Team:TU Eindhoven/Design

Design

Design

Brainstorming

During the first brainstorming session about a possible project subject, we decided that we wanted to fight a worldwide problem. Already in an early stadium of brainstorming, antimicrobial resistance came to our attention. Some of our team members looked further into the subject and crossed the topic of bacteriophage therapy. This was then discussed with our supervisors and they pointed out that therapy is a very difficult project subject. Fascinated by the specificity of bacteriophages for their bacterial hosts, while at the same time stunned by the relatively deep ignorance regarding these amazing creatures in the scientific community, we decided to look into the use of phages for diagnostics.

Antimicrobial resistance

Misuse of antibiotics is one of the causes of the increase in antimicrobial resistance [1]. Broad-spectrum antibiotics are administered when it is not yet known which bacterial strain is causing the infection. The disadvantage of these antibiotics is the development of resistance, both in pathogenic and non-pathogenic bacteria that encounter these antibiotics [2]. Currently, it takes about 72 hours or longer to determine which bacterium is the source of an infection [3]. In these 72 hours, the patient is often treated with broad-spectrum antibiotics, which are not always effective and promote antimicrobial resistance. Therefore, there is a need for faster and specific detection methods to be able to administer effective antibiotics earlier, decreasing antimicrobial resistance.

Fast and specific detection method

Bacteriophages bind specifically to a bacterial strain and this interaction is useful for the specific detection of bacteria. When a bacteriophage binds to a bacterium, it inserts its DNA and replicates (Figure 1). It then takes a variable amount of time before the phage lyses the bacterium and the resulting number of progeny phages (the burst size) varies as well. Since we want to design a faster detection method, the lysis time should be minimal, while the replication rate should be high, so we are able to detect enough phages within a narrow timeframe. We researched the lysis time and replication rate for several phages that target pathogenic bacteria. These include the phages T7, T4, PhiX174 and Lambda for E. coli species. The lysis time of these phages lies between 10 and 60 minutes [4, 5, 6], making fast detection possible. The replication rate varies between the different phages. The T7 phage produces up to 100 progeny phages within 25 minutes [7], while the T4 phage produces up to 150-200 progeny phages in 25 minutes [8]. In contrast, the lambda phage produces about 100 progeny phages in 40 minutes [9]. As you can see, the T7 and T4 phages are the most efficient.



Figure 1: Bacteriophages infect bacteria, insert their DNA, replicate and lyse the bacteria.

During replication, the DNA is packed into the heads of these progeny phages. At first, we thought that we could use the free DNA that is not packed for our detection system. However, Dr. Jean-Paul Pirnay and Dr. Maia Merabishvili told us that it is not known how much free DNA is present and this amount can also vary in different phages. He advised us to detect the phage DNA present in the heads of the progeny phages. To extract the phage DNA, a capsid denaturation step must be introduced into the system before DNA detection, for which we use heat.

In summary, our design idea is to use specific phages that target specific bacteria. When these phages bind to a bacterium, the phage DNA is amplified, and many progeny phages are produced. Therefore, an increase in phage DNA can be detected when a specific bacterium is present.

DNA detection

Still, we needed to find a system that can detect DNA. Phages can contain either double-stranded DNA (dsDNA) or single-stranded DNA (ssDNA). Therefore, we needed to think of detection systems for both dsDNA and ssDNA. For ssDNA detection, a good system has been developed at our university; the BRET-beacon (Figure 2). This beacon can detect ssDNA using Bioluminescence Resonance Energy Transfer, so you can observe a change in wavelength when ssDNA is bound or unbound [10]. We would be able to use this system for detection of ssDNA.




Figure 2: Overview of the BRET beacon system. The system contains a handle-oligonucleotide to which the NanoLuc luciferase is bound and an anti-handle oligonucleotide to which Cy3 is bound. This anti-handle oligonucleotide contains a stem loop to which target ssDNA can hybridize. The handle and anti-handle oligonucleotides hybridize and, if no target ssDNA is present, light at the wavelength of Cy3 can be observed. When target ssDNA is present, a double-helix is formed that causes NanoLuc and Cy3 to be separated. This results in light at the wavelength of NanoLuc. Retrieved from Engelen, et al. (2017) [10].

For dsDNA detection, we wanted to use the specificity of the dCas9 protein. We explored several articles and found the research of Zhang, et al. in which a split Firefly luciferase fused to dCas9 is used for dsDNA detection [11]. We decided to use a split version of NanoLuc linked to dCas9 as a sensor. NanoLuc is a smaller bioluminescent protein that has better stability and a >150-fold higher luminescence than Firefly luciferase [12]. Also, the substrate has a lower background activity [11], which can be beneficial in many applications. NanoLuc uses the substrate Furimazine. Thus, by using NanoLuc, we intend to have a lower detection limit of DNA than the current method with the Firefly luciferase.

When we went to Brussels, we discussed the ssDNA and dsDNA phages with Dr. Jean-Paul Pirnay and he pointed out that most bacteriophages that target pathogenic bacteria are dsDNA phages. The T7 phage, which is a dsDNA phage, is the most common phage to use when targeting E. coli bacteria. Therefore, we decided to use the dsDNA detection system.

Paired dCas9-Split-NanoLuc for detection of dsDNA

dCas9 is a catalytically inactive form of Cas9 that was primarily used for transcription regulation. dCas9 contains a few mutations in the RuvC1 and HNH domains of Cas9 to make it catalytically inactive. It can still form a complex with guide RNA (gRNA) and bind to specific target DNA, but it does not cleave the dsDNA [13]. In our design, the wildtype sequence of dCas9 is used, which we retrieved from Addgene [14].

Using a (GGS)5 linker, we linked the dCas9 protein to Split NanoLuc or NanoBiT. NanoBiT contains two subunits: SmallBit of 1.3 kDa and LargeBit of 18 kDa [15]. These subunits weakly associate and form the bioluminescent complex NanoLuc, resulting in luminescence at a wavelength of 460 nm when the substrate is present [16]. Two dCas9 proteins are needed to create a bright signal for detection. One dCas9 protein is linked to the SmallBit and the other dCas9 protein is linked to the LargeBit. Both proteins bind in close proximity to a specific target sequence on DNA. When they do, SmallBit and LargeBit associate and emit light (Figure 3). The affinity between SmallBit and LargeBit can be varied between Kd = 0.7 nM and Kd = 190 µM [15]. For our design, we chose a Kd of 2.5 µM, which is a relatively low affinity. This low affinity should minimize background signal caused by spontaneous association when dCas9 does not bind to a specific DNA sequence.




Figure 3: The paired dCas9-Split-NanoLuc system emits light at 460 nm when the dCas9-SmallBit and the dCas9-LargeBit proteins bind in close proximity to a specific target sequence on dsDNA.

In our design, the SmallBit and LargeBit are connected C-terminal to dCas9 (Figure 4A & 4B), which is based on previous research [11]. The GGS linker in our system is relatively long, but the interspace distance between the two dCas9 proteins on the DNA can be varied to find the optimal distance between SmallBit and LargeBit. Using a long linker, we hope to be able to get a significant signal for a variety of interspace distances.




Figure 4: A) dCas9-SmallBit, B) dCas9-LargeBit

Functioning of paired dCas9-Split-NanoLuc

dCas9 binds to target DNA using gRNA, which is an RNA strand containing 20 bp complementary to the target sequence. Next to the target sequence, a protospacer adjacent motif (PAM) sequence must be present on the DNA strand. This is a 3 nucleotide sequence, 5’-NGG-3', that is firstly recognized by dCas9 when it starts scanning the DNA [17]. When two dCas9 proteins bind in proximity, four different orientations of the gRNA are possible: PAM-In, PAM-Out, PAM-Left and PAM-Right (Figure 5). We did not test all these options, but we based our orientation on the research of Zhang et al. (2017) and chose PAM-In for our design [11]. Next to PAM-In, we also tested PAM-Out to check whether this also worked for our sensor. We observed a slight decrease in signal when PAM-Out was used, so PAM-In is indeed a better orientation for our sensor.




Figure 5: The PAM orientations used for our dCas9-Split-NanoLuc sensor.

Our paired dCas9-Split-NanoLuc is a modular system since the sequence of gRNA can easily be altered to detect different dsDNA sequences. This modularity was integrated into a gRNA Finder Tool, which can be used to find gRNA sequences for any given target. The genomic sequence of any DNA genome can be entered, the tool will search for suitable gRNA sequences based on PAM sequences that are separated by a certain interspace distance. The measurement results show that the range of interspace distances that result in detectable signal is remarkably broad, meaning a wide variety of gRNA sequences can be selected with the gRNA search tool. Although the interspace distance is not a limiting factor, we optimized the interspace distance for our sensor. The optimal interspace distance was determined to be 70 bp.

The limit of detection of the paired dCas9-Split-NanoLuc sensor was determined using different sensor concentrations. It was found that the limit of detection for a 2 nM sensor is 10 pM of target DNA concentration with an interspace distance of 20 bp. In addition, we tested whether the limit of detection could be lowered when multiple sensor proteins bind to multiple recognition sites on one target. We tested what the effect of three sensor proteins is on the limit of detection. The results show that the limit of detection is indeed lower when three sensor proteins bind to three different recognition sites on one target DNA sequence, since the measured limit of detection is 5 pM.

dCas9-BRET for detection of dsDNA

In addition to the dCas9-Split-NanoLuc sensor, a dCas9-BRET sensor was designed to detect the dsDNA of the bacteriophages. A study from Sternberg et al. (2015) showed that dCas9 undergoes a conformational change upon DNA binding (Figure 6) [18]. This conformational change takes place in the HNH domain. We designed a sensor that can detect this conformational change.



Figure 6: The difference in conformation between the HNH domain (shown in yellow) in sgRNA/DNA-bound state and in docked state. Retrieved from Sternberg et al., 2015 [18].

This detection would be possible using Bioluminescence Resonance Energy Transfer (BRET). BRET involves the Förster resonance energy transfer between a luciferase as donor and a fluorophore as acceptor [18]. This energy transfer occurs when the donor luciferase and the acceptor fluorophore are in close proximity and when the emission spectrum of the bioluminescent donor overlaps with the excitation spectrum of the acceptor. Subsequently, the energy emission of the donor can be detected relative to the energy emission of the acceptor. The emission of the donor will decrease, while the emission of the acceptor will increase when the donor and acceptor are brought closer together. This ratiometric measurement is an advantage since the ratio of these energy emissions is less susceptible to variations in conditions such as temperature, concentrations and time compared to an absolute intensiometric measurement. In our design, NanoLuc is the bioluminescent donor and Cy3 is the acceptor fluorophore. Cy3 is a cyanine dye that has a maximum emission wavelength of ~570 nm. Upon binding of the target DNA, dCas9-BRET will undergo a conformational change that causes NanoLuc and Cy3 to be brought close together, resulting in BRET between NanoLuc and the Cy3 dye (Figure 7).




Figure 7: Schematic overview of the BRET principle. When donor and acceptor are separated, the donor NanoLuc will emit blue light. Upon target DNA binding, the donor and acceptor are brought in close proximity, resulting in energy transfer from donor to acceptor and therefore orange light emission.

To detect a conformational change using BRET, the donor and acceptor should be separated when DNA is not present (no BRET) and they should be in close proximity when DNA is present (BRET). In our design, the donor or acceptor should be on the HNH domain and the other one should be close to the HNH domain. Since the HNH domain undergoes a conformational change, we do not want to hinder this. Therefore, it makes sense to couple the acceptor Cy3 dye to the HNH domain instead of the relatively bulky protein NanoLuc. Cy3 is coupled to the HNH domain via maleimide coupling, which is done via a cysteine residue. Therefore, a cysteine was incorporated at a specific location on the HNH domain. In addition, two cysteines in the dCas9 protein were altered to serines to prevent coupling of Cy3 at undesired locations.

The donor NanoLuc would be coupled to one of the termini of dCas9. However, both termini of dCas9 are not close to the HNH domain and can therefore not be used for BRET. To create new N – and C-termini closer to the HNH domain, circular permutation (CP) can be used. In CP, the primary sequence of the protein is altered by coupling the original N – and C-terminus with a linker, while splitting the sequence at a different position [20]. In a study by Oakes et al. (2019), different circular permutants of Cas9 were designed and their functionality was tested [21]. We based our location of circular permutation on one of their circular permutants. The circular permutant we chose was permuted at a convenient location close to the HNH domain where dCas9 binds to DNA, and the permuted variant was still functioning well in the study by Oakes et al. (2019).

In our BRET sensor design, NanoLuc and Cy3 are separated when no DNA is bound. This means that no BRET occurs since the distance for energy transfer is too big. Therefore, it results in bioluminescence at the wavelength of NanoLuc (460 nm). When dCas9 binds to DNA, the conformational change causes NanoLuc and Cy3 to get in close proximity, which means that BRET takes place (Figure 8). Therefore, the energy of NanoLuc is transferred to Cy3, resulting in light emission at the wavelength of Cy3 (570 nm).




Figure 8: Functioning of the dCas9-BRET sensor. When no DNA is present, the Cy3 and NanoLuc are separated, which results in the emission of light at the wavelength of 460 nm. When DNA is present, Cy3 and NanoLuc are brought close to each other, so BRET occurs. This results in the emission of light at the wavelength of 570 nm.

The complete detection system: dCastect

The entire detection system, called dCastect, is a fast, specific and modular system for the detection of bacterial infections. The system starts with bacteriophage infection and ends with the detection of phage DNA (Figure 9). The bacteriophage binds to a specific bacterium, inserts its DNA and replicates to form progeny phages, effectively amplifying detectable DNA. Within a small timeframe, the cell lyses and the progeny phages are released. Subsequently, the progeny phages are heated at 70 °C to extract their dsDNA. This dsDNA is then detected using either the paired dCas9-Split-NanoLuc sensor or dCas9-BRET sensor.




Figure 9: Schematic overview of the dCastect system. It starts with bacteriophage infection. The phage then inserts its DNA, replicates and lyses the bacterium. This results in the release of progeny phages, which are subsequently heated to extract their DNA. This dsDNA can be detected using the paired dCas9-Split-NanoLuc or single dCas9-BRET system.

We have performed a proof-of-concept of this system using T7 phages, E.coli bacteria and our sensor. To determine the optimal conditions for these proof-of-concept experiments, we modeled the dynamics of the interaction between bacteria and bacteriophages. In the proof-of-concept experiments, a dilution series of a T7 phage stock with known titer obtained from the QAMH was prepared. A heating step was executed to extract the phage DNA and it was then tested whether our paired dCas9-Split-NanoLuc system could detect this phage DNA. The results show that the sensor can indeed detect phage DNA, which shows that the system functions as expected.

References

  1. English, B.K. & Gaur, A.H. (2009). The Use and Abuse of Antibiotics and the Development of Antibiotic Resistance. Hot Topics in Infection and Immunity in Children VI, 659, 73-82.
  2. Melander, R.J., Zurawski, D.V. & Melander, C. (2018). Narrow-spectrum antibacterial agents. MedChemComm, 9(1), 12-21.
  3. Kelley, S.O. (2017). New Technologies for Rapid Bacterial Identification and Antibiotic Resistance Profiling. SLAS Technology, 22(2), 113-121.
  4. Heineman, R.H. & Bull, J.J. (2007). Testing Optimality with Experimental Evolution: Lysis Time in a Bacteriophage. Evolution, 61(7), 1695-1709.
  5. Wang, N. (2006). Lysis Timing and Bacteriophage Fitness. Genetics, 172(1), 17-26.
  6. Young, K.D., Anderson, R.J.& Hafner, R.J. (1989). Lysis of Escherichia coli by the Bacteriophage ΦX174 E Protein: Inhibition of Lysis by Heat Shock Proteins. Journal of Bacteriology, 171(8), 4334-4341.
  7. iro, R., Molshanski-Mor, S., Yosef, I., Milam, S.L., Erickson, H.P. & Qimron, U. (2013). Gene product 0.4 increases bacteriophage T7 competitiveness by inhibiting host cell division. Proc Natl Acad Sci U S A, 110(48), 19549-19554.
  8. Abedon, S.T., Hyman, P. & Thomas, C. (2003). Experimental examination of bacteriophage latent-period evolution as a response to bacterial availability. Appl Environ Microbiol, 69, 7499-7506.
  9. Szybalski, W. (1974). Bacteriophage Lambda. In King, R.C. (eds), Bacteria, Bacteriophages, and Fungi. Boston, MA: Springer.
  10. Engelen, W., van de Wiel, K.M., Meijer, L.H.H., Saha, B. & Merkx, M. (2017). Nucleic acid detection using BRET-beacons based on bioluminescent protein-DNA hybrids. Chem Commun, 53(19), 2862-2865.
  11. Zhang, Y., et al. (2017). Paired Design of dCas9 as a Systemic Platform for the Detection of Featured Nucleic Acid Sequences in Pathogenic Strains. ACS Synth Biol, 6, 211-216.
  12. England, C.G., Ehlerding, E.B. & Cai, W. (2016). NanoLuc: A Small Luciferase is Brightening up the Field of Bioluminescence. Bioconjug Chem, 27(5), 1175-1187.
  13. Qi, L.S., et al. (2013). Repurposing CRISPR as an RNA-Guided Platform for Sequence-Specific Control of Gene Expression. Cell, 152(5), 1173-1183.
  14. Zuris, J.A., et al. (2015). Cationic lipid-mediated delivery of proteins enables efficient protein-based genome editing in vitro and in vivo. Nat Biotechnol, 33(1), 73-80.
  15. Dixon, A.S., et al. (2016). NanoLuc Complementation Reporter Optimized for Accurate Measurement of Protein Interactions in Cells. ACS Chem Biol, 11(2), 400-408.
  16. Hall, M.P., et al. (2012). Engineered Luciferase Reporter from a Deep Sea Shrimp Utilizing a Novel Imidazopyrazinone Substrate. ACS Chem Biol, 7(11), 1848-1857.
  17. Anders, C., Niewoehner, O., Duerst, A. & Jinek, M. (2014). Structural basis of PAM-dependent target DNA recognition by the Cas9 endonuclease. Nature, 513, 569-573.
  18. Sternberg, S.H., LaFrance, B., Kaplan, M. & Doudna, J.A. (2015). Conformational control of DNA target cleavage by CRISPR-Cas9. Nature, 527, 110-113.
  19. Xu, Y., Piston, D.W. & Johnson, C.H. (1999). A bioluminescence resonance energy transfer (BRET) system: Application to interacting circadian clock proteins. PNAS, 96(1), 151-156.
  20. Yu, Y. & Lutz, S. (2011). Circular permutation: a different way to engineer enzyme structure and function. Trends in Biotechnology, 29(1), 18-25.
  21. Oakes, B.L., et al. (2019). CRISPR-Cas9 Circular Permutants as Programmable Scaffolds for Genome Modification. Cell, 176, 254-267.