To investigate the antigen specificity of the monovalent and bivalent nanobodies we decided to perform an ELISA experiment on almost the complete range of known influenza subtypes, meaning we went beyond the scope of previous literature. We ordered the hemagglutinin proteins listed in table 1 from BEI resources, which we received as his-tagged protein solutions. Because of the presence of this his-tag on our hemagglutinin antigens, we were not able to use our anti-6xhis HRP conjugate antibody as used previously in our Western blot results. For this reason we opted to use an anti-myc HRP conjugate antibody, which will bind to the C-terminal myc-tag present in all our nanobodies.

Table 1. Scope of avian influenza hemagglutinins used for ELISA experiments. The range encompasses H1 to H11 with the exception of H6 and H8 subtypes. All hemagglutinin proteins originate from an influenza A virus. Some hemagglutinin proteins originate from a pdm09 strain, which stands for “Pandemic Disease Mexico 2009”, colloquially known as swine flu.

The ELISA results, as displayed in figure 6, show little to no binding to any hemagglutinin, even to the ones who have shown to bind in literature. The absorbance values we found are significantly lower compared to typical binding values found in literature. Considering this low absorbance is also true for our positive control, it is most likely a technical error somewhere in our ELISA protocol. To optimize our ELISA protocol to scientific standards is beyond the scope of our iGEM timeframe. For future ELISA optimization experiments we are looking at increasing the concentration of hemagglutinin on the 96 wells plate. On top of this, we aim to coat the hemagglutinin on the 96 wells plate in a bicarbonate buffer, which could, due to the buffers higher pH, increase the rate of hydrophobic interactions between amino acid side chains and the plastic surface of the plate [2]. Furthermore, the nanobody production yield requires further optimization, such that protein purification becomes more reliable. As of now, we are aiming to implement the suggested optimizations and we are looking forward to presenting them at the Grand Jamboree!

Figure 6. Anti-avian influenza nanobody reactivity to a wide range of avian influenza hemagglutinins. Each hemagglutinin subtype was coated on the 96 wells plate at 1 μg/mL in PBS overnight at 4 °C. The coated wells were subjected to the pooled elution fractions of his-tag purified nanobodies. Nanobody binding to hemagglutinin was measured at an absorbance of 450 nm using Benzene-1,2-diamine (OPD) as chemiluminescent substrate. For the positive control a well was coated with H7N9 hemagglutinin, which was subjected to rabbit serum against H7N9 (BEI resources NR-48765). Then a secondary anti-rabbit alkaline phosphatase conjugate antibody was used to measure the binding. For the negative control an H7N9 coated well was subjected to PBS instead of a nanobody solution. The monovalent R1A-B6 and R1A-A5 nanobodies are shown in grey and yellow, respectively, while the bivalent R1A-B6 and R1A-B6 linked with R1A-A5 are shown in orange and blue, respectively.

Biocontainment

The Nanobuddy project involves the development of a therapeutic L. reuteri which can be seeded into the lung mucus of chickens. Such an in situ deployment of a bacterium requires tight biocontainment, ensuring that none of the bacteria escape into the environment. We thus developed a 2-input OR-gated killswitch, designed to be capable of activating a CRISPR-Cas9 kill-mechanism. The kill mechanism gets activated when either the temperature drops below a set number or when there is exposure to light. The specific design-choices and assembly strategy can be found on the engineering page.

For brevity we will not show all the steps and results from the cloning of all the constructs, but in the end we successfully assembled the plasmids for the testing of individual promoters, the TetR inverter, and the temperature sensors.

Temperature sensor testing

As outlined in the experiments and the engineering pages, we performed 96-well microtiter plate-based growth and fluorescence experiments on the successfully constructed strains. Here we pipetted 200 µL of our strains with a dilution of approximately 0.01 OD600 into M9 minimal medium with glucose and ampicillin, into the 96-well plate. These plates were then incubated in a TECAN-100 -or a SYNERGY HTX MM microtiter plate reader.

The gain settings of the plate reader were optimized and set to 100. The runtime was optimized and set to 10 hours and measurement frequency was set to every 2.5 minutes. As optimization was time consuming, each measurement had slightly different settings and the results yielded are not very comparable. However, they do indicate some trend and functionality of the constructs as can be seen below.

The first construct of interest is BBa_K4233041, where the J23100 promoter controls mRFP1. This construct was transformed into DH5α and BL21, and tested on fluorescence over time. The fluorescence was first normalized to the M9 blank, followed by division by OD600 per time-point to determine fluorescence/OD600. The result is shown below in figure 7.

Figure 7. Normalized fluorescence (575/620) / OD over time (hours) of the BBa_4233041 construct, consisting of a J23100 constitutive promoter powering mRFP1, in BL21.

These results (gain: 255, em/ex: 575/620, runtime: 5 hours, measurement frequency:3 min.) show that there are high levels of fluorescence from this construct quite consistently, with an initial drop possibly due to the organism’s fluorescence level adapting to the lower temperature after growth at 37℃. As the mRFP1 slowly degrades older proteins. This data is quite promising for an initial experiment, however, as it shows the construct is functional with relatively consistent expression. Though it also appears that the temperature also significantly affects expression here. We recommend future experiments with more temperatures, and to compare BL21 with DH5α, where consistent run-times, measurement frequency and other parameters are used.

The same experiment was also performed using other parameters (gain: 200, em/ex:575/620, runtime:5 hours, measurement frequency: 2.5 min) and at 37℃. The construct here was only available in DH5α, which exhibited low growth, which caused normalization to drop the relative fluorescence to below 0. This prevents direct quantitative comparison with the previous dataset. However, the general trend is visible, which shows an increase in fluorescence per cell as is expected unlike the previous results. This can be explained by the fact that the initial growth temperature is the same as the experimental temperature, so the cells had already been expressing at the correct temperature.

Figure 8. Normalized fluorescence (575/620) / OD over time (hours) of the BBa_K4233041 construct, consisting of a J23100 constitutive promoter powering mRFP1, in DH5α.

Finally, two of our temperature sensors were also tested, pTlpA1-36C (BBa_K4233045) and pTlpA2-36C (BBa_K4233049). These were almost identical, except with differently designed promoters (Mk1, the original from the paper and Mk2, see the engineering page), the latter of which was hypothesized to function more effectively. In this experiment the parameters were (gain: 200, em/ex: 575/620, runtime: 5 hours, measurement frequency:2.5 min) at 37℃. The constructs were transformed in DH5α, and the results are shown below in figure 9.

Figure 9. Normalized fluorescence (575/620) / OD over time (hours) of the BBa_K4233045 and BBa_K4233045 constructs, consisting of TlpA-based temperature sensor mechanisms (Mk.1 and Mk.2 respectively) powering mRFP1, in DH5α.

This data appears to have the same issue as the previous graph, where the low cell-growth and the relatively strong M9 background fluorescence causes the normalized fluorescence to drop below 0. The same recommendation repeats here, to optimize cell-growth. Additionally it is possible to use a standard, like a constitutive promoter construct used on every plate to determine fold-change compared to that standard, or to do the same with a serial dilution of the Interlab fluorescent dyes. This would allow for results to be compared more effectively between different plates.

As for this dataset on its own, it appears that maximum promoter activity is the same for both promoter variants, though the original pTlpA1 promoter speeds up more quickly.

For the temperature sensor mechanism itself, since the temperature is only one degree above the inhibition limit, this data is not reliable. But it seems like the repression is not present at this temperature at the very least. To make a conclusion about TlpA repression and temperature sensing, more temperatures above and below the repression limit need to be tested.

References

[1] Hufton SE, Risley P, Ball CR, Major D, Engelhardt OG, Poole S. The breadth of cross sub-type neutralisation activity of a single domain antibody to influenza hemagglutinin can be increased by antibody valency. PLoS One. 2014 Aug 1;9(8):e103294. doi: 10.1371/journal.pone.0103294. PMID: 25084445; PMCID: PMC4118869.

[2] What is the difference between bicarbonate and PBS as a coating buffer in ELISA assay? AAT Bioquest, https://www.aatbio.com/resources/faq-frequently-asked-questions/What-is-the-difference-between-bicarbonate-and-PBS-as-a-coating-buffer-in-ELISA-assay