Team:Glasgow/Project/Switch

The Switch

The switch (BBa_K1463000) is the central part of the project, and the key to making our new system function. The idea is to place a promoter between two recombination sites: PhiC31 attB and attP. The two sites are in inverted (head-to-head) orientation so that expression of the PhiC31 integrase protein will flip the orientation of the DNA segment containing the promoter. Recombination between attB and attP produces two new sites (attL and attR) which are not recombined by the integrase, so the switching is unidirectional. We constructed the recombinase-activated switch using BBa_K1463501, a reversed version of the BBa_J23100 promoter. Prior to recombination, this will direct transcription of genes to the left of the switch. After recombination, genes to the right will be transcribed. We incorporated a terminator (BBa_B0010) to the right of the reverse promoter to prevent any read-through transcription of the genes we wish to stay OFF. We also incorporated a spacer BBa_K1463030 containing HindIII and BamHI sites, to make sure the att sites were far enough apart (>200bp) for efficient recombination and to allow us to monitor recombination by restriction digestion and gel electrophoresis.

Figure 1: How the Recombinase Switch works

The next section will cover in vivo testing of the switch.

Click HERE to jump to the Integrase section of the report.
Click HERE to jump to the analysis of the switch Promoter.
Click HERE to jump to the in vitro section of the report.

What else did we have to do? The genes to the left of the switch have to be transcribed from right to left. We therefore had to come up with our own standard for reversed open reading frame biobricks and design an efficient reverse ribosome binding site (RBS) biobrick.

• Made a reverse RFP biobrick (BBa_K1463520) from BBa_E1010 by using PCR to attach the standard prefix just after the bar code of BBa_E1010 and the standard suffix just before the ATG start codon. Standard biobrick assembly will then give a reversed version of the BBa_E1010 ORF.
• Tested the existing reverse RBS BBa_I742130 and our own more efficient reverse RBS (BBa_K1463560) designed using the Ribosome Binding Calculator of Salis et al. (see BBa_K1463560 for more details)
• Used B0034 upstream of E0040 GFP (recreating BBa_J85201) and cloned downstream of the BBa_K1463050 switch to give BBa_K1463001
• Cloned the reverse RBS-RFP construct upstream of BBa_K1463001 to produce our finished RFP - switch - GFP BBa_K1463000 biobrick
  This BBa_K1463000 biobrick was cloned into a low copy number vector with a pSC101 origin. The nature of a switch requires a low copy number within the cell both for efficient switching and to avoid ‘leakage’ of the signal. This leakage is caused by some RFP being transcribed when GFP should be or vice-versa. Therefore having fewer plasmids in a cell reduces the leakage problem.

Once created, the RFP –switch – GFP construct needed to be placed into the vector best suited for it. We needed to put our construct into a low copy number plasmid, as the efficiency of the switch flipping would increase. We decided on a plasmid with a pSC101 origin and carrying a kanamycin resistance gene, as the expression plasmid for the Integrase had chloramphenicol resistance. We didn’t have an empty plasmid available, so we had to modify an existing pSC101-based plasmid. After removing what the plasmid was previously used to carry, a biobrick compatible multiple cloning site was inserted. This MCS was of our own design, and consisted of the prefix and suffix, with a spacer in the middle containing a HinDIII site.

We next attempted to make the plasmid fully biobrick compatible. In the pSC101 origin region there was a SpeI site, and we wanted to remove this. There were also two PstI sites on either side of the kanamycin resistance gene, but it would have been difficult to remove these as they were flanked by long repeat sequences. We designed a mutagenic oligo to remove the SpeI site, and attempts were made to remove the site by site-directed mutagenesis. Unfortunately, this did not work, possibly because of the changes this rendered to the replication origin. A previous iGEM team had removed this site, but it resulted in the plasmid no longer having a low copy number, and instead had a great variability in the number of copies each cell contained. The change we attempted was different from the previously attempted one and was more conservative, but in the end it didn’t work. We were left with the problem of having a plasmid that we couldn’t cut uniquely with SpeI or PstI, as it would cut outside of the MCS.

Our solution was to cut the vector with EcoRI and XbaI in MCS, and ligate in the BBa_K1463000 switch biobrick cut with EcoRI and SpeI. Transformation into DH5α gave us a low copy number kanamycin resistant plasmid containing the RFP-switch-GFP BBa_K1463000 that we could use for in vivo assays.

φC31 integrase attP and attB sites (BBa_K1463040 and BBa_K1463041) within the recombinase switch require φC31 integrase to control switching. To make a φC31 integrase biobrick, prefix and suffix sequences were added to the integrase gene by PCR and it was cloned into PSB1C3. The integrase gene has an EcoRI site within it so in order to remove it a double-stranded oligo was designed to change the guanine at the beginning of the EcoRI site into a cytosine. This change did not alter the resulting protein. The vector was digested with PstI and EcoRI, removing the first 40 bases from the integrase gene. This digest was run on a gel before the vector and integrase fragments were extracted and purified. The double-stranded oligo was ligated to these fragments, reconstituting the integrase gene plus the prefix, making it a biobrick. The integrase gene was cloned downstream of an arabinose-inducible promoter in the plasmid pBAD33 (Guzman et al 1995 PMID: 7608087).

The directionality of φC31 integrase recombination can be reversed using the recombinase directionality factor, gp3 (Khaleel et al 2011 PMID: 21564337). This protein allows the integrase to recombine attL and attR, thus potentially reversing the directionality of our recombination switch. The φC31 gp3 gene was cloned into pSB1C3, but all attempts to clone it downstream of a BBa_J23100 promoter + BBa_B0034 RBS failed, so the gene was not tested for functionality.

pBAD-int containing the φC31 integrase gene under the control of the arabinose regulate promoter on pBAD33 was co-transformed into DS941 cells with the switch construct (BBa_K1463000) carried on the low copy number pSC101 vector. This allowed in vivo switching capabilities to be tested by restriction digestion of the switch DNA and by looking at the fluorescence produced. φC31 integrase was induced using the sugar arabinose. The arabinose inducible promoter on pBAD33 promoter is well described and tightly regulated and was therefore ideal as a proof of concept. In the future it is hoped that any inducible promoter could be used to switch on integrase expression only in the desired conditions.

The gel below shows what happened to the DNA when the cells were grown in glucose (no integrase) or in arabinose (integrase expressed). The experiment was carried out in triplicate, all worked equally well. There was no change in restriction pattern of the switch plasmid when cells were grown in glucose, while all of the switch plasmid changed to the expected restriction pattern when the cells were grown on arabinose.

Gel One:

Figure 2: Gel 1, Lane Experiment. Cells grown in glucose or arabinose

1. pBAD-int on its own
2. Switch #2 on its own
3. Switch #2 + pBAD-int glucose
4. Switch #2 + pBAD-int arabinose
5. Switch #3 on its own
6. Switch #3 + pBAD-int glucose
7. Switch #3 + pBAD-int arabinose
8. Switch #4 on its own
9. Switch #4 + pBAD-int glucose
10. Switch #4 + pBAD-int arabinose
11. pBAD33 gvpAC (ignore)
12. 1kb+ marker

All are cut with BamHI.
pBAD-int gives the biggest band about 7kb
The unrecombined switch gives two bands about 2.4 and 2.7 kb
The recombined switch gives two bands about 2.5 and 2.6 kb.

Gel 2:

Figure 3: Gel 2 showing supercoiled DNA

2. Switch #2 + pBAD-int glucose
3. Switch #2 + pBAD-int arabinose
4. Switch #3 + pBAD-int glucose
5. Switch #3 + pBAD-int arabinose
6. Switch #4 + pBAD-int glucose
7. Switch #4 + pBAD-int arabinose
8. Switch #2 on its own
9. Switch #3 on its own
10. Switch #4 on its own
11. pBAD-int only

This gel shows uncut supercoiled plasmid DNA from the same assay. Addition of arabinose leads to DNA inversion which does not change the size of the switch plasmid, so no change is seen. However a reduction in plasmid yield is noticeable in the presence of glucose.

The first fluorescence scan (040914-fluorescence below) shows the exact same  cells the DNA was extracted from in the same order as lane 2-11 on gel 2 above, and finally pBAD33 gvpAC as a negative control.
(red, green, red, green , red, green, red, red, red, almost black,  almost black). The two rows are just two identical 200 ul samples of each to show any pipetting errors or flecks of dust in the wells.

Figure 4:Fluorescence Scan of 200 ul samples of overnight cultures in a 96 well plate using a Typhoon FLA9500 scanner. Overlay of red and green fluorescent images using 532 nm laser and LPG filter and 473 laser and BPB filter respectively. (Scan 040914)

On the second scan (050914-fluorescence below), the top two rows are the same samples as on the first scan, but now one day old.

Figure 5: Fluorescence Scan produced as in figure 4. (Scan 050914)

The next two rows show the same samples now grow another 10 generations  (overnight) in glucose. The ones that had not recombined, still fluoresce red, the ones that had recombined stay recombined and fluoresce green. So the system "remembers" that it has been exposed to arabinose.

The final two rows are the same as the two above, but they have been grown overnight with no glucose. Everything remains the same but the fluorescence is a bit brighter than the glucose samples, probably  because glucose reduces the switch plasmid copy number (see gel). These results prove our concept. If this switch is placed between two other genes, e.g. with flagellar or other motility genes on the left and gas vesicle genes on the right, it should in theory turn off flagella formation when the recombinase is induced by any substrate, turning on vesicle genes. This would stop the cells swimming and float them to the top of the media.

In order to determine how efficient the switch promoter is, in comparison to the J23100 promoter without the att sites, a plasmid was generated with J23100 promoter upstream of the GFP with the B0034 RBS. The fluorescence of transformants containing this plasmid was compared with the fluorescence of transformants containing K1463000 (RFP/switch/GFP) exposed to integrase which fluoresce green (figure 6). Measuring the fluorescence levels revealed that the switch promoter is less efficient compared to the J23100 promoter. This suggests that the att sites and intervening sequences between the promoter and the RBS interferes with the strength of the promoter.

Figure 6:Relative fluorescence of Switch/GFP vs J23100/GFP

Figure 7: Relative fluorescence of switch/RFP vs J23100/RFP

To assess the switch, several rounds of recombination reactions were carried out in vitro. These were performed using plasmid DNA containing the switch and GFP. This construct can be seen in Figure 8. Two versions of this plasmid were used, one with GFP translated from a strong B0034 RBS (GFP34), and another with GFP translated from a weaker B0032 RBS (GFP32) (See parts K1463001 and K1463002).

Figure 8: Diagram showing the construct used for the in vitro characterisation of the switch.
Diagram shows the integrase switch containing a HindIII restriction site. Once exposed to integrase, the inserted DNA sequence results in the promoter (shown by a red arrow) to be orientated towards GFP, resulting in it’s expression. Also seen here is the HindIII site contained within the switch, which moves when the DNA is inverted. (See parts K1463001 and K1463002)

Figure 9: Fragment sizes obtained from restriction digest of plasmids exposed to different concentrations of integrase. The 34 and 32 denote different promoter strengths.
1. Switch & GFP34 – 8uM int. 2. Switch & GFP34 – 4uM int. 3. Switch & GFP34 – 0uM int.
4. Switch & GFP32 – 8uM int. 5. Switch & GFP32 – 4uM int. 6. Switch & GFP34 – 0uM int.
7. Switch & GFP34 – 8uM int. 8. Switch & GFP34 – 4uM int. 9. Switch & GFP34 – 0uM int.
10. Switch & GFP32 – 8uM int. 11. Switch & GFP32 – 4uM int. 12. Switch & GFP34 – 0uM int.

This DNA was mixed with a purified integrase protein, isolated originally from the bacteriophage phiC31, in a special buffer. This reaction was incubated at 30˚C for 2 hours. Next, some of this reaction was digested with the restriction enzymes PstI and HindIII. These were chosen because there is a HindIII site within the switch, which moves when the DNA inverts. PstI cuts in the biobrick suffix and was chosen as when cut along with this HindIII, visibly different fragment sizes are produced when run on an agarose gel (Figure 2).

Figure 2 shows the results of integrase reactions, which proves that the switch works. In the lanes exposed to integrase it can be seen that the DNA is present in both the inverted and non-inverted forms. The non-inverted only can be seen in the lanes where the DNA was not exposed to integrase. Though it is not 100%, it can be seen that it does switch.

This plasmid DNA can also be transformed into cells. In those which carry a switch in which the DNA has recombined, the promoter should now be in the correct orientation to drive GFP expression, and therefore should fluoresce. Those which carry a plasmid in the un-inverted form should not express GFP, and therefore should not fluoresce. Figure 3 shows colonies which have been transformed with plasmid DNA containing the switch plus GFP. It can be seen that some of the colonies are indeed fluorescing, but none of those containing DNA not exposed to integrase are (Figure 3: 9-12).

Figure 10: Fluorescing and non-fluorescing colonies. Image shows duplicated transformation of plasmid DNA into TOP10.
1. GFP34 + 8uM int. (1) 2. GFP34 + 8uM int. (2) 3. GFP32 + 8uM int. (1) 4. GFP32 + 8uM int. (2)
5. GFP34 + 4uM int. (1) 6. GFP34 + 4uM int. (2) 7. GFP32 + 4uM int. (1) 8. GFP32 + 4uM int. (2)
9. GFP34 + 0uM int. (1) 10. GFP34 + 0uM int.(2) 11. GFP32 + 0uM int. (1) 12. GFP32 + 0uM int. (2)

Figure 11: Short streaked colonies containing switch. 20 colonies from transformation plates short streaked out on each half of the plate. A diagram showing the experimental setup is also shown. Labelling on Short Streak plates show colonies selected to be taken on for further analysis. 4 colonies marked ‘Black’ which are expressing GFP, 4 colonies marked ‘White’ which are not expressing GFP, and 2 colonies marked ‘Grey’ which appear to be expressing GFP at a low level. Hereon after referred to as B1,2,3&4; W1,2,3&4; and G1 & 2 respectively.

Colonies were short streaked out, containing a mixture of both fluorescing and non-fluorescing colonies. These can be seen in Figure 4. Those colonies marked “Black” are ones which fluoresce, and which would therefore be expected to carry an inverted version of the plasmid. Those marked “White” are those which are not fluorescing, and would thus be expected to carry a non-inverted copy of the plasmid. There were some colonies which were obviously fluorescing, but not as brightly as the others, these are marked as “Grey”. We hypothesised that this was either due to the cell containing two different copies of the plasmid, or due to the plasmid being maintained at a lower copy number than the other colonies. The marked colonies were run on a gel to check this.

It can be seen in the gel below that all of the colonies marked B1-4 are in the inverted form as expected, and all those marked W1-4 are in the non-inverted form. This is as expected, and shows that the switch does indeed correctly switch in the presence of integrase, and that the now correctly orientated promoter does indeed drive expression of GFP. It can also be seen in the colonies marked G1 & G2 that both the hypotheses with how the greyer colonies could result appear to be correct. G1 shows the presence of two bands, indicating the colony had both an inverted and non-inverted plasmid. G2 on the other hand shows one band, indicating the presence of a flipped switch. As the DNA came from a less-brightly fluorescing colony, it could be the result of the plasmid carrying the switch being present at a lower copy number.

Figure 12: Gel showing Switch plasmids obtained from selected short streak colonies. The plasmids were extracted from the selected colonies of the short streak plates, and again digested with HinDIII and PstI. Lanes marked B1,2,3 & 4; W1,2,3 & 4 and G1 &2 correspond to the colonies marked as such in Figure 11

Alongside experiments to test the switch, the efficiency of two different ribosome binding sites were also tested. A strong RBS (B0034) and a weak RBS (B0032) were used in front of GFP (Parts K1463001 and K1463002 respectively). It can be seen that in cells exposed to the same concentration of integrase, but carry different ribosome binding sites, that there is a visible difference in the fluorescence (Figure 12). Having seen that the stronger RBS was indeed brighter and had no negative impact on the growth of the cells, it was incorporated into the final switch construct.

Figure 13: Comparison of promoter strength.
1. GFP34 + 8uM int. 2. GFP32 + 8uM int. (1) 3. GFP32 + 0uM int.