Engineering Success

Our team designed, built, tested, and re-designed genetic elements used to control gene expression in response to pH changes. We designed and built reporter constructs to test performance. We used DNA sequencing data and modeling to predict function and revisit design plans. We set design specifications for pH-sensor and adjustment circuits, and used what we learned though the testing of reporter constructs to feed into the design.


Since the goal of our project was to create a device that could sense and neutralize the pH of its environment in real time, our team began searching the iGEM parts repository for pH-responsive genetic elements. Our investigation led us to the acid-sensitive promoter pASR (BBa_K1231000) and alkaline-sensitive riboswitch Alx (BBa_K2348000). We noticed Alx included both the riboswitch and alkaline-sensitive promoter. We decided to redesign Alx to remove the promoter, so that transcription of this device could be constitutive, and not require induction. We obtained the sequence for PRE (pH Response Element), the riboswitch component of the Alx riboswitch and an improved version of the riboswitch, PREmR34, one where the native RBS was replaced with Bba_B0034 (1).

Before proceeding with the design of our base and acid producing circuits, we first wanted to test the functionality of these pH sensing genetic elements, and determine if they could work simultaneously within a chassis. To do this, we built reporter constructs using the pH responsive control elements that were used to drive expression of BFP or RFP (Figure 1).


Figure 1: SBOL diagrams of the reporter constructs used in this study


Construction and characterization of an acid-responsive DNA construct


After designing the acid-responsive DNA construct, pASR-RFP (BBa_K5097008), we added RCF10 prefix and suffix and had the DNA synthesized by IDT. We ligated the EcoRI/PstI-digested pASR-RFP gBlock into pSB1C3, and, transformed the products into E. coli DH5ɑ cells, screened clones using colony PCR, and then selected a positive clone to confirm by DNA sequencing.


Figure 2: Colony PCR confirms successful cloning of pASR-RFP into pSB1C3.


Acid-responsive promoter pASR fused to RFP was ligated into plasmid pSB1C3. Bacterial clones were subjected to colony PCR and analyzed by agarose gel electrophoresis. The ~1200 bp band in lanes 6 and 9 indicate the positive clones. Lanes 1 and 2 consist of positive and negative control PCR reactions that included the same primers and polymerase mix.

After confirming successful cloning of the pASR-RFP reporter, we performed a growth study to test expression of the reporter gene in response to varying pH levels. To do this, we grew the cells in LB supplemented with buffering agents as described in table 1 below.


Table 1: Buffering agents used to adjust the pH of LB for growth studies.


In parallel, we measured the OD595 of the culture and its fluorescence (Ex. 555 nm, Em. 596 nm). We then normalized the absorbance to cell density and calculated the change in this normalized value over the course of the experiment.


Figure 3: Normalized RFP expression from cells expressing pASR-RFP when grown at different pH values. The pH indicated was created by adding a buffering agent to LB to adjust the pH. UB, unbuffered LB.


These results demonstrate that, as expected, RFP is expressed at acidic pH (3.5 - 5), with maximal expression at pH 4.5. This result is consistent with experience reported by other iGEM teams using pASR. From this we concluded that pASR will function correctly, and can be used in our system as an acid-sensor to drive the expression of a base-producing gene.

Construction and characterization of an base-responsive DNA construct


We designed two base-responsive DNA reporter constructs, PRE-BFP (BBa_K5097004) and PREmR34-BFP (BBa_K5097005) so that we could test the claim that replacement of the native RBS within the riboswitch resulted in better translation (1). Prior to synthesis, we added the RCF10 prefix and suffix. We built the construct by ligating EcoRI/PstI-digested gBlocks into pSB1C3, and after transforming E. coli DH5ɑ cells with the construct, we tested screened clones using colony PCR, and then selected a positive clone to confirm by DNA sequencing.


Figure 4: Colony PCR confirms successful cloning of Alx riboswitch-BFP into pSB1C3. Alx riboswitch flanked upstream with a strong constitutive promoter and downstream with either the (lanes 1-5) weaker or (lanes 6-9) stronger m34 ribosome binding site were ligated into plasmid pSB1C3. Bacterial clones were then subjected to colony PCR and analyzed by agarose gel electrophoresis. The ~1200 bp band in lanes 1-6 indicate the positive clones


While the PREmR34-BFP construct sequencing confirmed that the construct was correct, we noticed that there were mutations within the PRE-BFP construct’s riboswitch segment. We selected new clones, sent them for sequencing, and again, noticed mutations present. In total, we sequenced at least 9 colonies, some were sent multiple times for sequencing, and each time, a significant number of errors were noted (Figure 5). When sequencing reactions were repeated on the same clones, we often received different results.


Figure 5: Multiple Sequence Alignment by CLUSTALW showing the results obtained for the PRE riboswitch clones using Sanger Sequencing (2). Dashes indicate missing nucleotides, N means unidentified nucleotide. Some samples were sequences twice, with the second sequence indicated as _1. WPS, whole plasmid sequencing


We reached out to IDT technical support, as they generously donated the synthetic DNA used in this work. Although the PRE riboswitch sequence was not flagged with a high complexity score upon submission, they couldn’t rule out the possibility that this particular sequence (which contains secondary structures) was difficult to construct.

We then considered the possibility that there is nothing wrong with the PRE riboswitch sequence, and instead there is something wrong with the interplay between the DNA sequencing technology and the riboswitch. The MGH CCIB DNA core facility, in their Sanger sequencing troubleshooting page on their website do indicate that regions with complex folding can interfere with Sanger sequencing reads (3). They go on to indicate that there are procedure variations involving alternate dyes that may help the results, or other sequencing technologies that can be employed that may generate better results. Indeed, when we selected a few clones to send for whole plasmid sequencing, we received consistent results (figure 6).

In the midst of efforts to identify a PRE-BFP reporter construct that did not contain a mutation, we decided to go ahead and test some of the clones to determine if they demonstrated base-responsive control of gene expression. We performed another growth study, only this time, using E. coli DH5ɑ cells transformed with the clones now designated PREm2-BFP, PREm13-BFP, and PREw9-BFP (figure 6). Again, we inoculated cultures and measured the OD595 of the culture and its fluorescence (Ex. 374 nm, Em. 446 nm). We then normalized the absorbance to cell density and calculated the change in this normalized value over the course of the experiment.


Figure 6: Whole plasmid sequencing results for PRE mutant clones m2, m13, and w9 selected for further study.


Figure 7: Normalized BFP expression from cells expressing PRE-BFP when grown at different pH values.The pH indicated was created by adding a buffering agent to LB to adjust the pH. UB, unbuffered LB.


One of the mutant clones, PREm13-BFP, demonstrated strong induction of BFP expression in response to alkaline pH, while the other mutants exhibited a significantly weaker response (figure 7). This indicates that the part is performing as expected, despite sequencing results suggesting that the riboswitch contains mutations.

We then decided to compare the activities of PREm13-BFP and PREmR34-BFP. The latter has an engineered ribosome binding site, and we wanted to determine if this would improve expression of our reporter. When we conducted another growth study, we were surprised to find that the PREm13-BFP reporter produced higher levels of BFP at alkaline pH than the PREmR34 with the alternative ribosome binding site (figure 8). Until we can repeat the growth study, we cannot rule out the possibility that the mR34 read at pH 9 was poor. Alternatively, since the ribosome binding site lies within the end of the riboswitch sequence (4), it is possible that changes within the mR34 RBS (relative to the m13 RBS) affect the secondary structure of the riboswitch and subsequent expression of the BFP.


Figure 8: Normalized BFP expression from cells expressing PREm13-BFP and PREmR34-BFP when grown at different pH values. The pH indicated was created by adding a buffering agent to LB to adjust the pH. UB, unbuffered LB.

Modeling the PRE riboswitch for functional insight

Sequencing results for the PRE-BFP colonies kept indicating the presence of mutations in clones screened, yet at least some mutants appeared to retain (or even improve function), while others appear to knock out pH responsive regulation of gene expression. Because of this, we decided to model the folding of the riboswitches to visualize what impacts, if any, the identified mutations might have on the folding of the riboswitch. We began by modeling the expected structure of PRE and comparing that against the engineered version PREmR34 (Figure 9). We observed no difference in the overall structure of the riboswitch, which is consistent with previously published results (1).


Figure 9: Structure of PRE and PREmR34 riboswitches. The sequences of these two iterations of the PRE riboswitch were modeled in RNAFold, and visualized in RNAsketch (4,5). No significant differences in the structure of the riboswitch are apparent, save for a longer stem structure in PREmR34, created by the replacement of the native RBS with B0034 at the end of the RNA, and the modification of the complementary region at the beginning of the riboswitch.


We next decided to model what effects the mutations identified in the PRE mutant clones might have on the structure of the riboswitch (figure 10). We selected the most complete sequences we generated using Sanger sequencing, and modeled the RNA structure using RNAfold (4). We observed that the m13 and m2 iterations of PRE retained the overall stem-loop structure of the PRE riboswitch. The w9 iteration however, displayed a distinctive riboswitch structure, with an additional stem loop present. These results suggest that major disruptions of the PRE riboswitch structure, as seen in PREw9 do appear to prevents pH-dependent induction of BFP expression as measured in (figure 7), but also that alteration of even a few nucleotides within the riboswitch, as seen in PREm2 in comparison to m13, can lead to major changes in the pH responsiveness of the switch.


Figure 10: Structure of PRE riboswitch mutants. The sequences of three clones identified by Sanger sequencing as expressing mutant PRE riboswitch were modeled in RNAFold, and visualized in RNAsketch (4,5). PREm13 and m2 structures highly resemble that of PRE (Figure 8), while PREw9 displays significant differences in several regions

pH Adjusting Construct Design


Based on the performance of the pASR and PREm13 pH responsive elements in testing, we felt confident in moving forward to begin design of a prototype of the neutralization circuits that are required for our device.

To remediate acidic wastewater, we planned to fuse the acid-responsive pASR promoter to genes encoding base-producing pathways. We decided on three specifications for the base producing system:

1. The chemical must be produced by a variety of cell types and be common in the environment.

2. The pathway must be streamlined, to make cloning more achievable without significant optimization,

3. The pathway should utilize a single, easy to obtain, and relatively inexpensive substrate

We first considered engineering the urease pathway into our device, as E.coli, and many other bacterial species with well characterized genomes are known to carry these genes (6). The urease pathway catalyzes the hydrolysis of urea into ammonia, a base produced naturally by many organisms. We thought that Helicobacter pylori would be a good source of the urease pathway, as this species is an acidophile, whose enzyme structure would be well adaptive to function in low pH environments (7). However, we revaluated this choice, when, in researching the urease pathway, we realized that there are eight enzymes required for this pathway, making the challenge of engineering it into a genetic circuit harder. We therefore sought out another pathway, and selected the putrescine synthesis pathway of E. coli as our base producing system (Figure 9). Two enzymes cooperate to produce the polyamine, putrescine, speA and speB (8). which is a molecule that acts as a base and is produced by many organisms.The circuit’s design involves constructing a bicistronic gene in which the pASR promoter is fused to an RBS and the speA and speB coding sequences.


Figure 9: The putrescine pathway. L-arginine is converted into putrescine in a two-step process catalyzed by speA and speB. Figure adapted from (9).


We also devised a strategy to generate an alkaline-sensing, acid-producing construct. To do so, we planned to fuse the best performing iteration of the PRE riboswitch, PREm13 to an upstream constitutive promoter, and an acid-producing gene. For this gene, we selected ldha, which encodes Lactate dehydrogenase. This enzyme catalyzes the conversion of pyruvate to lactic acid, which would then acidify alkaline wastewater to neutralize it before release


Figure 10: SBOL diagram of the acid-sensing, base-producing (pARS-PUT1) and base-sensing, acid-producing (PREm13-LDHA) circuits designed for this study.


Last, we also considered generating one more variation of the Alx riboswitch-containing construct that could have improved its activity. A 2022 publication from Stephen et al. (10) found that sequences downstream of the native Alx riboswitch dramatically increase its dynamic range. Whereas a gene containing the Alx riboswitch on its own is induced ~7-fold by alkaline pH, a gene including these additional downstream sequences (which amounts to ~100 more nucleotides) is induced 63-fold. Utilizing more of the native sequences may also reduce the possibility of unintentionally disrupting the structure of the riboswitch.

We devised plans to build acid and base-responsive constructs that could actually neutralize wastewater, but we were unable to complete their construction prior to the wiki freeze. We had just begun constructing the pASR-PUT1 using Gibson Assembly. Before beginning construction of PREm13-LDHA, we wanted to confirm the sequence of the riboswitch using a sequencing method other than Sanger sequencing. Once the structure and improved activity of PREm13 is confirmed, we will know if this is the iteration of the riboswitch that should be used to drive expression of LDHA.

References

1. Pham, Hoang Long et al. “Engineering a riboswitch-based genetic platform for the self-directed evolution of acid-tolerant phenotypes.” Nature communications vol. 8,1 411. 4 Sep. 2017, doi:10.1038/s41467-017-00511-w

2. Thompson, J D et al. “CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice.” Nucleic acids research vol. 22,22 (1994): 4673-80. doi:10.1093/nar/22.22.4673

3. “Sanger DNA Sequencing: Troubleshooting.” Center for Computational and Integrative Biology DNA Core, Massachusetts General Hospital, dnacore.mgh.harvard.edu/new-cgi-bin/site/pages/sequencing_pages/seq_troubleshooting.jsp#:~:text=intensity%20drops%20dramatically.-,What%20is%20the%20cause?,the%20rates%20by%20clicking%20here. Accessed 25 Sept. 2024.

4. Gruber, Andreas R et al. “The Vienna RNA websuite.” Nucleic acids research vol. 36,Web Server issue (2008): W70-4. doi:10.1093/nar/gkn188

5. Stefan Hammer, Birgit Tschiatschek, Christoph Flamm, Ivo L. Hofacker, and Sven Findeiß. “RNAblueprint: Flexible Multiple Target Nucleic Acid Sequence Design.” Bioinformatics, 2017. https://doi.org/10.1093/bioinformatics/btx263.

6. Mobley, H L et al. “Molecular biology of microbial ureases.” Microbiological reviews vol. 59,3 (1995): 451-80. doi:10.1128/mr.59.3.451-480.1995

7. Mobley HLT. Urease. In: Mobley HLT, Mendz GL, Hazell SL, editors. Helicobacter pylori: Physiology and Genetics. Washington (DC): ASM Press; 2001. Chapter 16. Available from: https://www.ncbi.nlm.nih.gov/books/NBK2417/

8. Boyle, S M et al. “Expression of the cloned genes encoding the putrescine biosynthetic enzymes and methionine adenosyltransferase of Escherichia coli (speA, speB, speC and metK).” Gene vol. 30,1-3 (1984): 129-36. doi:10.1016/0378-1119(84)90113-6

9. Morris, David R. et al. “Putrescine Biosynthesis in Escherichia coli” Journal of Biological Chemistry vol. 244,22 (1969): 6094-99. doi.org/10.1016/S0021-9258(18)63510-0.

10. Stephen, Christine, and Tatiana V Mishanina. “Alkaline pH has an unexpected effect on transcriptional pausing during synthesis of the Escherichia coli pH-responsive riboswitch.” The Journal of biological chemistry vol. 298,9 (2022): 102302. doi:10.1016/j.jbc.2022.102302