Engineering Success

Engineering Success

The goal of our project was to equip the probiotic strain E. coli Nissle 1917 with a genetic circuit that allows our genetically modified organism (GMO) to sense the inflammation in the gastrointestinal tract (GIT) of IBD patients and thereafter secrete therapeutics to counteract the inflammatory signalling. Elevated nitric oxide (NO) levels are a sign of inflammation 1 and are sensed by our GMO. The proinflammatory cytokine tumour necrosis factor (TNF-α) is a major driver of inflammation and already targeted by several IBD therapies. Our modified probiotic strain senses the inflammation via nitric oxide and secretes anti-TNFα single domain antibodies (nanobodies) that locally inhibit the inflammatory activity of TNFα and therefore reduce the inflammation. In order to achieve this final stage, we went through several engineering cycles:

  • Cycle 1:
    • Characterization of the nitric oxide sensor
    • Purification of 9 nanobody constructs and testing their functionality
  • Cycle 2:
    • Modification of the NO-sensing genetic circuit
    • Testing secretion of functional nanobodies
  • Cycle 3:
    • Combination of NO sensor with nanobody secretion

Cycle 1

Design

We started by designing the different components of our final genetic circuit. This comprises the 9 different nanobody constructs and the nitric oxide sensing circuit. The sequences for three anti-TNFα nanobodies were obtained from a patent. 2 We additionally designed bivalent constructs by combination of the three single nanobody candidates. This resulted in three monovalent, as well as three homogeneous and three heterogeneous bivalent nanobody constructs.
PnorVβ is an optimised nitric oxide sensitive promoter where the second integration host factor (IHF) was deleted from the promoter pNorV to exhibit better dose response at a low range of inducer (DETA/NO). 3 PnorVβ was then coupled to sfGFP and to a positive feedback loop with NorR, the corresponding transcription factor. Our plasmid maps are available here.

Build

The amino acid sequences of the nanobodies were converted to DNA and codon optimised for E. coli. We inserted our nanobody constructs into the expression vector pSBinit using Golden Gate assembly. The pSBinit vector already contains a myc-tag which is attached to the C-terminal end of the nanobody. We transformed the common lab strain E. coli MC1061 with the plasmid. The nanobodies were then expressed and purified via periplasmic extraction or whole cell lysis for bivalent constructs (see protocols).
Our nitric oxide sensing genetic circuit was inserted into a high copy number backbone (colE1, Twist Bioscience) and several primers were ordered for later Gibson assembly of various constructs. Furthermore, we designed and ordered a negative control plasmid as well as a control plasmid with pNorV instead of pNorVβ. All constructs were transformed and measured in E. coli Nissle 1917.

Test

Purified nanobodies were then tested for their ability to bind biotinylated TNFα with an ELISA (see Experiments). Successful capturing of TNFα by the nanobody results in a colour change.

Figure 1A: ELISA performed with all purified nanobody constructs. One bivalent nanobody (N7 = linked VHH#3E + VHH#2B) is missing. First four wells show the negative and postive control.
Figure 1B: Dose response curves of GFP/OD for the different plasmids to various DETA/NO concentrations, with pNorVβ with 2RBS being the most expressive promoter, while pNorV of the 2016 ETH iGEM team being the least expressive promoter.

Additionally, we were interested in developing an immunoassay to test the efficacy of the nanobodies on TNFα-stimulated immune cells. For this, we used the human monocytic cell line THP-1. Nanobodies were added to the cells, which were then stimulated by addition of different concentrations of human recombinant TNFα. After 24 hours, we isolated the RNA. Via quantitative real time polymerase chain reaction (qRT PCR) we compared the expression levels of the proinflammatory human IL-1β using as a control the housekeeping gene human GAPDH (see Experiments for more details). When nanobodies were added, we observed a reduction of IL-1β, indicating the successful inhibitory effect of our nanobodies on TNFα-induced monocytic immune response. Go to the Results for the corresponding graph.

Concurrently, we tested the induction of GFP by DETA/NO in the different plasmids. For this, we prepared several 96-well plates with the constructs we wanted to compare in E. coli Nissle 1917 and added different concentrations of DETA/NO. The OD600 and the fluorescence intensity were measured over the next 16 hours. We quickly realised that pNorVβ induced a higher response to NO than pNorV (Figure 1B), although the sensitivity of our construct did not reach the IBD relevant range.

Learn

We learn from our preliminary data that our nanobody constructs are able to bind TNFα and that the nitric oxide sensor is functional, responding well to a wide NO-range, although it would need to be further optimised. Additionally, the immunoassay with human monocytes gives us confirmation of the anti-inflammatory effect induced by our nanobodies upon TNFα-binding.


Cycle 2

Design

To further enhance the response of our NO promoter, we designed 3 different constructs that differ in the amount of ribosome binding sites (RBS) in front of the sfGFP and ordered the corresponding primers. This was done because our reference paper 3 developed a nitric oxide sensitive system with 3 ribosome binding sites in front of GFP and we wanted to see how the number of ribosome binding sites affected the output of our circuit. Additionally, we also wanted to test whether the removal of the feedback loop could improve the sensitivity of pNorVβ at low NO concentrations. For that, we designed and ordered further Gibson primers.

Cycle 1 demonstrated the functionality of our nanobodies, so we were ready for the next step: implementing the nanobodies in an inducible plasmid to test their secretion in our chassis, E. coli Nissle 1917. We decided to use the hemolysin A secretion system, a one-step secretion system originally isolated from uropathogenic E. coli strains. (4) Adding the HlyA secretion tag to the nanobodies allows their secretion and together with the myc-tag their detection in the bacterial supernatant.

Build

We assembled the construct by site-directed mutagenesis (quick change PCR) of our primary construct. This construct had one RBS, and plasmids with 2 and 3 RBS were produced. The rest of the plasmid remained unchanged. Removal of the NorR in constructs with 1, 2 and 3 ribosome binding sites was achieved by Gibson assembly. Once ready, the plasmids were transformed into E. coli Nissle 1917.

In the nanobody lab, we designed a high copy number plasmid (colE1, Twist Bioscience) that contains the inducible arabinose system pBAD for a controlled nanobody expression. The nanobodies were cloned into the vector by Golden Gate assembly and were fused with a myc-tag for detection and a HlyA-tag for secretion. Our plasmid maps are available here. The components for the secretion system (HlyB and HlyD) were integrated into a medium copy plasmid (p15A, Twist Bioscience) where a constitutive promoter controls their expression. For more information about the hemolysin A secretion system, go to the respective part BBa_K4387987. Our end chassis E. coli Nissle 1917 was then double transformed with both plasmids and the same was done with the lab strain E. coli MC1061 as a control.

Test

We prepared several 96-well plates of the plasmids with 1, 2, and 3 ribosome binding sites, with and without NorR in E. coli Nissle 1917 and added the same concentrations of DETA/NO as used previously. Next, we measured the OD600 and the fluorescence intensity for 16 hours. From earlier experiments, we had indeed observed that measuring over 16 hours was necessary to enter the plateau phase, where expression of GFP reaches constant levels.

The expression of nanobodies was induced by adding arabinose to the liquid bacterial cultures. Our transformed bacteria were then incubated at 37°C overnight. We validated the successful secretion of functional nanobodies by performing a Western blot of the supernatant we obtained from the liquid bacterial cultures. Additionally, we confirmed their functionality with an ELISA.

Figure 2A: Double transformed E. coli MC1061 were induced by arabinose and incubated at 37°C overnight. Anti-myc antibodies were used in the Western blot to detect secreted nanobodies in the bacterial supernatant. N1 = VHH#2B, N2 = VHH#3E, N3 = VHH#12B, N4 = biv. VHH#2B, N5 = biv. VHH#3E, N6 = biv. VHH#12B, N7 = linked VHH#3E + VHH#2B, N8 = linked VHH#2B + VHH#12B, N9 = linked VHH#3E + VHH#12B.
Figure 2B: ELISA comparing the TNFα-binding capabilities of secreted vs purified nanobodies obtained from MC1061. Adalimumab: positive control. Sybody against a membrane protein: negative control.
Figure 3A: Double transformed E. coli Nissle 1917 were induced by arabinose and incubated at 37°C overnight. Anti-myc antibodies were used in the Western blot to detect secreted nanobodies in the bacterial supernatant. N1 = VHH#2B, N8 = linked VHH#2B + VHH#12B.
Figure 3B: ELISA comparing the TNFα-binding capabilities of secreted vs purified nanobodies obtained from MC1061 and E. coli Nissle. Adalimumab: positive control. Sybody against a membrane protein: negative control.
Figure 4: Response of the different constructs to DETA/NO induction
Learn

Our data taught us that more ribosome binding sites increased the GFP expression, whereby the nitric oxide sensing circuit with two ribosome binding sites yields the highest response to NO induction. In this construct it was observed that the background levels of GFP expression were higher, meaning that this system is more leaky than the others. However, due to the high reponse of the two RBS construct, the background expression relative to maximum expression was not higher than other constructs. On the contrary, removing the feedback loop showed decreased response to No induction.
Additionally, we were able to successfully secrete monovalent and bivalent nanobody constructs in both E. coli strains and we learnt that secretion does not affect the nanobody’s ability to bind TNFα.


Cycle 3

Design

After characterizing the nitric oxide sensor and testing the secretion of nanobodies, we were able to combine everything and reach our final stage: testing the nanobody secretion triggered by NO. We took the plasmid with two ribosomal binding sites and exchanged the sfGFP with a nanobody that contains the myc-tag and HlyA-tag.

Build

We exchanged the sfGFP with a nanobody by Gibson assembly and double transformed again both E. coli strains with the secretion system plasmid and the newly synthesized plasmid containing a nanobody that is expressed upon NO-induction (for more details, see part BBa_K4387978).

Test

We grew liquid cultures of both strains and induced the secretion of the nanobody with DETA/NO at 37°C overnight. The supernatant containing the secreted nanobodies was then analyzed with a Western blot and ELISA.

Figure 5A: Double transformed E. coli Nissle 1917, able to secrete the monovalent nanobody VHH#2B, were induced with different DETA/NO concentrations and incubated at 37°C overnight. Anti-myc antibodies were used in the Western blot to detect secreted nanobodies in the bacterial supernatant.
Figure 5B: ELISA showing the TNFα-binding capabilities of the secreted monovalent nanobody VHH#2B upon NO induction, obtained from E. coli Nissle 1917. Adalimumab: positive control. Sybody against a membrane protein: negative control.
Learn

The first two bands of the Western blot in figure 5A showing the bacterial samples that have not been induced with DETA/NO are visible, indicating that the promoter is leaky. To investigate further, we compared the intensity of the bands that we received from the Western blot and could show that the highest induction with 2mM DETA/NO displays on average a 41% bigger intensity than the non-induced bands (see Results for more information). We can conclude that the induced nanobody expression is still significantly higher than background expression. A possible explanation for the leakiness might be the two ribosomal binding sites that follow the promoter, leading to an enhanced promoter activity but also to more leakiness. Additionally, the bacterial cultures were grown overnight for about 15 hours at 37°C, leading to a dense E. coli culture. It is possible that over time nitric oxide might have been metabolically processed by the bacteria and accumulated, leading to an increasing self-induction over this long period of time.


Plasmid maps:

Here you can find the maps of all our plasmids. The plasmids with two and three ribosome binding sites in front of sfGFP, as well as the plasmids without NorR were derived from piGEM2.



Future steps and limitations

We are able to show with these results that an induced secretion of functional anti-TNFα nanobodies produced by the probiotic strain E. coli Nissle 1917 is possible. We can now go further and modify the nitric oxide sensing promoter to become less leaky and even more specific to IBD-relevant NO ranges. Additionally, we can start working on biosafety aspects concerning the integration of the whole genetic circuit into the bacterial genome and the implementation of a kill switch. For more information on our biosafety strategy, go to the Project Safety page.


References:

  1. Qingdong Guan, 2019, “A Comprehensive Review and Update on the Pathogenesis of Inflammatory Bowel Diseas”
  2. Silence, Karen, Lauwereys, Marc, De Haard, Hans, et al. "Single domain antibodies directed against tumour necrosis factor-alpha and uses therefor", Int. Publication Number: WO 2004/041862 A2, 21 May 2004
  3. Xiaoyu J. Chen et al., 2021, Rational Design and Characterization of Nitric Oxide Biosensors in E. coli Nissle 1917 and Mini SimCells
  4. Ruano-Gallego, D., Fraile, S., Gutierrez, C. et al. Screening and purification of nanobodies from E. coli culture supernatants using the hemolysin secretion system. Microb Cell Fact 18, 47 (2019)