Hi I'm Staph. Please wait a minute for me to prepare the website... Hi I'm Staph. Please wait a minute for me to prepare the website...

Overview

1 / 2
Protein Structures
2 / 2
Mathematical Models

Based on curiosity, we simply simulated our polypeptides’ structures with AlphaFold2. Next, we pick the part of the cleavage site structure to do protein docking simulation by Autodock. We can see our AMPC and V8 protease binding affinity before doing the experiment. Moreover, to continue predicting effective time by the following model, we can determine which AMPC we designed is the most suitable for our system.

To find out the effective time of our system, we have to monitor the change of substrate and product amounts to check the activity of the enzyme. Though it can be done by experimentation, we don’t have enough time to finish it. To get a more efficient way to predict the achievement of our system, we built a simulation model with every antimicrobial peptide complex(AMPC) we tried.

There have three combination types of our polypeptides, including ABABAB, AAB, and AB. A and B mean different AMPs individually. There are 1 AMPC in combination AB, 5 in AAB, and 1 in ABABAB. E in the simple sequence representation means the linker, glutamic acid. For easier understanding, our naming convention of AMPC in the model is A/B.

Sequnces of Different Combination Types

Sequence Number Part Name Description Name In Model
No.1 BBa_K4336000 [LL37]E[LL37]’E[Lysostaphin] 2LL37/Lyt
No.2 BBa_K4336001 [HNP1] E [HNP1]' E [Lysostaphin] 2HNP1/Lyt
No.3 BBa_K4336002 Lysostaphin Lyt
No.4 BBa_K4336003 E(LL37) E (hBD-3) E (LL37)’ E (hBD-3)’ E(LL37) E (hBD-3) 3LL37/3hBD3
No.5 BBa_K4336004 E(LcCCL28) E (Lysostaphin) LC28/Lyt
No.6 BBa_K4336005 [Ranalexin] E [Ranalexin]' E [Lysostaphin] 2Ran/Lyt
No.7 BBa_K4336006 [HBD3] E [HBD3]' E [Lysostaphin] 2hBD3/Lyt
No.8 BBa_K4336007 [Histatin 5] E [Histatin 5]' E [Lysostaphin] 2Hst/Lyt

See Parts page for detail information.

AMPC Structure

Description

Driven by curiosity and preparation for a model, we use AlphaFold2 open source on google colab to simulate our protein structures. Though most of the AMPs are attached to Lysostaphin, the simulated structures are not similar. We think maybe the part of Lysostaphin may change its structure when connect with different AMPs. However, though there are some positions in our structure that have low IDDT values, the prediction of AlphaFold2 is still the most accurate software we’ve used. We can have confidence in our prediction and we use part of the structure to predict the binding affinity between our AMPC and V8 protease which can find detail in our protein docking results.

CL28/Lyt
2(Ran)/Lyt
2(LL37)/Lyt
2(Hit5)/Lyt
2(HNP1)/Lyt
2(HBD3)/Lyt
3(LL37)/3(hBD3)

Protein Docking

Description

After predicting our AMPC structure, we want to know more about how they will bind with the V8 protease. Because the cleavage site preference of V8 protease cares about the amino acids near the glutamic acid(P1), we use three amino acids of the cleavage site (P2-P1-P1’) of our substrates be ligands to predict the binding activity. We first check what are the possible cleavage sites we have and the naming convention is P2-P1-P1’.

We use Autodock to achieve this part. Referring to past research and trying several positions of the grid box, we set the parameters to 46 × 34 × 62 around the substrate-binding region and set the center at the position that our substrates mostly probably bind to. We selected the Lamarckian genetic algorithm (LGA) to find the most favorable interactions. Below are the docking parameters we use in Autodock.

Results

Below is the result including our probable cleavage sites. We choose the run with the lowest binding energy to present. With these images, we can better realize how well our substrates bind with V8 protease with different amino acid combinations.

Binding Result

The pdf file represents the binding energy part of our binding result. (To find out the overall log file, please scroll to the bottom of the page to see the attachments.) We can obtain the lowest binding energy order here:
glutamic(-7.10) < YEA(-4.29) < CEF(-4.03) < SEA(-3.60) < YED(-3.59) < SEL(-3.53) < KEA(-3.22) < SEG(-3.01) < CEA(-2.89) < KEG(-2.86) < KEL(-2.49)

CEA

CEF

SEA

SEG

SEL

KEA

KEG

KEL

YEA

YED

Conclusion

Mostly, the docking result is the same as the past research which points out that when the P1’ site is A or G and the P2 site is Y, V8 protease would prefer to cleave. Furthermore, we can simply guess that the AMPC which will be best cleaved is CL28/Lyt and 2Hst5/Lyt which involve YEA and YED at the cleavage site. The second place for best-cleaved AMPC might be 2LL37/Lyt which involves SEL and SEA at the cleavage site. Comparing this result with the following effective time model result, we may see that CL28/Lyt and 2LL37/Lyt show good performance among AMPC candidates and can have a different perspective to determine our best AMPC.

System Effective time v1

Description

This is the primary design of our model. We planned to get three variables, including the reaction rate constant, minimum inhibitory concentration, and the degradation rate of each antimicrobial peptide. Below is how we designed the simple version. If you want to see the final result, please refer to System Effective time v.2. There are three steps for us to build the model. First, use MATLAB to do curve-fitting by the first-order reaction equation to find the reaction rate constant which would be on the measurement page. Next, list out all the equations that our AMPC is being cleaved. Last, differentiate the equations with time and apply them to python to get the result.

Reactions and Functions

According to the literature, we tried to assume their reaction rate constant k_1 and k_2 for different cleavage sites.
In our project, we found glutamic acid with leucine, aspartic acid, and phenylalanine were the V8 protease preference sites, and glutamic acid with glycine, alanine, and lycine were not.


According to this reaction, we can list out equations of the substrate and products below.


According to this reaction, we can list out equations of the substrate and products below.


(Click ☝ )
According to this reaction, we can list out equations of the substrate and products below.

(Click ☝ )

Assumptions

Python Code

Please check out our GitLab.

Conclusion

This model is simple and can be easily understood. After discussing our model with other iGEM teams and our advisors, we thought this model design was too simple to meet the realistic situation. In version 2, we would include Enzyme Kinetic equations and consider more about the interaction between V8 protease and our AMPC.

System Effective time v2

Description

This is the modified model design. We apply Enzyme Kinetic formulas in this version. What is very different from the beginning design is that we consider more about how well the V8 protease bind with our substrate which will make our model more likely to fit the real situation of our project.

Before we start building the new model, we found that the value of K_m and k_cat of V8 protease are 28.4 mM and 0.29 s^(-1) respectively. For easier use, we transfer these two data into 28400 μM and 1044 hr^(-1).

Next, refer to the product information of V8 protease of Sigma-Aldrich, they recommend using a ratio of 3/100 (w/w) of the enzyme to the substrate, and we diluted the concentration of V8 protease to 1 μg/ml. That is, after transferring the unit, we used the concentration of \(0.037μM)\ of V8 protease to do the experiment to check the effectiveness of our project. The ideal original concentration of different AMPC calculated is documented below.

We used the above data for experiments because of the lack of AMPC. However, we still wanted to know how much of the concentration of V8 protease and AMPC can reach the appropriate level that we expected. Therefore, we increased the concentration of V8 protease and AMPC 100 times higher like below to run our model.

The last step in preparation for building our model is obtaining each AMP’s minimum inhibitory concentration (MIC). These MICs help us make the result graph clearer to see how long the effective time is. Below are the MIC ranges of each AMP.

Reactions and Functions

Below is the standard reaction of Enzyme Kinetic. Before writing python, we have to list out the equations of each AMPC and AMP when they are being a substrate.


With each type of AMPs combination, below are their reactions individually.


According to this reaction, we can list out equations of the substrate and products below.


(Click ☝ )
According to this reaction, we can list out equations of the substrate and products below.

(Click ☝ )


(Click ☝ )
According to this reaction, we can list out equations of the substrate and products below.

(Click ☝ )

Assumptions

Python Code

Please check out our GitLab.

Results

Before looking at the graph of the result, please note that

  1. The lines of final products mean line A(t) and line B(t).
  2. The naming convention of our AMPC is A/B
  3. On the labels of the graph, ub means upper bound, and lb means lower bound.
  4. The >effective time means the time range that the lines of final products on the graph exceed the MIC lines respectively.

    5. First, for AMPC LC28/Lyt, the lines of final products can exceed the MICs of LC28 and Lyt. The effective time is 12 and 25 hours for LC28 and Lyt respectively.

    Next, there are five kinds of AMPC in the >AAB combination, including 2Ran/Lyt, 2LL37/Lyt, 2HNP1/Lyt, 2hBD3/Lyt, and 2Hst5/Lyt. Most of the lines of products can go beyond a single AMP’s MIC upper bound respectively, except for the line of Hst5 can not reach the MIC bound. The effective time ranges of each AMP are 16, 30, 30, 5, 0, and 25 hours for Ran, LL37, HNP1, hBD3, Hst5, and Lyt respectively.

    Last but not least, for the AMPC 3LL37/3hBD3, we can see the lines of products are all above the MIC of hBD3 and LL37. What’s more, the effective time of LL37 and the one of hBD3 are 40 hours and 15 hours respectively.

    Conclusion

    In this model, we analyzed the effective time by comparing each AMP’s concentration and MIC levels. The results show that the drug's effective time can last for more or less than a day with this level of concentration. If what we predicted happens, it means that we don't have to use the patch for a long time but only 30 minutes then can have a long effective time. Furthermore, our system functions like two AMPs kill the bacteria at the same time and may be very different from the one that has only one kind of AMP. Therefore, We think that the effective time of reality may be better than the above result. That is, we don’t need that much concentration of AMPC to totally reach the upper bound of MIC but can still kill the bacteria well because of the help of another AMP.

    This model helped us a lot to get an insight into our project, but for a more complete and accurate prediction of the efficacy of our system, we plan to involve the capacity of the microneedle and the concentration of S. aureus and V8 protease in the patient’s skin in the model parameters in the future.

    Reference

    1. Frey AM, Chaput D, Shaw LN. Insight into the human pathodegradome of the V8 protease from Staphylococcus aureus. Cell Rep. 2021 Apr 6;35(1):108930. doi: 10.1016/j.celrep.2021.108930. PMID: 33826899; PMCID: PMC8054439.
    2. Manne K, Narayana SVL. Structural insights into the role of the N-terminus in the activation and function of extracellular serine protease from Staphylococcus aureus epidermidis. Acta Crystallogr D Struct Biol. 2020 Jan 1;76(Pt 1):28-40. doi: 10.1107/S2059798319015055. Epub 2020 Jan 1. PMID: 31909741; PMCID: PMC6939437.
    3. Jumper, J., Evans, R., Pritzel, A. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596, 583–589 (2021). https://doi.org/10.1038/s41586-021-03819-2
    4. Domhan C, Uhl P, Kleist C, Zimmermann S, Umstätter F, Leotta K, Mier W, Wink M. Replacement of l-Amino Acids by d-Amino Acids in the Antimicrobial Peptide Ranalexin and Its Consequences for Antimicrobial Activity and Biodistribution. Molecules. 2019 Aug 17;24(16):2987. doi: 10.3390/molecules24162987. PMID: 31426494; PMCID: PMC6720431.
    5. Yang XY, Li CR, Lou RH, Wang YM, Zhang WX, Chen HZ, Huang QS, Han YX, Jiang JD, You XF. In vitro activity of recombinant lysostaphin against Staphylococcus aureus isolates from hospitals in Beijing, China. J Med Microbiol. 2007 Jan;56(Pt 1):71-76. doi: 10.1099/jmm.0.46788-0. PMID: 17172520.
    6. Su J, Li H, Hu J, Wang D, Zhang F, Fu Z, Han F. LcCCL28-25, Derived from Piscine Chemokine, Exhibits Antimicrobial Activity against Gram-Negative and Gram-Positive Bacteria In Vitro and In Vivo. Microbiol Spectr. 2022 Jun 29;10(3):e0251521. doi: 10.1128/spectrum.02515-21. Epub 2022 May 26. PMID: 35616397; PMCID: PMC9241943.
    7. Wanmakok M, Orrapin S, Intorasoot A, Intorasoot S. Expression in Escherichia coli of novel recombinant hybrid antimicrobial peptide AL32-P113 with enhanced antimicrobial activity in vitro. Gene. 2018 Sep 10;671:1-9. doi: 10.1016/j.gene.2018.05.106. Epub 2018 May 30. PMID: 29859288.
    8. Komatsu T, Salih E, Helmerhorst EJ, Offner GD, Oppenheim FG. Influence of histatin 5 on Candida albicans mitochondrial protein expression assessed by quantitative mass spectrometry. J Proteome Res. 2011 Feb 4;10(2):646-55. doi: 10.1021/pr100861k. Epub 2010 Dec 6. PMID: 21080726; PMCID: PMC3033980.
    9. Midorikawa K, Ouhara K, Komatsuzawa H, Kawai T, Yamada S, Fujiwara T, Yamazaki K, Sayama K, Taubman MA, Kurihara H, Hashimoto K, Sugai M. Staphylococcus aureus susceptibility to innate antimicrobial peptides, beta-defensins and CAP18, expressed by human keratinocytes. Infect Immun. 2003 Jul;71(7):3730-9. doi: 10.1128/IAI.71.7.3730-3739.2003. PMID: 12819054; PMCID: PMC162002.
    10. Kakudo S, Kikuchi N, Kitadokoro K, Fujiwara T, Nakamura E, Okamoto H, Shin M, Tamaki M, Teraoka H, Tsuzuki H, et al. Purification, characterization, cloning, and expression of a glutamic acid-specific protease from Bacillus licheniformis ATCC 14580. J Biol Chem. 1992 Nov 25;267(33):23782-8. PMID: 1429718.
    11. https://2021.igem.org/Team:CCU_Taiwan/Model