Difference between revisions of "Part:BBa K3697004"

Latest revision as of 00:40, 28 October 2020

Recombination-based Detection System for B. Subtilis (manP)

This system, once incorporated into the B. Subtilis will act as a detection system for a customizable nucleic acid sequence corresponding to the sequence with homology to the "homology arms" of the system. When exposed to the target sequence a recombination event will be triggered causing the excision of the negative selection marker that is flanked by the homology arms. More information about the specific negative selection marker used in this system can be found in the documentation for part BBa_K3697002 and more information about the specific set of homology arms used in this system can be found in the documentation for part BBa_K3697003.

The 2020 Stanford Team's Usage and Overview of Relevant Biology

This system works in B. subtilis because of the way that it integrates with the competence and genomic recombination systems in B. subtilis. A brief overview of both of these systems is given below, but more information about how the homology arms of the system trigger site specific recombination events can be found in the documentation for part BBa_K3697003.

B. subtilis' competence system is naturally triggered when they are put on environmental stress, but strains of B. subtilis have been engineered to have inducible competence. Some examples of strains with inducible competence systems that were successfully used by the 2020 Stanford iGEM team are 1A976 and 1A1276. Both of these strains were obtained through the Bacillus Genetic Stock Center (http://www.bgsc.org/). Once DNA is taken in through B. subtilis' competence system, B. subtilis will allow it to be recombined into the genome if there is sufficient levels of homology to the sequences already in the genome. In general, ~1000 base pairs of homology split between two homology arms flanking the site of integration into the genome are generally used to trigger a recombination event [1], but a recombination event with less effeciency can be triggered by regions with less homology.

This detection system was used in conjunction with the mannitol inducible competence system of Bacillus subtilis strain 1A1276. The first step of using this detection system in B. subtilis was assembling this system into PBS1C. This was done via a Gibson Assembly. Next, once the this system was put into PBS1C it was transformed into 1A1276 and selected for on media with chloramphenicol. Next, the cells with the detection system were transformed with a sample of DNA. In the event that the cells were exposed to a sample with the target sequence of interest, a second recombination even will be triggered causing the excision of the manP negative selection marker. If the target sequence was not in the sample, then this second recombination event will not be triggered and the negative selection marker will remain in the cells. The manP negative selection marker can selected for by exposing these cells to mannose. If the cells live and grow when exposed to media with mannose in it, then the second recombination event occurred and the target sequence was in the sample. If the cells die and/or don't grow then this second recombination event has not occurred and the target sequence was not detected in the sample.

Figure 1:Graphic of the procedure for the detection system described above

Modeling the System's Effectiveness

A model for the effectiveness of this system is described and illustrated below:

This system uses cell survival rate as readout for detection. The cells will be equipped with a landing in their genome which contains a gene rendering the sugar mannose toxic to the cell. When the target sequence is introduced to the cell interior, it will recombine with the landing pad region with a certain recombination efficiency and excise this gene, allowing the cell to survive in mannose. We can model the variance in the bulk survival rate of an entire colony and from that extract a confidence level in a positive result for a given measured survival rate.

In a colony of B. subtilis cells (green), only some will be competent, or able to take up environmental DNA (blue).

When DNA is introduced, out of these competent cells, only a few will pick up the target DNA sequence (yellow). Once inside the cell, only a certain fraction of these recombine successfully into the genome (red)

When exposed to mannose, the cells which successfully recombined have a very high survival rate, while almost all of the cells that didn't recombine will die (black)

Cultures which do not have any target DNA exhibit a lower survival rate, but can we distinguish the detecting case from the non-detecting? We first attempt to model the fraction of cells which can pick up the target sequence. We assume that a B. subtilis cell will uniformly and independently sample strands of DNA from the environment of a fixed length as it uptakes DNA. The probability that any one such strand contains the target sequence is given by the equation below:

The cell cannot take up the target at all if Lt > Lf, and we assume that no fragment contains two copy of the target. We can then say that for a competetent cell taking in many fragments from the environment:

And combining this with the fraction of uptaking cells:

However, for any particular culture of cells, the fraction of uptaking cells will be random. We can model the actual fraction of uptaking cells as coming from a normal distribution, since it is sampled from a binomial distribution.

Now we can feed the value I as drawn from this distribution into the expression for survival rate

Now, we can again sample from this binomial distibution N times to get the distribution of survival rates for any colony of cells. We will make the simplification that the uncertainity of I does not propagate into the uncertainity of the deviation of S, in fact we can make two approximations. The standard deviation of S - E(s) is between (sqrt(.25/N)) for the upper bound, and (sqrt(F(1-F)/N)) below. We will see that these correspond to the more sensitive and more specific models of the test, respectively. Since variances add, we get:

In the case where there is no target DNA in the environment, we can replicate the above expression and solve for I = 0.

We have expressions for what fraction of cells we should expect to survive in the case where there is target DNA in the environment and when there isn't, so for a given measurement of cell survival rate, we can calculate the chances that target DNA was present.

Because we modelled these as normal distributions, we can explicitly evaluate this expression for given values of I, E, F, and N.

These are both gaussian distributions, and we can graphically evaluate (28) for the given values.

Distributions for Sd (red) and S (blue) for E = .7, F= .04, N=1000, and I =.27 (estimated below). (green) The probability of target presence given p = .07 for given Sm, under specific model.

Estimates of I, the fraction of cells taking in the target DNA with Nf = 50, Lf = 2000, LT = 1000, X = .7. We estimated above using chi = .01. These estimates use reasonable figures from the literature. B. subtilis has been reported to take in around 100,000 bases of DNA in chunks of thousands of nt. The target we used is 1000nt in length. Super competent B. subtilis strains have been shown to achieve up to 70% competence. The MiniBacillus project reported recombination efficiencies of around 70%, and survival rate of unrecombined cells in mannose is likely even lower than 4%, the prior is a reasonable value within the historical prevalence of COVID-19. Using a modest colony of only 1000 cells, we are able to achieve an incredibly specific test. In the realistic distributions of measurements, the probability of target presence is very close to either 0 or 1. We have an extremely high confidence in the positive or negative results of this test. For example, from the model it appears that this kind of system would have a false positive rate of less than .1% and a rate of over 99.9% for detection of positive results. This model demonstrates that the high sensitivity and power of this system is robust to altering the conditions of the data. Large changes from the parameters above will still maintain the sharp discrimination of this model. It is only in extreme cases, where I approximately equal to .05 (pictured below) or E is approximately equal to .1 that cases with low confidence start to occur with any noticeable frequency. Even in situations where both of those figures prevail, a culture of 100000, which is still highly practical, achieves high discrimination.

[1] Dubnau D. Sonenshein AL, Hoch JA, Losick R. Genetic exchange and homologous recombination, Bacillus subtilis and Other Gram-positive Bacteria: Biochemistry, Physiology, and Molecular Genetics, 1993Washington, DCASM(pg. 555-584) Sequence and Features

Sequence and Features