Overview
Chromoprotein DBTL
L-lactate detection DBTL

To realise our living diagnostic tool Colourectal, we did a lot of research. We did experiments regarding interaction, detection, signalling, modularity and biosafety. During all these experiments we went through multiple Design-Build-Test-Learn (DBTL) cycles. On this page we highlight two DBTL cycles we were particularly fond of. One highlights the DBTL cycle of the chromoprotein secretion results page, while the other highlights and combines the DBTL cycle of the L-lactate detection results and modelling page.

To see if our living diagnostic has detected colorectal cancer, a coloured protein from the chromoprotein (CP) family will be secreted and colour the stool. However, chromoproteins were not observed to naturally diffuse outside the bacterial cell [1], therefore a secretion system needed to be integrated into our design. By adding a signal peptide (SP) sequence to the C- or N-termini of a protein, the protein can be secreted.
Design
We began by researching in the literature which SPs were native to E. coli and were proven to work. We decided to test different SPs from the main secretion systems used for recombinant protein expression, these include the type I and type II secretion system [2,3]. In addition, we also tested the type VIII secretion system [4] as we could see from the literature that both type I and type II had some disadvantages, such as the sequences remaining attached to the protein and the accumulation in the periplasmic space respectively [5,6]. The following SPs from the different systems were thus tested (Table 1).
Table 1: The signal peptides tested and their properties.
   
Signal Peptide   
   
iGEM registry   
Type

Pathway

Protein state

Position
   
HlyA   
   
BBa_K554002   
   
Type I   
   
Unfolded   
   
C-terminus   
   
OmpA   
   
BBa_K208003   
   
SEC (Type II)   
   
Unfolded   
   
N-terminus   
   
PelB   
   
BBa_J32015   
   
SEC (Type II)   
   
Unfolded   
   
N-terminus   
   
TorA   
   
BBa_K1012002   
   
TAT (Type II)   
   
Folded   
   
N-terminus   
   
SEC-N22   
   
BBa_K2895007   
   
Type VIII   
   
Unfolded   
   
N-terminus   
Build
From our research, we built our first expression system. This system consists of an IPTG inducible promoter, pLac (BBa_R0010), followed by the RBS (BBa_B0034), the corresponding signal peptide (Table 1), CP (amilCP (BBa_K592009) , Ultramarine (UM) (BBa_K4244001), spisPink (BBa_K1033932) or anm2CP (BBa_K2387001)) and the terminator (BBa_B0015) .
Test
We could see from our plates and overnight cultures that our system was leaky as the colonies were coloured even though IPTG was not added and that spisPink did not exhibit any colouration (Figure 1). We also observed that due to the long maturation time the colour could take more than 24 hours before being visible.
Plate experiment with chromoproteins Figure 1: E. coli expressing chromoproteins on LB Agar plates supplemented with kanamycin after 36 hours (non-induced). From top-left to bottom-right: amilCP, spisPink, Ultramarine and anm2CP.
Learn
We took the time to study how the pLac promoter worked and found that lacI, a gene that codes a repressor, binds to the operator region of pLac and gets removed when IPTG is added. lacI is already present in the genome of EcN but due to the relatively high copy number of our plasmid, it cannot compete with all the operator sites and therefore expression happens without IPTG induction.
Design
To balance the ratio of repressor and plasmid we added the lacI to our backbone and repeated the assembly. This fixed the leakiness but the colours were not as intense as with the leaky plasmid upon IPTG induction. We then decided to change the expression system to test the signal peptide for secretion using a constitutive promoter and fluorescence proteins. This will give us an indication if secretion is possible in our strain although the secretion of CP will have to be confirmed by another mean as the amino acid sequence and tertiary structure affect the secretion efficiency of the signal peptide [7].
Build
The new expression system was assembled using a medium constitutive rubisco promoter, rbcL (BBa_K4244057) with its terminator (BBa_K4244058) and super folder green fluorescent protein (sfGFP) (BBa_K3143689). The promoter rbcL was chosen to avoid high levels of expression as high levels of expression might cause inclusion bodies clogging the channel [8]. In addition, chloroplast promoters are eubacterial in origin and hence compatible with gene expression in bacteria [9]. sfGFP is more suitable for secretion assays as it matures faster than CPs, and the fluorescence can be quantified more easily.
Test
A time series measurement was performed where three fractions, the supernatant, the pellet and the total were compared.
Plate experiment with chromoproteins Figure 2: Time series of total, pellet and supernatant fraction of sfGFP under different signal peptides expressed with a constitutive promoter. After 5 hours of growth, the fluorescence reached a plateau in the total and pellet fractions. No significant fluorescence was observed in the supernatant fraction for any of the signal peptides tested.
As observed in Figure 2, no fluorescence was detected in the supernatant over 8 hours of growth, signalling that secretion of sfGFP was not achieved with any of the constructs under a constitutive promoter.
Learn
We were not sure what could be the reason for the non-secretion of sfGFP but we hypothesized that the expression level caused by the constitutive promoter was still too high causing inclusion bodies [8] and clogging the protein channel. We looked for another promoter that could be induced and where the expression level can be controlled based on the amount of induction.
Design
Therefore we replaced the promoter by a rhamnose inducible promoter that was introduced in EcN ΔrhaB ΔrhaT to allow for better titration [10].
Build
The promoter pRha (BBa_ K914003) was thus used for the induciction, followed by the RBS (BBa_B0034), the SP, CP and the double terminator (BBa_B0015).
Test
An endpoint measurement was done after 5 hours of growth induced with rhamnose (500μM) (Figure 3), as we observed from Figure 2 that protein production reached a plateau after 5 hours. Different rhamnose concentrations (0, 25, 50, 100, 250 μM) were tested leading to the same conclusion that no fluorescence was detected in the supernatant (data not shown).
Plate experiment with chromoproteins Figure 3: Endpoint measurement of total, pellet and supernatant fraction of sfGFP under different signal peptides expressed with rhamnose inducible promoter. Measurement was performed 5 hours after rhamnose induction at 500 μM. No significant fluorescence was detected in the supernatant fraction for any of the signal peptides tested.
Learn & Conclusion
We still have to figure out why we do not see any secretion knowing that the signal peptides were shown to work in E. coli species [11–13]. One reason could be that it is strain specific and our strain, EcN, does not use the same system or cannot recognise the SP sequence from another E. coli. Another reason could come from our setup not being sensitive enough. To help with that a Western blot can be performed to see if any protein is present in the supernatant. A lot remains to be done but due to time, we were not able to perform all the experiments wanted and show the successful secretion of CP. However, if you want to have a closer look at our results please visit the results section.

To make our living diagnostic capable of detecting colorectal cancer, we set out to make it sensitive to two cancer biomarkers, one of which is L-lactate. In healthy colons L-lactate is present at concentrations ranging between 1.5 – 3 mM, while in colon tumours the concentration is higher ranging between 10 – 25 mM [1].

Design

Some bacteria, such as Escherichia coli can use lactate as a substrate for growth. The operon responsible for this is the lldPRD operon [2]. This operon contains a transcription factor called LldR capable of recognizing two operator sequences located upstream and downstream of the operon’s promoter. By binding to these operator sequences LldR represses the expression of lldPRD. However, if lactate is present, LldR changes conformation which results in activation of gene expression. We used this concept to develop a synthetic biosensor in E. coli Nissle 1917 (EcN) to sense L-lactate. A problem with the native lldPRD operon promoter, however, is that it is repressed by the presence of glucose and absence oxygen, which are conditions found in the colon [3]. Fortunately, Zúñiga et al. developed a lactate sensitive promoter which is not repressed by the above-mentioned conditions, called A Lactate Promoter Operating in Glucose and Anoxia (ALPaGA) [4].

Build

Using this concept we built a lactate-sensing genetic circuit (Figure 1). In this circuit a reporter gene, super folder green fluorescent protein (sfGFP) , was placed after the lactate inducible promoter ALPaGA, while LldR was constitutively expressed. A genetic construct like this should only show fluorescence when lactate is present.

Figure 1: Configuration of lactate sensing genetic circuit, the circuit consists of a transcription factor LldR (red), a lactate inducible promoter ALPaGA (A Promoter Operating in Glucose and Anoxia) and super folder green fluorescent protein (sfGFP) (Green). 1A When lactate is absent sfGFP is repressed. 1B When lactate is present sfGFP is expressed.
Test

This genetic circuit was then introduced into EcN, and the fluorescence output was measured at different L-lactate concentrations. From this we learned that EcN was able to sense L-lactate using such a genetic circuit since an increase in fluorescence with increasing L-lactate concentrations was observed. However currently, the biosensor is not able to differentiate healthy levels of lactate (1-3 mM) from levels associated with cancer (10-30 mM), highlighted as the green and red areas in Figure 2. This is because at both healthy and cancer associated levels of L-lactate significant fluorescence is observed.

Lactate Figure 2: The response of ALPaGA, expressed as OD600 corrected fluorescence output, at different L-lactate concentrations. The green and red areas signify the L-lactate concentrations in the microenvironment of healthy and cancerous colon cells, respectively. ** P ≤ 0.01 *** P ≤ 0.001.
Learn

Because of this we set out to find a way to change the dose response in such a way that it will not respond to healthy levels of L-lactate anymore. When diving into literature we found that there are many tools to regulate gene expression in bacteria, of which two examples are Clustered Regularly Interspaced Palindromic Repeats interference (CRISPRi) and antisense RNA (asRNA) [5–7]. We hypothesized that we could modify the operational range of the dose response from Figure 2 by adding these two tools to the genetic circuit.

Design

To test this hypothesis, we designed a CRISPRi-asRNA genetic circuit. This circuit is organized as follows: a constitutively expressed CRISPRi cassette represses sfGFP. The cassette consists of a deactivated CRISPR associated (dCas) protein and a single-guide RNA (sgRNA) complementary to sfGFP. Additionally, an asRNA complementary to the sgRNA would be made inducible by lactate. This can be made possible by placing the ALPaGA promoter in front of the asRNA and constitutively expressing LldR. Designing a genetic circuit in this way means that the reporter gene is repressed if lactate is absent but expressed when lactate is present. The initial hypothesis was that by changing the binding affinity between the asRNA-sgRNA complex, the activation efficiency would change. If the binding affinity was low, more asRNA, and thus more lactate, would need to be present to sufficiently counteract the CRISPRi cassette. This would thus result in the operational range shifting to higher lactate concentrations.

Lactate Figure 2: The response of ALPaGA, expressed as OD600 corrected fluorescence output, at different L-lactate concentrations. The green and red areas signify the L-lactate concentrations in the microenvironment of healthy and cancerous colon cells, respectively. ** P ≤ 0.01 *** P ≤ 0.001.
Build

Due to insufficient time, we were not able to finish building and testing the construct. Fortunately, the construct was modelled in silico and fitted to experimental data from the previous experiment as well as data from literature [7].

Test

The model predicted that only the sensitivity, meaning the slope of the dose response curve, would change when adding the CRISPRi-asRNA genetic circuit (Figure 3). The added CRISPRi and asRNA genetic components were not predicted to affect the operational range of the dose response curve. See Figure 4 for a comparison of the modelled dose responses. To find parameters that can change the operational range of the dose response, we performed sensitivity analysis.

Lactate Figure 4: Modelled dose responses for L-lactate of the genetic circuits with and without Clustered Regularly Interspaced Palindromic Repeats interference (CRISPRi) and antisense RNA (asRNA) in orange and teal, respectively. The green and red areas signify the concentration of L-lactate around healthy and cancerous colon cell microenvironments, respectively.
Learn

From the sensitivity analysis we learned that the parameter described as $k\_f_{LldRcomplex}$ was able to change the operational range, see Figure 5. This parameter describes the rate of activation of LldR. We hypothesized that, in biological terms, this rate is influenced by the intracellular L-lactate concentration. This means that, if we could control the internal concentration of L-lactate, we could control the operational range of the dose response. We expect we can control the intracellular concentration of L-lactate by knocking out or knocking in lactate dehydrogenases.

Lactate Figure 5: Modelled dose responses for L-lactate where the parameter k_f_LldRcomplex is increased or decreased by a factor of 10. The green and red areas signify the concentration of L-lactate around healthy and cancerous colon cell microenvironments, respectively.
Design

To test this hypothesis, we dove into literature to find what enzymes are responsible for the degradation of L-lactate in EcN. We found that EcN, like other E. coli strains, contains two genes responsible for L-lactate dehydrogenases, namely lldD and ykgF [8]. Besides, EcN contains another lactate dehydrogenase nldH, not found in other E. coli strains [9]. We designed a strategy to knock out these genes in EcN. If time had allowed, we wanted to test our lactate biosensor in this knock out strain and compare it to the biosensor in a normal EcN strain.

Conclusions

During our experiments regarding the detection of cancer biomarker L-lactate we went through multiple DBTL cycles. We did this by combining literature, experimental data, and modelling to test multiple hypothesis and gain relevant insights into our L-lactate biosensor circuit.