overview

Modeling of a synthetic biological wastewater treatment

engineering system

Yuanxin Wang, Jianjie Zhao, Ruihao Li

Overview

Our project mainly focus on designing gene circuits to gather copper ions, degrading cyanide, detoxifying fluoride and suggesting whether the water is safety for further use. With these giant goals, the first thing we needed to do is using computational method to simulate the biological process and figure out whether our design is feasible. We established DDEs (delay differential equations) to see whether our instructors are trustable and give some further information for the detective part of our toolkit. Then we tested the robustness and sensitivity to get a broader insight of biological system both in single cell level and multicellular level. By doing this, we can get their properties for better application.

Single cell level

DDEs simulation

There are two kinds of E.Coli in the project—workers and instructors. The former ones produce some proteins binding with copper ions in the polluted water and the latter ones tell us whether the water is safe enough for further use. Since the thing we care about most is the safety of the water and the workers will be dedicated to remove the ions in the water before we decided to let them flow to the following pool, we established some equations to simulate the biological process of instructors. Considering about it will take some time for the transcription and translation process before a protein can bind with some certain promoters, we use DDEs instead of ODEs to make our simulation closer to the reality. And here are the equations:

parameter	description	value	reference
copynum	copy number of pACYDuet-1 plasmid	18~22	[1]
trc1	transcription rate of mCII	w/o inducing: 0 inducing:	[2]
deg1	transcription rate of mCII	0.12	[3]
deg2	degradation rate of CII	0.1	[4]
deg3	degradation rate of mCI	0.12	[3]
deg4	degradation rate of CI	0.042	[4]
deg5	degradation rate of mGFP	0.13	[5]
deg6	degradation rate of GFP	0.017	[6]
deg7	degradation rate of mRFP	0.13	[5]
deg8	degradation rate of RFP	0.017	[6]
trl1	translation rate of CII	0.12	[4]
trl2	translation rate of CI	0.09	[3]
trl3	translation rate of GFP	5.4	[7]
trl4	translation rate of RFP	5.4	[7]
Vmax1	maximum transcription rate when induced by CII protein	0.9	[4]
Vmax2	maximum transcription rate when induced by CI2 protein	0.66	[4]
τ1	time for CII transcription, translation and folding	0.24min	estimated the same as CI2
τ2	time for CI2 transcription, translation and folding	0.24min	[3]
km1	apparent association constant for CII binding with pRE promoter	0.398	[8]
km2	apparent association constant for CI2 binding with pR promoter	1.58*10-3<>	[3]
k1	reaction constant for CI forming CI2	3	[4]
k2	reaction constant for CI2 disassociating to CI	30	[4]

[1]

[2] Copper-inducible transcriptional regulation at two promoters in the Escherichia coli copper resistance determinant pco D. A. Rouch and N. L. Brown Microbiology (1997), 143, 1191-1202

[3] Kinetics of lambda phage manipulate genes, regulatory networks and protein interaction lysogenic state / cleavage transition characteristics. Hui Ding, et al. Journ al of Inn er Mongolia University Sep. 2007 Vol. 38 No. 5

[4] Stochastic Kinetic Analysis of Developmental Pathway Bifurcation in Phage l-Infected Escherichiacoli Cells  Adam Arkin ,et al. Genetics 149: 1633–1648(August 1998)

[5] Global analysis of mRNA decay and abundance in Escherichia coli at single-gene resolution using two-color fluorescent DNA microarrays, Jonathan A. Bernstein, et al. PNAS July23, 2002, vol.99 no.15, 9697–9702

[6] http://bionumbers.hms.harvard.edu/

[7] https://2009.igem.org/Team:PKU_Beijing/Modeling/Parameters

[8] Kinetic analysis of mutations affecting the cII activation site at the PRE promoter of bacteriophage λ, MING-CHE SHIH, et al. Proc. Natl. Acad. Sipi. USA, Vol. 81, pp. 6432-6436, October 1984, Genetics

The results of simulation are shown in the graphs below:

As you can notice in the picture, the expression level of fluorescent protein is changed a lot between polluted and non-polluted water. Thus, by detecting the fluorescence intensity of each protein, we can gain the information about whether the water is safe for further use. Considering about the severe consequences about taking in too much copper ions, we should make sure that our data is credible and the information we get from it is accurate.

We simulated the whole process of the water-dealing procedure. In the view of that the transcription rate of the copper sensitive promoter is related to the concentration of copper in the water, we divided the dealing process into several parts with different transcriptional rate and combine all the data eventually to make our simulation closer to the reality. The result showed below indicates that detecting one of the fluorescent intensity only is enough to get the information we want. But to detect the other fluorescent intensity redundantly can make the conclusion more trustable.

Robustness and sensitivity analysis

Considering that there are so many parameters and variables in the equations, we chose to use numerical solutions to analyze the robustness of the equations. Since trc₁, τ₁ and τ₂ are three changeable parameters that may contribute most to the results of the output, we decided to put our focus on these three parameters in this part. And here are some graphs representing the expression states under different trc₁ values.

From left to right, up to down, the values of trc₁ are: 0, 0.002, 0.005 and 1. These four graphs represent four different stable states with different trc₁ values. Although the final expression levels are different, they all achieve a stable states. In other words, we can judge whether the water is safe for further use by just detecting limited numbers of data.

Then we shift our focus onto the specific parameters. The first thing we did is to analyze how the expression levels of GFP and RFP are sensitive to the value of trc₁. We made the value of trc₁ be changed from 0 to 0.1 at the step length of 0.001. The picture showed below is the simulating result.

As you can see in the picture, the green curve represents the GFP expression condition and the red one represents that of RFP. The fewer and more scattered the curves are, the faster the final output changes. Based on the picture showed above, we can conclude that the expression of GFP and RFP is less sensitive to trc₁ with increasing value of it from 0 to 1. When trc₁ approaches to ∞, the expression level of GFP is close to what the black curve indicates while that of RFP is close to what the blue one indicates.

To test the effects of the expression condition to GFP and RFP caused by trc₁ a step further, we pictured the fluorescent intensity of GFP and RFP at t equaling 200min (an estimated stable states) under different trc₁ values.

trc1*10-2	[0 1]	[1 2]	[2 3]	[3 4]	[4 5]	[5 6]	[6 7]	[7 8]	[8 9]	[9 10]
GFP	2332.70	11.86	3.33	1.71	1.08	0.76	0.57	0.45	0.36	0.30
RFP	423.87	282.56	201.85	151.43	117.81	94.28	77.16	64.32	54.44	46.68

The left picture above shows the fluorescent intensity of GFP and RFP on the stable state under different trc₁ values, while the right one shows the rate of change. The chart above shows the digital number changes of GFP and RFP when trc₁ changes at the step length of 0.01.

According to the data and graphs above, we will find it’s not hard to make the conclusions below:

1. The expression level of GFP and RFP is rather sensitive to trc₁, when value of trc₁₁ becomes larger, this kind of influence reduces and will finally have nothing to do with the expression level of GFP and RFP. (In the extreme state)

2. The influence of trc₁ to the expression condition of GFP is short but obvious. When the value of trc₁ changes between 0 and 0.01, the change of relative fluorescent intensity of GFP is about 2322. While that of trc₁ changes between 0.09 and 0.1, the change of relative fluorescent intensity of GFP is only about 0.3, which is close to 0. In addition, you can easily observe from the graph on the right that the rate of change of the GFP fluorescent intensity reaches the maximum to at the site A and reduces tremendously to near 0.

3. The influence of trc₁ to the expression condition of RFP is mild but endless. When the value of trc₁ changes between 0 and 0.01, the change of relative fluorescent intensity of GFP is about 424. When that of trc₁ changes between 0.09 and 0.1, the change of relative fluorescent intensity of RFP is about 47. The fluorescent intensity of RFP is still increasing when the value of trc₁ equals 0.1. Actually, this kind of increasing will be kept even the value of trc₁ is over 1.

4.At the site 1 on the left graph, the value of trc₁ is 0.004, where the fluorescent intensity of GFP and RFP is exactly the same. And at B site on the right graph, the value of trc₁ is about 0.0099, where the rate of change of the fluorescent intensity of GFP and RFP is the same.

Considering that the dealing process is a long-lasting period, during which the concentration of copper ions change a lot, the transcription rate is different from time to time. We tested the time that would consume to reach a new stable state with the change of trc₁.

This picture represents the simulated condition with value of trc₁ changed from 0.005 to 0.002. We defined that when the related error equals or is less than 5%, it reached a stable state. So as you can see in the picture above, when t equals 373 (site E), the fluorescent intensity of GFP and RFP are equaled. When t equals 554 (site F), the expression level of RFP reaches the stable state. And when t equals 646 (site G), the expression level of GFP reaches the stable state.

The above is what we did to analyze the output sensitivity to trc₁, and next we tested how the parameter τ (time delay) affect the output result. We changed the value of τ from 0.12 to 0.36 at the step length of 0.01 to see the result.

As you can see in the picture above, the fluorescent intensity of GFP and RFP is extremely insensitive to the change of τ (time delay). When the parameter τ (time delay) changes from 0.12 to 0.36, the fluorescent intensity of GFP in the stable state only changes at the value of 5.28, while that of RFP only changes at the value of 2.23.

Circuit improvement

As mentioned above that our system is rather sensitive to the concentration of copper ions, the detected results may fail to tell what the concentration of cooper ions is exactly. If the basic transcriptional level of the promoter is higher than 0.005, the expression level of RFP is always comparatively much higher than that of GFP. Unfortunately, according to the results from our wet lab, the promoter we chose at the first time has a quit severe leakage. Thus we will have no idea about whether the water is safe enough for further use by detecting the fluorescent intensity. Since it’s the expression level of RFP is higher than our expectations, we tried to reduce it by improving our gene circuits. According to the designing, it’s CII protein that induce the expression of RFP, so we thought adding a degradation tag may be a good solution. Then we tested how many times of the original degradation rate is needed. And here is the result with the transcription rate in the state of the water is safe according to the national standard. Considering about that there are always some oscillations of parameters in the biological system, we also did a gradient analysis to see whether these oscillations may affect the result and whether there is any superior choices.

As you can see in the picture above, a specific times of the degradation rate is needed and there is no much room for the parameter oscillation. Since the degradation rate cannot be predicted accurately in vivo especially when some tags are added (the accelerated degradation rate largely depends on the proteins inside cells), we try to find some other promoter that is more suitable for our project.

Fortunately, we found the pcoa promoter. As you can see in the graph below, there is no significant difference of the relative fluorescent intensity between the pET28a plasmids containing and not containing the pcoa promoter. So it can be predicted that the transcription rate of this promoter can meet our need.

Multicelluar level

Environmental factors

Since the wastewater we tried to deal with come from the process of industry producing, we must considering some environmental factors that can affect the treatment procedure. There are some other ions in the water that can form some chelate compounds combining with copper ions. And these ions have a competitive relationship with copper-binding proteins. By analyzing the whole environmental surroundings, we can get the information about how many copper-binding proteins estimated do we need, thus to calculate how many bacteria we should paint on the surface of the RBC (rotating biological contactor). According to the previous literature review, we found the major existing forms of copper ions in the wastewater from industries are [Cu(CN)3]^2- and [Cu(NH3)4]²⁺. There are 4 major reactions occur in the water, and here are equations.

After looking up some papers, we found the concentration of [Cu(CN)₃]^2- is estimated to be 50mg/L^[9] and that of [Cu(NH3)4]²⁺/ is about 20.55mM^[10] in the wastewater from industry. Since our E.Kungfu can also oxidize cyanide, CN^- can be removed from the water, so the balance of the process will move to the direction in favor of degrading [Cu(CN)₃]^2-. Thus, most of [Cu(CN)₃]^2- will be transformed into Cu⁺. Then disproportionation reaction occurs, all Cu+ will be transformed into either Cu or Cu²⁺. As for [Cu(NH₃)4]²⁺, it’s because one of the degradation products is NH3, which is a kind of gas that will be released from water, that the balance of the process will also move to the direction in favor of degradation. So the amount of CBP we need is just the exact amount of copper ions that we try to remove from the wastewater. (The apparent dissociation constants for Cu(I)^- binding proteins and ligands of low-mass is about 1015 and we will constantly remove bacteria containing CBPs from the wastewater, so we assumed that all the copper ions can be adsorbed by these copper binding proteins.) According to the data mentioned above, the number of CBPs we need to treat the wastewater to make it meet the national standard is about 3.09*1021. By estimating how many copper bind proteins an E.Coli can bear, we can gain the data about how many E.Kungfu we need to plant onto the rotating disks. And combining with the diameter of the rotating disk, we will know how thick of the cultural medium should be needed to fixate enough gene modified worker bacteria.

[9] The research of treating sodium hypochlorite containing [Cu(CN)₃]^2- complex ions in wastewater. Shiqian Wei Vol 26 No. 5 Xuchang University Vol. l26. No. 5, sept. 2007

[10] The research of treating copper ammonia complex in wastewater use TMT. Dongmei Liao, Chinese Water Supply and Drainage Vol. 22, Sept. 2006
RBC modeling