The biological realization of CryptoGErM consisted of three
subprojects: integration, decoy and key deletion.
Integration
The first subproject is the integration, the key and message
sequences were integrated into the genomic DNA of two separate
Bacillus subtilis strains. In order to achieve this we had the key
and the message sequence synthesized by IDT. Then we cloned it in
pSB1C3 to amplify it from there and cloned it in the B. subtilis
integration biobrick BBa_K823023. This vector can be used to
integrate sequences into the amyE locus of Bacillus subtilis. From
there we integrated it in the genome of B. subtilis 168 trp+. For
message and key transmission these B. subtilis strains were
sporulated.
The location in the genomic DNA makes it harder to retrieve the
data, since whole genome sequencing is required. The message is
protected by a digital lock: the key, and thus doesn’t need further
biological protection.
The key is not encrypted and has to be protected by a biological
lock. We followed different approaches to design a multi-layered
biological lock.
We followed two main ideas, namely hiding the key and deleting
the key when unauthorized parties handle it.
Decoy
Hiding the key became called the decoy approach.The
key-containing spore will be send in a mixture of different decoy
spores. The recipient has to be aware of the special treatment that
is required to select the correct spores from the decoy. We have
been working on a photoswitchable ciprofloxacin compound.
Only the knowledge about the right wavelength allows the
recipient to activate the added ciprofloxacin and thus start
selection of the right spore strain.
Our design of the decoy approach included the biobricks for
ciprofloxacin resistance, a super folder GFP and a pATP promotor.
Key deletion
Another biological security layer is provided by our key
deletion system. We have been working on two different approaches.
The first is a nucA killswitch which is made out of an assembly of
different biobricks. Atc or tetracycline have to be added to
inhibit the tetR promoter to stop the expression of the nucA and
digestion of the key sequence.
The second approach makes use of a CRISPRcas system which will
delete the key from the genome if no special treatment is applied.
Addition of Atc or tetracycline will stop the cas9 expression.
This system is highly flexible and biological layers can easily
be added or modified to the wishes of the user.
Proof of concept
We chose to save and transmit our encrypted message and the
associated key in the genome of B. subtilis, which is our
CryptoGErM. The message and key were generated by our Encryption machine. For the integration we tried two different integration
plasmids, one is the BioBrick BBa_K823023 from iGEM Munich 2012 and
the other one is the pDR111. Construction of the plasmids can be
seen here (link to plasmid construction). Both plasmids integrate
in the amyE locus of B. subtilis. The amyE gene, which codes for a
starch degrading enzyme, is destroyed in the process of integration
and therefore colonies can be screened with the starch test for
successful integration. Both plasmids were constructed for both
message and key however the following part will only focus on the
BBa_K823023 integration plasmid.
The transformation into the B. subtilis was performed according
to the protocol Transformation of B. subtilis with the key sequence in
BBa_K823023 and message sequence in BBa_K823023. Since B. subtilis
is naturally competent it is easy and efficient to integrate the
message/key into the genome of this bacterium. We also tested the
transformation efficiency of the BBa_K823023 backbone (link to
transformation efficiency K823023).
Figure 1: CryptoGErM (B. subtilis) colonies after
being transformed with either the message or the key K823023
plasmid.
Figure 2 Starch test for the B. subtilis colonies
potentially carrying message or key.
The starch test in Figure 2 shows that almost all colonies are
not able to degrade starch therefore it can be assumed that they
integrated our message/ key into the amyE locus. The starch test
offers a quick and cheap way to screen multiple colonies.
Now, that we had a B. subtilis strain with our message and
another one with our key, we performed the sporulation protocol
Preparation of the spore stock of B. subtilis to obtain a spore batch carrying the key and
another one with the message.
Figure 3. (A) B. subtilis spores carrying the
message under the phase contrast microscope. (B) CryptoGErM
(left, with sunglasses) carrying the message and (right, cute)
carrying the key.
At this point it was time to find out if we could actually send
our message spores somewhere and retrieve the message back. The same procedure works for the key.
29/08. First the sending of the spores was simulated in the lab
by leaving the 10 μl of message spores in a tube in an envelope for
24 h.
30/08. After the sending simulation, the spores were streaked
out on LB agar with 150 μg/ml spectinomycin LB agar plates.
01/09. A single colony was picked from the restreaked spores
plate and used as template for colony PCR to amplify the message
sequence. Colony PCR with the F message sequence, R message
sequence primers and F message sequencing, R message sequencing
primers (primer sequences can be found in our primer list).
Figure 4 (A) Spores of Bacillus subtilis carrying
our encrypted message during the sending simulation. (B)
Colonies obtained from germinating the spores on LB agar.
Figure 5.Colony PCR on colonies obtained from Bacillus subtilis. Primers used: (o) F/R message sequence. Product 572 bp. (n) F/R message sequencing. Message product 916 bp. For the key the F/R message sequencing primers were used. Key product 178 bp.
The expected size for the message product of the first primer pair was 572 bp and for the second one it is 916 bp. Both bands can be seen on the agarose gel (see Figure 5). The key product was 178 bp and the gel shows that this product was successfully amplified from the genome of B. subtilis. The PCR product was subsequently cleaned up with the kit DNA Clean-up (PCR Purification Kit – Jena Bioscience). The sample 3n and 4n were sent for sequencing with the primers F/R message sequencing Sequencing (Macrogen).
Sequencing (Macrogen).
Figure 6. (A) PCR product, amplified message
sequence, on the way to be sequenced by Macrogen. (B)
Chromatogram of the sequencing result. (C) Obtained sequence.
05/09. The moment of truth! We copied-pasted the encrypted message sequence and the key sequence in our decryption machine. The key is converted to plain text. For this proof of conept we chose the key “Autoclave after reading”. The result of the decryption was: The world is full of obvious things which nobody by any chance ever observes.” CryptoGErM works! We successfully integrated an encrypted message into the genome of B. subtilis, the key in another strain and received the same message back from the spores. As instructed by the key, all cryptoGErMs were autoclaved in the end of the experiments.
.
Figure 7. Decryption program with the message
sequence
The next step was to actually send the message to another iGEM
team to demonstrate the functionality under real world conditions. Therefore read Collaborations.
Figure 8. (A) Members from iGEM team Eindhoven 2016 decrypting our message sent in spores. (B) Also the iGEM team Wageningen is happy to decrypt our message. Blue arrow indicates decrypred message
Decoy
The spores containing the DNA sequence that encodes our key will be
sent in a mixture with decoy spores. These strains were constructed
with our BioBricks BBa_K1930002 (key) and BBa_K1930006 (sfGFP) The high
ratio of decoy spores makes it hard for unauthorized parties to
retrieve the correct key if they try to sequence the entire sample by
brute force. In this experiment we tried to determine how fine our
system is in selecting the spores from the decoy once the right
treatment is applied.
We prepared a Bacillus subtilis strain containing a superfolder GFP
and a spectinomycin resistance cassette in the genome. Then we prepared
mixtures of that mutant with wild-type B. subtilis in different ratios.
In this experiment living cells were used instead of spores. After
growing in different conditions, the final mixture of mutant vs
wild-type strains was determined microscopically and in a flow
cytometer.
Due to the high ratio of decoy cells, any non-approved party trying
to sequence the entire sample will not be able to distinguish the
key-sequence from the background noise. In the scientific literature,
using standard sequencing techniques it has been possible to detect one
mutant out of 150 wild-type molecules [1][2]. However, fine-tuned
technologies, such as Duplex Sequencing, have shown to increase that
number to one mutant in 10,000 wild-type cells. That same technique is theoretically
able to detect one mutant out of 10 million decoys[3]!
Experiment setup
We combined different ratios of sfGFP bacteria and decoy bacteria.
LB medium was inoculated from glycerol stocks of the sfGFP strain and
the wild-type strain which were grown overnight at 37 °C, shaking at 220 rpm
in a 3 ml culture. On the next day the corresponding dilutions were
made and grown again overnight at 37 °C in a shaking liquid 3 ml
culture. The antibiotic was added to both the preculture and the
diluted culture.
The next morning the cells were visualized in the microscope
Time-lapse microscopy/Phase-contrast microscopy and additionally diluted 50 times in 1X PBS buffer to
analyze in the flow cytometer.
Figure no.
Concentration spectinomycin [µg/ml]
Initial ratio sfGFP:decoy
Final ratio sfGFP:decoy
1,2
0
0:1
0:1
3,4
0
1:0
25:1
5,6
150
0:1
0:1
7,8
150
1:0
160:1
9,10
0
1:1
12:1
11,12
150
1:1
230:1
13,14
0
1:150
10:1
15,16
150
1:150
70:1
17,18
0
1:10,000,000
0:1
19,20
150
1:10,000,000
0:1
Table 1. The initial ratio of mutant vs wild-type
strains was screened from 1 to 10 million. The final ratio was
measured as the relative (green:gray) area under the curve (AUC)
obtained in the flow cytometer (Figures 2, 4, 6, 8, 10, 12, 14, 16,
18 and 20). [Using: Flowing Software 2.5]
Results
The mutant strain containing the superfolder GFP can be seen green
in the microscopy images and is also marked green in the flow cytometer
graphs. The wild-type decoy cells are gray in both cases.
Figure 2 & 3: Wildtype without spectinomycin
Figure 4 & 5: sfGFP strain without spectinomyin
Figure 6 & 7: Wildtype with spectinomycin
Figure 8 & 9: sfGFP strain with spectinomyin
Figure 10 & 11: 1:1 without spectinomycin
Figure 12 & 13: 1:1 with spectinomycin
Figure 14 & 15: 1:150 without spectinomycin
Figure 16 & 17: 1:150 with spectinomycin
Figure 18 & 19: 1:10.000.000 without spectinomycin
Figure 20 & 21: 1:10.000.000 with spectinomycin
Discussion
As control groups we used samples that contained either only the
decoy wild-type strain or the spectinomycin-resistant sfGFP mutant.
While the wild type grew well without addition of spectinomycin
(Figures 1 and 2), no growth could be observed if the antibiotic was
added (Figures 5 and 6). On the other hand, the spectinomycin resistant
sfGFP strain grew well both in its presence or absence (Figures 3, 4, 7
and 8). In both cases, growth of cells not expressing sfGFP was
observed. This could be due to not fully developed cells that do not
yet express sfGFP in their current cell cycle.
A mixed culture in the ratio 1:1 without the addition of
spectinomycin showed presence of both wild-type and sfGFP cells (Figures 9,
10, 11 and 12) as expected. However, the unexpected higher ratio of
sfGFP strain under conditions that do not give advantage over the
wild-type strain leads us to assume that the mutant generally grows
faster than the wild-type strain. Similarly, adding the antibiotic increases
20 times the fraction of sfGFP cells, in this case by killing the
non-resistant wild-type.
The samples with a 1:150 ratio showed consistent results (Figures
13, 14, 15 and 16) compared to the 1:1 ratios. Without the addition of
spectinomycin the mutant outgrew the wild-type strain (10:1) even
though the initial ratio was not in its favor (1:150).
We went to an extreme of using a ratio of 1 mutant in 10 million wild-type cells.
In this conditions, no growth of mutants was observed. The ratio is too
high to allow the mutant strain to grow even in the presence of
antibiotic that would give it an advantage over the wild-type strain.
Conclusion
Our experiment shows that a specific strain, in this case containing
a sfGFP and a spectinomycin resistance cassette can be selected from a
larger number of decoys. We could not determine the optimal ratio that
would strengthen this layer of biosecurity. For further experiments the
ratio of decoys should be fine tuned to determine the maximum ratio of
spores:decoys that could be used, thus reassuring that unauthorized
parties will not be able to recover the key by sequencing the whole
sample but the intended recipient will still be able to recover it.
References:
[1]Fox EJ, Reid-Bayliss KS,
Emond MJ, Loeb LA (2014) Accuracy of Next Generation Sequencing
Platforms. Next Generat Sequenc & Applic 1: 106.
doi:10.4172/jngsa.1000106
[2]Pochon (2013). Evaluating
detection limits of next-generation sequencing for the surveillance
and monitoring of international marine pests. PLos One
8(9):e73935
[3]Schmitt MW, Kennedy SR,
Salk JJ, Fox EJ, Hiatt JB, et al. (2012) Detection of ultra-rare
mutations by next-generation sequencing. Proc Natl Acad Sci U S A
109: 14508-14513.
Transformation Efficiency of B.subtilis plasmid backbone
(BBa_K823023) created by iGEM LMU Munich 2012
The integration vector from team LMU Munich BacillusBiobrickbox
2012 (BBa_K823023) can be used to integrate an insert of interest
in B. subtilis. The cloned insert will be integrated within the
amyE locus in B. subtilis after transformation (see figure 1 for
the integration locus). The amyE gene encodes for the alpha-amylase
protein, which degrades starch. After transformation with
BBa_K823023, the AmyE locus will be interrupted by the insert. The
successful integration disrupts the ability of the bacteria to
degrade starch.
Two strains of B. subtilis were chosen to test the
transformation efficiency - B. subtilis 168 trp+ and B. subtilis
168 trp-. Each strain was transformed with three different
concentrations of BBa_K823023; 1 µg/ml, 100 ng/ml, 10 ng/ml. This
experiment was done in triplicate. The transformation was
performed as indicated in Transformation of B. subtilis and colonies
were selected on LB agar plates containing 5 μg/ml chloramphenicol.
Subsequently 9 colonies were screened for correct integration of
K823023 with the starch test Integration check: Starch test. They were grown on
agar plates containing starch. The amylase activity of amyE is
visible as a clear zone (halo) after addition of lugol’s iodine
suggests the amyE gene was still intact and functional, which leads
to the conclusion that the integration was unsuccessful. The
colonies not forming the distinctive halo suggests a successful
integration into amyE, disrupting its amylase activity which
degrades starch.
Results
From the transformed bacteria suspension 50 µl were plated on LB
agar plates with 5 µl/ml chloramphenicol. All the plates had colony
formation, as seen in figure 1 and 2.
Fig. 1 Colonies after transformation of B.
subtilis 168 trp+ with K823023. Concentration first row is
1 µg, second row is 100 ng, and third is 10 ng plasmid DNA.
Fig. 2 Colonies after transformation of B.
subtilis 168 trp- with K823023. Concentration first row is
1 µg, second row is 100 ng, and third is 10 ng plasmid
DNA.
After addition of lugol’s iodine, there were no clear zones
around any B. subtilis colonies, see figure 3. This result
demonstrated that the amyE locus in B. subtilis had been replaced
successfully. Apart from this transformation efficiency experiment,
our team has been using BBa_K823023 as a plasmid backbone for our
message and key for integration in B. subtilis.
Fig. 3 Addition of lugol’s iodine to colonies grown
on starch plates. There is only growth on the left plate 3,4
and 9. On the right plate there is only growth on 1,7 and 9.
The other colonies might died from a too heat inoculation loop.
In addition, the colonies in the plates were counted. Plates
which contain a lot of colonies were divided in 16 areas as seen in
figure 4. The area with the estimated average amount of colonies
were counted.
Fig. 4 counting colonies after transformation.
For the counted colonies the transformation efficiency is
calculated with the following formula:
(# Colonies on plate/ng of DNA plated) X 1000 ng/µg =
CFU/µg of DNA
The amount ng of DNA plated, could be calculated with the following:
Volume of plasmid used in µL x concentration of DNA in
ng/µl x (volume plated / total reaction volume)
The first calculation is given as example:
400 µl of bacteria suspension is transformed with 3,6 µl
plasmid. This plasmid had a concentration of 276 ng/µl. The plated
volume is 50 µl. The mean of the amount of colonies overnight was
1882,7 cfu.
The result of the calculation is:
(3,6 µl * 276) * (50 µl / 403,6 µl) ≈ 123 ng DNA
plated
(1882,7 colonies / 123 ng plated DNA) * 1000 ng/µg =
1,5E+04 CFU/µg of DNA.
The results are summarised in the graphs below:
Fig. 5. Results of the transformation
efficiency. The mean amount of colonies has been used to
calculate the CFU/µg DNA. Fig. 6 Results of the transformation
efficiency. The mean amount of colonies has been used to
calculate the CFU/µg DNA.
We can infer from the graphs that lower amount of DNA resulted
in higher colony forming units. The recommendation is not to use
1000 ng of DNA for transformation with BBa_K823023. Using 100 ng or
10 ng DNA for transformation would be slightly better.
