Project Description | TU-Braunschweig

Motivation

According to the German Society for Bipolar Disorder (DBGS), 2.5 million people in Germany are affected by bipolar [1]. Bipolar is characterized by alternating episodes of mania and depression and can greatly affect a patient's day to day life. The condition can be treated with lithium compounds, since they act as mood stabilizers. The therapy has shown high efficacy, particularly in preventing suicides[2]. Due to the risk of lithium intoxication, regular monitoring of lithium levels in the blood is necessary. In the first four weeks of therapy, this monitoring is done weekly[3]. An at home test system can provide an excellent alternative to time-consuming doctor visits and laboratory analysis of blood samples, which are required to avoid accidental under- or overdosing of lithium.

Summary

With our project Li+onSwitch, we aim to develop a testing system that will enable patients with bipolar undergoing lithium therapy to independently monitor their lithium levels.

The measurement of lithium holds significant importance for individuals with Bipolar, lithium administration is a therapeutic approach for this condition. Monitoring lithium levels in the blood is crucial, as therapeutic and cytotoxic lithium concentrations are closely intertwined, making accurate and reliable measurement challenging. Currently, regular blood tests conducted by doctors are the only means of monitoring lithium levels.

We present a novel measurement method for lithium, utilizing a lithium-sensing riboswitch. Riboswitches are mRNA elements capable of sensing small molecules and regulating gene expression accordingly. We aims to exploit the unique properties of the nhaA riboswitches described by White et al. (2022) to enable the specific and sensitive measurement of lithium ions.

The synthesis of the riboswitch is planned to be carried out in a cell-free system, ensuring precise control over the experimental conditions and simplifying the production process. To quantify lithium, the reporters LacZ/beta-Galactosidase, superfolder GFP, NanoLuc and mScarletI-3 are evaluated in conjunction with the riboswitches.

Solution

For the detection of lithium, we utilize the recently discovered Li+-II riboswitch that can sense lithium ions. In the biological context riboswitches function as a type of switch that can turn gene expression on or off. We utilize this riboswitch in our system to generate a signal in the presence of Li+.

The Li+-II riboswitch is a translational riboswitch. Translational riboswitches are encoded on the DNA (Figure 1 (A)). Upon transcription, they form a secondary structure in the 5‘-UTR of mRNA that masks the ribosome binding site. As a result the ribosome cannot bind and the mRNA is not translated (Figure 1 (B)). Once a ligand binds to the aptamer structure of the riboswitch, its structure changes, revealing the ribosome binding site and translation can occur (Figure 1 (C)).

Accordingly the nhaA-II riboswitch will activate the expression of a reporter system in response to lithium, indicating whether the lithium concentration has reached a certain level. The signal allows us to quantitatively determine the lithium level in samples.

Figure 1: Activation Mechanism of a Translational Riboswitch. Upon transcription, the riboswitch forms a three dimensional structure on the mRNA (A) that can be linearized when a ligand binds (B), thus enabling translation of the mRNA (C).

Riboswitch Composition

A riboswitch is an RNA construct with the ability to exert control over gene expression, either positively or negatively, upon binding to a specific ligand. These molecular devices operate at either the transcriptional or translational levels within the cell by forming two domains: An aptamer domain catalyzes the folding into a three-dimensional structure, thereby forming the ligand binding domain (Figure 1). The ligand binding domain exhibits extraordinary specificity and selectivity for certain ligands, which are often mono- or divalent metal ions but also non-metal ions or certain chemicals.

Why does our riboswitch detect lithium?

Our literature research led us to the publication of White et al. (2022) which describes two translational riboswitches believed to confer lithium toxicity resistance in certain bacteria (Figure 2). This made us think about other possible applications of the described riboswitches. Finally, it came to our minds that we could revolutionize the detection and quantification of lithium by integrating the lithium sensitive riboswitch into a reporter system. The riboswitch classes Li+-I and Li+-II are classes of riboswitches that can be found in many bacteria. The sequences that are shown are consensus sequences that White et. al. developed from several of these riboswitches.

These riboswitches have RNA sequences with conserved regions which form aptamers, thereby allowing ligand binding to regulate expression of downstream genes. These RNA motifs are often associated with nhaA genes in bacteria. NhaA proteins are membrane-localized Na+/H+ antiporters that also transport Li+, thereby enabling the bacteria to eject toxic Li+ from the cell.

Figure 2: Comparison of Li+-I and Li+-II Ribowitches. A shows the Li+-I riboswitch and B shows the Li+-II riboswitch sequence. Conserved regions are marked in red. C shows additional information. Source: White et al. (2022)

Decision On Riboswitch Motif

Because of this protein association, the riboswitches were called Li+-I (nhaA-I motif) and Li+-II (nhaA-II motif). The riboswitches show structural differences but similar activity. The researchers created consensus sequences based on sequence alignments from different bacteria (Figure 3) and were able to characterize them regarding specificity and activity.

Figure 3: The Li+-II riboswitch DNA sequence and two dimensional mRNA structure. A shows the DNA sequence of the riboswitch with conserved regions marked in red and the possible Lithium Binding Site in pink. B shows the mRNA sequence of the riboswitch in a two dimensional representation of the structure it takes on after transcription with conserved regions shown in red. Source: White et al. (2022)

For our project, we opted to focus exclusively on the Li+-II riboswitch due to concerns surrounding the Li+-I riboswitch: In the paper, the researchers suggested that slight nucleotide modifications might shift the specificity of the Li+-I riboswitch from lithium to sodium, which would lead to a disastrous outcome for our project. Furthermore, the Li+-I riboswitch exhibited structural similarities to a sodium riboswitch, strengthening our resolve to work solely with the Li+-II riboswitch (Figure 4).

Figure 4: Structural comparison of Li+-I riboswitch (A) and a sodium specific riboswitch. The structural similarities are obvious at first glance: The folding of the riboswitches is almost identical. The figure is only supposed to show the structural similarity but the colourful dots are explained in Figure 1C. Source: White et al. (2022).

Challenges

The function of riboswitches depends on multiple factors. Adjacent sequences can influence the folding and therefore the behavior of the riboswitch. We also plan to make the testing system applicable from home. Therefore, we need to develop a cell-free system which poses an additional challenge since it could affect the sensitivity and function of the riboswitch.

Furthermore, the riboswitch itself provides a qualitative signal. To detect the concentration of lithium in saliva, we need a suitable reporter system that is quantifiable. This system must function reliably and provide reproducible results, as toxicity begins at a blood lithium concentration of 1.5 mM.[3]

To tackle these challenges, we decided to test combinations of different riboswitch constructs and reporters. This would allow us to choose the reporter-riboswitch combination which characteristics fits the best to the requirements we have for our test system.

Construct Design

Our team introduces a new composite part consisting of a T7 promotor, the nhaA-II riboswitch and a reporter system. The general DNA and mRNA construct is shown in Figure 5.

Figure 5: Schematic Representation of the Lithium Quantification Construct.

We performed various tests with different reporter systems and different riboswitch variations. We used the reporters LacZ/beta-Galactosidase, superfolder GFP, NanoLuc and mScarletI-3 (Figure 6).

Figure 6: Schematic Representation of mRNA Constructs.The lithium-senstitve riboswitch is shown upstream of the beginning of an nhaA gene and a complete reporter gene.

In total, we had 4 different riboswitch constructs with 4 different reporters which lead us to 16 constructs that we successfully cloned by Golden Gate Cloning.

Construct R1

For our first construct, we cloned the riboswitch sequence upstream of the T7 promoter, followed by the cloning of a reporter gene downstream of the riboswitch. We chose the T7 promoter because it is known that the combination of the T7 promoter with the T7 RNA polymerase results in high translation efficiency.

An important insight into the design of the construct came from the expertise of Prof. Dr. Mario Moerl, who emphasized the importance of maintaining the natural environment of the riboswitch. This is crucial because it influences the folding and functionality of the riboswitch. We therefore made a conscious decision to retain the spacer sequence between the promoter and the riboswitch. In addition, we retained the spacer sequence containing the ribosome binding site (RBS) and the 5' region of the nhaA gene between the riboswitch and the reporter sequence (Figure 7). For this, we used the same sequence configuration that White et al. used in their experiments.

Figure 7: Our first construct consisting of T7 Promoter, Spacer Region 1, Li+-II Riboswitch, Spacer Region 2 and the beginning of the nhaA gene

These carefully considered steps in our experimental design were instrumental in preserving the integrity and functionality of the riboswitch, ensuring that our project was built on sound scientific foundations.

In addition to our original DNA construct we wanted to examine how changes to the construct affect the riboswitches performance. Therefore, we ordered three additional constructs:

Construct R2

Our construct R2 is missing the Spacer Region 1. We designed this to check if this negatively influences the riboswitches three-dimensional folding and thereby the sensitivity (Figure 8). This would be useful if we were to find out that the system is too sensitive and needs downregulation.

Figure 8: Construct R2 with deleted Spacer Region 2

Construct R3

In our construct R3, we exchanged the predicted RBS with a consensus Shine Dalgarno Sequence which is known to be a well working RBS (Figure 9). We did this to see if this influences the riboswitch folding and improves reporter gene expression. This could be useful if the system isn’t sensitive enough, which could be solved with a higher riboswitch concentration in the test system.

Figure 9: Construct R3 with Shine Dalgarno RBS

Construct R4

In our construct R4, we exchanged the predicted RBS with the g10T7 RBS sequence derived from the T7 bacteriophage (Figure 10). We also did this to see if we could improve reporter gene expression with this strong RBS for the same reasons as in construct 3.

Figure 10: Construct R4 with g10T7 RBS

Reporter constructs

As mentioned above, we choose four different reporters generating a fluorescence, luminicence or a visual signal to characterize in combination with our ribowitch constructs. You can find detailed descriptions of our reporters below.

sfGFP:

sfGFP is a constitutively green fluorescent protein with a short folding time. Its excitation maximum is at 487 nm, and the emission maximum is at 509 nm. It has a molecular weight of 26.8 kD. The folding time is 13.6 minutes.

Figure 11: Excitation (Ex) and emission (Em) spectra of sfGFP. Source: https://www.fpbase.org/protein/superfolder-gfp/

Genetic Origin

sfGFP belongs to the family of green fluorescent proteins. The protein originates from Aequorea victoria, a species of jellyfish found in the Pacific Ocean, and was published in 2005.

Adaptions by Our Team

With sfGFP, we have chosen an optimized version of GFP that is better suited for our project. It is crucial for our testing system to obtain results as quickly as possible. SuperfolderGFP is a particularly fast-folding variant that allows us to achieve this and is already widely used. For all our reporters, we have removed the start codons since we already have a start codon upstream in the nhaA gene. Additionally, we have added stop codons downstream.

Furthermore, we have adjusted the sfGFP sequence upstream and downstream to enable Golden Gate Cloning. Here, we utilize the ability of Type IIIS restriction enzymes to cut outside their recognition sequence, allowing for a seamless transition between two components of our plasmid. The sequence has been extended on both sides with the recognition sequence.

Figure 12: Sequence of our sfGFP construct

mScarlet-I3:

mScarlet-I3 is a constitutively red fluorescent protein with a short folding time. The excitation maximum is 568 nm, and the emission maximum is 592 nm. It has a molecular weight of 25.8 kD, and the folding time is 2.0 minutes.

Figure 13: Excitation (Ex) and emission (Em) spectra of mScarlet-I3. Source: https://www.fpbase.org/protein/mscarlet-i3/

Genetic Origin

mScarlet-I3 belongs to the family of red fluorescent proteins and was not isolated from an organism but was derived from a synthetic construct, originally derived from mCherry.[6]

Adaptions by Our Team

With mScarlet-I3, we have chosen an optimized version of mScarlet that is better suited for our project. It is important for our testing system to obtain results as quickly as possible. With mScarlet-I3, we have a variant that folds particularly rapidly, allowing us to achieve this, and it already has a wide range of applications. For all our reporters, we have removed the start codons, as we already have a start codon upstream in the nhaA gene. Additionally, we have added stop codons downstream.

Furthermore, we have modified the sequence of Scarlet-I3 upstream and downstream to enable Golden Gate Cloning. Here, we are utilizing the ability of Type IIIS restriction enzymes to cut outside their recognition sequence, allowing for a seamless transition between two components of our plasmid. To achieve this, the sequence has been extended on both sides with the recognition sequence.

Figure 14: Sequence of our mScarlet-I3 construct

NanoLuc:

NanoLuc is a bright blue luminescent protein with its associated substrate being Fumirazine, and it requires no cofactors. It has a molecular weight of 19.1 kD and emits light at a wavelength of 460 nm. NanoLuc was derived from OLuc, a luciferase found in Oplophorus gracilirostris. Its brightness is 100 times higher compared to other luciferases like firefly or Renilla luciferase, all while having a significantly lower molecular weight (19 kD). This makes it possible for NanoLuc to respond to even the slightest changes within a test system.

Furimazine reacts with O2 to form CO2 and Furimamide, resulting in the production of light.

Figure 15: Furimazine [9]

Figure 16: Furimamide [9]

Genetic Origin

NanoLuc is derived from OLuc, a luciferase found in Oplophorus gracilirostris, a deep-sea shrimp.

Adaptions by our team

For our testing system, it is crucial to have a reporter variant that does not require any excitation and is highly sensitive. Since NanoLuc is a luciferase and emits light signals, the expression can be easily measured without the need for excitation, unlike fluorescent proteins. NanoLuc is particularly interesting for our project because NanoLuc luciferase appears 100 times brighter than firefly or Renilla luciferase. Additionally, its molecular weight is only one-third (19 kD) that of firefly luciferase.

For all our reporters, we have removed the start codons since we already have a start codon upstream in the nhaA gene. We have also added stop codons downstream. Furthermore, we have adapted the sequence of NanoLuc upstream and downstream to enable Golden Gate Cloning. Here, we utilize the ability of Type IIIS restriction enzymes to cut outside their recognition sequence, allowing for a seamless transition between two components of our plasmid. To achieve this, the sequence has been extended on both sides with the recognition sequence.

Figure 17: Sequence of our NanoLuc construct

lacZ:

lacZ is a gene that encodes for ß-galactosidase. This enzyme is typically responsible for breaking down lactose into galactose and glucose, allowing for the metabolism of lactose to occur.

The enzyme occurs in the natural biology of Escherichia coli and plays a significant role in carbon metabolism. In the natural environment of E. coli, beta-galactosidase is involved in breaking down complex sugars, especially lactose, into their individual sugar units. This allows the bacterium to utilize lactose as a source of carbon and energy, which becomes particularly relevant when other preferred carbon sources are scarce. The expression of beta-galactosidase in E. coli is regulated by the lac operon, which is activated when lactose is available as an external sugar source.

As long as no lactose is present in the vicinity of the cells, a repressor binds to an operator region, blocking the formation of downstream enzymes.

However, when lactose is present, it can bind to the repressor, preventing it from inhibiting the formation of enzymes. As a result, the enzymes are only produced when lactose is present and the formation of the enzymes is beneficial.

The natural occurrence and function of beta-galactosidase in E. coli have contributed to its use as a reporter enzyme in molecular biology studies, establishing a valuable connection between its biological function and its application as a research tool.

Usage in Our Project

In our experiments, we initially employed X-Gal (5-bromo-4-chloro-3-indolyl-beta-D-galactopyranoside) as a substrate for beta-galactosidase. What's special about X-Gal is that it produces a blue dye when cleaved. The dye produced during the metabolism of X-Gal belongs to the family of indigo dyes, the very same dyes used in the production of blue jeans. The formation of this dye as a colorimetric indicator in beta-galactosidase assays requires the enzymatic action on two X-Gal molecules.

Figure 18: X-gal reaction. Source: https://en.wikipedia.org/wiki/X-gal#/media/File:X-Gal_reaction.png

X-Gal is a chromogenic substrate that, when cleaved by beta-galactosidase, generates a blue precipitate. This blue coloration allows for straightforward visual detection of gene expression and provides an immediate and qualitative signal of promoter activity. However, it is important to note that X-Gal's color change has a rather binary outcome, making it less suitable for precise quantitative measurements.

To achieve quantitative measurements of beta-galactosidase activity, we transitioned to using o-nitrophenyl-beta-D-galactopyranoside (ONPG) as our substrate in the assay. ONPG is advantageous in quantitative assays due to its ability to produce a yellow-colored product upon hydrolysis by beta-galactosidase. This yellow coloration can be quantified by measuring the absorbance at approximately 420 nm (OD420), offering a continuous scale for assessing enzyme activity.

Miller units are a common metric used in the assay to quantify the activity of beta-galactosidase. This unit of measurement is named after its developer, Jeffrey H. Miller, who introduced it as a means to express the enzymatic activity of beta-galactosidase in a quantitative manner. In the ONPG assay, Miller units are calculated by measuring the rate of hydrolysis of ONPG by beta-galactosidase and normalizing it to the time of reaction and the amount of cell culture used. The formula for Miller units typically involves the change in absorbance at 420 nm (OD420) over time, corrected for the optical density at 600 nm (OD600). OD550 is a correction factor which allows for light-scattering by E. coli correction.

Miller units = 1000 x [OD420 - (1.75 x OD550)/t x v x OD600]

t is the total time of the reaction expressed in minutes and v is the volume of culture used in the assay

To perform these enzymatic assays safely and in compliance with laboratory safety standards, we conducted the reactions under a fume hood. This was particularly important because certain reagents, such as beta-mercaptoethanol and chloroform, used in the assay are toxic and require proper ventilation to minimize exposure to hazardous fumes.

Instead of using a plate reader for absorbance measurements, we opted for a conventional spectrophotometer. This choice was made due to the ease of fitting the spectrophotometer under the fume hood, ensuring a safe environment while obtaining accurate OD420, OD550, and OD600 readings. This approach allowed us to conduct quantitative measurements with precision and in a manner consistent with laboratory safety protocols.

During our experiments involving the ONPG assay, a significant challenge arose. It was observed that E. coli BL21(DE3) naturally possesses the Lac Operon, which can lead to the expression of beta-galactosidase and subsequent metabolism of the substrate even in the absence of the permissive conditions provided by the riboswitch. This inherent genetic feature of E. coli BL21(DE3) posed a considerable hurdle, necessitating a workaround to ensure accurate and controlled experimentation. To address this issue, we determined that the experiment could only be reliably conducted in a knockout strain such as E. coli KRX (Promega), which lacks the Lac Operon, thus allowing for precise regulation of beta-galactosidase expression and substrate utilization under the influence of the riboswitch. This strategic choice of bacterial strain was instrumental in achieving the experimental control essential for our research objectives.

Figure 19: BL21(DE3) and KRX cell cultures, with and without X-Gal. From left to right: KRX without X-Gal, KRX with X-Gal, BL21(DE3) without X-Gal, BL21(DE3) with X-Gal. Blue color indicates ß-Galactosidase activity.

Genetic Origin

lacZ is a gene of the lactose operon of Escherichia coli and encodes for the enzyme ß-Galactosidase. The lac operon serves to regulate enzymes necessary for utilization, including ß-Galactosidase. The enzyme occurs in the natural biology of Escherichia coli and plays a significant role in carbon metabolism. In the natural environment of E. coli, beta-galactosidase is involved in breaking down complex sugars, especially lactose, into their individual sugar units. For the design, we used the genetical sequence of the lacZ-Gene, found in Escherichia coli K12, provided by the National Center for Biotechnology Information (NCBI).

Adaptations by Our Team

For our testing system, it is important to have a reporter variant that does not require any stimulation to be used in regions with insufficient technical support. lacZ expression, in combination with X-Gal or ONPG, is easily visible to the naked eye and can enable this. With all our reporters, we have removed the start codons since we already have a start codon upstream in the nhaA gene. Additionally, we have added stop codons downstream. Furthermore, we have adjusted the lacZ sequence upstream and downstream to enable Golden Gate Cloning. Here, we utilize the ability of Type IIIS restriction enzymes to cut outside their recognition sequence, allowing for a seamless transition between two components of our plasmid. The sequence has been extended on both sides with the recognition sequence. Due to the high sequence length, the reporter could not be sequenced in one piece. Therefore, we divided the reporter into two parts (lacZ_a and lacZ_b), which we were able to assemble in Golden Gate cloning thanks to the additional interfaces for the Type IIIS restriction enzymes between the parts.

Figure 20: Sequence of our lacZ_a construct

Figure 21: Sequence of our lacZ_b construct

Cell-Free Gene Expression System

A cell-free expression system is an in vitro method of gene expression to perform transcription and translation without the presence of living cells.

A key consideration for us, both in terms of safety and technical feasibility, is the avoidance of providing genetically modified organisms to end-users. Additionally, the cell-free gene expression system (CFE) offers distinct advantages for our research objectives. One notable advantage is the ability to tailor the reaction environment, thereby expediting the translation process. This adaptability allows us to fine-tune the conditions for optimal protein synthesis, achieving faster and more efficient results compared to in vivo methods.

Furthermore, the cell-free system circumvents a significant hurdle in our research. In our specific context, it eliminates the need for lithium to first enter the bacterial cell before being recognized by our riboswitch constructs. By bypassing the necessity for lithium uptake by living cells, the cell-free system streamlines the process, allowing us to focus directly on the desired molecular interactions and regulatory mechanisms, ultimately enhancing the precision and effectiveness of our experiments.

In our specific application, we opted for an E. coli lysate-based cell-free system. Since our previous experiments were carried out exclusively in E. coli strains. This led us to believe that using a CFE based on the same bacterium would lead to the fewest complications in the transfer of the experiment. Crucially, the E. coli strains used in this process already contain the T7 RNA polymerase, a vital component for gene expression.

After evaluating the riboswitch-reporter constructs mentioned above, we evaluated the two most promising candidates in the cell-free system. You can find more information on our Results page.

Sources

Deutsche Gesellschaft für bipolare Störungen e.V. (DGBS). (2023, April 19). dgbs.de. https://dgbs.de/bipolare-stoerung/verlauf
Song, J., Sjölander, A., Joas, E., Bergen, S. E., Runeson, B., Larsson, H., Landén, M., & Lichtenstein, P. (2017). Suicidal Behavior During Lithium and Valproate Treatment: A Within-Individual 8-Year Prospective Study of 50,000 Patients With Bipolar . American Journal of Psychiatry, 174(8), 795–802. https://doi.org/10.1176/appi.ajp.2017.16050542
Lewitzka, U., Haußmann, R. Lithium - ein Update. InFo Neurologie 24, 44–53 (2022). https://doi.org/10.1007/s15005-021-2201-4
Parkin, G. M., McCarthy, M. J., Thein, S. H., Piccerillo, H. L., Warikoo, N., Granger, D. A., & Thomas, E. A. (2021). Saliva testing as a means to monitor therapeutic lithium levels in patients with psychiatric disorders: Identification of clinical and environmental covariates, and their incorporation into a prediction model. Bipolar Disorders, 23(7), 679–688. https://doi.org/10.1111/bdi.13128
White, N., Sadeeshkumar, H., Sun, A. et al. Lithium-sensing riboswitch classes regulate expression of bacterial cation transporter genes. Sci Rep 12, 19145 (2022). https://doi.org/10.1038/s41598-022-20695-6
Gadella, T.W.J., van Weeren, L., Stouthamer, J. et al. mScarlet3: a brilliant and fast-maturing red fluorescent protein. Nat Methods 20, 541–545 (2023). https://doi.org/10.1038/s41592-023-01809-y
Superfolder GFP at FPbase, Lambert, https://www.fpbase.org/protein/superfolder-gfp/
mScarlet-I3 at FPbase, Lambert, https://www.fpbase.org/protein/mscarlet-i3/
England, Christopher G., et al. “NanoLuc: A Small Luciferase Is Brightening up the Field of Bioluminescence.” Bioconjugate Chemistry, vol. 27, no. 5, American Chemical Society, Apr. 2016, pp. 1175–87. https://doi.org/10.1021/acs.bioconjchem.6b00112.
X-gal https://en.wikipedia.org/wiki/X-gal#/media/File:X-Gal_reaction.png