Team:CCU Taiwan/Design

Team:CCU Taiwan/Implementation -

Team:CCU Taiwan/Implementation



There is an urgent demand for advanced antimicrobial dressings to defend against superbug infection on the wounds of patients with weak immunity or chronic disease. Through an extensive literature search, we found that antimicrobial peptides (AMPs) are one of the best candidates to overcome the antibiotic resistance of superbugs. However, the constitutive release of AMPs may stimulate unexpected immune responses [1]. Inspired by the 2019 iGEM team Linkoping_Sweden, we applied and improved the thrombin-based mechanism of AMPs release. AMPs alone are not enough to protect patients from secondary infection or promote recovery. Therefore, we designed a protective outer layer to physically block further infection and a cell-penetrating AMP (TAT-AMP) to combat superbug latency. We further combined an absorptive layer and a collagen layer to provide a suitable environment and materials for wound recovery. Finally, we applied a collagen binding domain (CBD) to anchor AMPs and TAT-AMPs onto the collagen layer. Together, we hope that our dressing, namely AgenT will provide a new solution to block the spread of superbugs.

Survey and design of antimicrobial components in

Survey of Antimicrobial peptides (AMPs)

AMPs are short peptides that consist of 10-50 amino acids. Typically, AMPs are composed of positive net charge and amphiphilic α-helical structure. The mechanism of AMPs can simply be divided into two categories, membrane disruption, and metabolism interference. Most AMPs interact with negatively charged bacterial cell walls based on electrostatic attraction. Afterward, it disrupts the integrity of the cell membrane, resulting in leakage of vesicles and cellular components. Besides cell membrane disruption, some AMPs can diffuse through the cell membrane and inhibit DNA replication, disrupting RNA and protein synthesis, which in turn leads to cell wall lysis [2][3].

▲ Figure 1: The mechanism of AMP-Cell disruption (upper), inhibit DNA replication, RNA and protein synthesis.

Based on a chart summarizing AMP characteristics [2], we first selected nine AMPs with favorable characteristics for safety, Minimal Inhibitory Concentration (MIC), and cytotoxicity (Table 1). We then focused on AMPs with activity against S. aureus or MRSA (Table 1), including D2A21, DPK-060, Mersacidin, and IDR-1. Mersacidin was later excluded because the modified peptides could not be generated through biobricks. IDR-1 modulates immunity to combat MRSA, which did not suit our project goals. Finally, we decided to use D2A21 and DPK-060 in our project.


D2A21 is a synthetic AMP with amino acid sequence FAKKFAKKFKKFAKKFAKFAFAF from N- to C- terminal [3]. The enriched positive charges allow D2A21 to exert its antibacterial effect by inserting and formatting voltage-sensitive channels in the cell membrane of bacteria [4].


DPK-060 is a cationic peptide with random coil structures derived from the endogenous human protein kininogen [3]. The amino acid sequence of DPK-060 is GKHKNKGKKNGKHNGWKWWW from N- to C- terminal. Three tryptophans (W) are added to the C-terminal of DPK-060 to promote its absorbability to Gram (+)/(-) bacteria, and to enhance the peptide stability [5]. The antimicrobial mechanism of DPK-060 is similar to that of D2A21, generating voltage-sensitive channels in the bacterial cell membrane.

AMP Name Sequence Mechanism of Action Activity Against Side Effects
Inhibition of DNA/RNA synthesis Gram (+), (−);
fungi; E. coli;
K. pneumoniae
Safety reported
(phase III)
Membrane disruption Gram (+), (−);
Gram (+), (−);
S. aureus ;
E. coli ;
P. aeruginosa
No side effects reported
(phase II)
Membrane disruption and immunomodulation Gram (+), (−); fungi;
S. aureus
No side effects reported
Mersacidin AAbFAbLPGG
Inhibition of cell wall Gram (+), MRSA; Clostridium spp. Safety reported
(phase I)
Immunomodulation MRSA; VRSA Safety reported
(IDR-1 derivative;
phase III)
RIVPA Immunomodulation Gram (+), (−) No side effects reported
(phase II/III)
Membrane disruption Gram (+), (−);
P. aeruginosa
No cytotoxicity and no resistance reported; no staining of human cornea
(phase I/II)
Membrane disruption Gram (+), (−);
P. aeruginosa
No cytotoxicity and no resistance reported; no staining of human cornea
(P113; histatin 5 analog;
phase IIb)
Membrane disruption and immunomodulation Gram (+), (−); Candida spp.;
A. baumannii;
P. aeruginosa;
E. cloacae
No cytotoxicity reported

▲ Table 1: The list of AMP candidates [3].

Design of thrombin-releasable AMPs linked with CBD

To ensure that AMPs stably anchor on the collagen layer, we decided to link a collagen binding domain (CBD) to the AMPs. The fibronectin 1 (FN1) protein is a well-known binding partner of collagen. Further exploring references suggested that the four domains (I6-II1-II2-I7) from FN1 showed the strongest interaction with collagen [6].

▲ Figure 2: Domain structure of fibronectin [6].

Inspired by the 2019 iGEM team Linkoping_Sweden, we found that the thrombin present on the open wound is an ideal candidate to release AMPs from CBD. However, the thrombin cleavage will leave two extra amino acids at the N-terminal of AMPs, which would hamper the activity of released AMPs [7]. To avoid this problem, we wanted to replace the P1’ and P2’ positions of the canonical thrombin cleavage site with the first two amino acids of AMPs. Fortunately, the first two amino acids at N-terminal of DPK-060 are G and K, which show high cleavage activity and are respectively located at P1’ and P2’ positions of thrombin cleavage site [8]. However, the first two amino acids of D2A21 are F and A, which may affect thrombin cleavage. Nevertheless, we decided to conduct thrombin release experiments to examine whether the modification of the thrombin cleavage site would affect the AMP release. Click Result for more details.

Survey of Cell-penetrating peptides (CPPs)

CPPs are short peptides whose lengths are less than 30 amino acids. CPPs can carry non-permeable cargoes, such as peptides and proteins, into human cells through macropinocytosis. Most CPPs are enriched with amino acids with positive charges, such as arginine and lysine residues [9].

To select CPP candidates in our project, we first investigated CPPs with data confirming their functions and safety in vivo or in clinical trials. The result showed that four CPPs fit this requirement (Table 2) [10]. TAT48-57 was the first reported CPP and the research on TAT48-57 is the most extensive. In addition, TAT48-57 is in clinical trials, suggesting its low cytotoxicity and relative safety in medical therapy. Therefore, we decided to use
TAT48-57 in our project and conducted an experiment to examine the penetrating ability of TAT48-57. Click Result for more details.

▲ Figure 3: The mechanism of CPP.


TAT48-57 is derived from the transcription protein TAT in HIV-1. TAT48-57 is a non-amphipathic peptide [9], and like general CPPs. TAT48-57 has highly positive net charges at physiological pH. The obvious difference between TAT48-57 and other CPPs is the uptake mechanism. The docking of TAT48-57 on the negatively-charged proteoglycan on the human cells will activate the Rac pathway to induce cell skeleton reorganization and macropinocytosis [11].

CPP Name Sequence Source Design Approach Structural Features Development Stage
TAT 48-57 GRKKRRQRRR Natural (transcription protein of HIV-1-positions 48–57) Virus derived material Unstructured in buffer solutions; random coil In clinical trials
Penetratin RQIKIWFQNR RMKWKK Natural (Drosophila Antennapedia homeodomain) Derived from natural Antennapedia homeoprotein Secondary amphipathic; forms helices or β-sheets depending on the environment In clinical trials
Polyarginine (R9) RRRRRRRRR Design inspired by entry 1 and 2 Designed to be R-rich Flexible; unstructured; random coil In clinical trials
TP10 AGYLLGKINL KALAALAKKIL Fusion CPP: N-terminal amino acids from galanin (AGYLLGKINLK) linked to mastoparan (ALAALAKKIL) Derived from the neuropeptide galanin linked to a toxin from the wasp venom Primary amphipathic; forms helices in the presence of phospholipids In vivo data

▲ Table 2: The list of CPP candidates [10]

Design of thrombin-releasable TAT-AMP linked with CBD

Similar to the releasable AMPs, P1’ and P2’ positions of the canonical thrombin cleavage site are modified into G and R, the first two amino acids at N-terminal of TATs. We also decided to conduct a thrombin release experiment to examine whether this modification is cleavable. For more details of thrombin cleavage assay, click Result.

Survey of Intracellular AMP and lysosomal endopeptidase
cutting site

The TAT will bring AMPs into macrophages through macropinosomes and fuse with phagolysosomes. In phagolysosomes, endopeptidases cathepsin D, F, O, L, and S are responsible for degrading foreign objects [12]. The survey of preference cleavage sequence of the cathepsin family members indicated that DPK-060 and D2A21 would not be attacked by cathepsins in the lysosome (Table 3). This result further suggested that cathepsins are good candidates to release the AMPs from TAT in phagolysosomes. Since cathepsin S (CTSS) is specifically expressed in antigen presenting cells (e.g., macrophages), we selected DPK-060 as intracellular AMPs and added a CTSS cleavage site to release DPK-060 in lysosomes in macrophage. Because it is not easy to get CTSS, we decided to perform molecular docking to examine the potential CTSS cleavage site. Click Model for more details.

Lysosomal Endopeptidase Expression P2 P1 References
Cathepsin D Ubiquitous V/E L/F [13]
Cathepsin F Ubiquitous L/K/V/F K/P/Y/R [14][15][16]
Cathepsin O Ubiquitous F R [14][17]
Cathepsin L Ubiquitous L/V/F/I K/R [14][18]
Cathepsin S Antigen-presenting cells L/V/F K/R [14][18]
Legumain Ubiquitous N/D [15][18]

▲ Table 3: Endopeptidases in lysosome

Survey and design of supporting components in

Excellent dressings should prevent further infection and accelerate wound recovery, as well as treating any current infection. Accordingly, we aim to generate a composite dressing to provide the following functions.

  • Protective barrier against secondary infection.
  • Absorbing excessive exudate while maintaining a moist environment at the wound interface.
  • Antimicrobial activity, particularly against superbug infection.
  • Non-toxic and biocompatible.
  • Easily removed from the wound.

  • PU film-the outer protective barrier

    The main function of the outer layer is blocking external contamination. After referring to the commonly used secondary dressing in clinical settings, we considered using gauze and polyurethane film. After discussing with the nursing staff at Yi'en home health care center, we understood that a user-friendly product is even better than a high-tech one. A user-friendly dressing should be convenient for patients and caregivers, produce less waste, and stay stably on the wounds when patients move. (Click Integrated‌ ‌Human‌ ‌Practice‌ for more detail.) To this, we selected polyurethane (PU) because of its self-adhesive characteristics. The PU film can be applied to the wounds directly without special training. Compared to the gauze, it is easily removed as it does not stick to the wound or injure the surrounding fragile skin. Furthermore, taking advantage of its pore size, the perspiration readily departs from the skin, while water outside the outer layer cannot enter. Thus, PU film is an appropriate material for assembling the outer layer of AgenT Dressing.

    ▲ Figure 4: Common polyurethane film on the market.

    Chitosan and Alginate-generating an environment suitable for wound recovery

    Chitosan is considered to be a promising biomedical material, due to its hemostatic and antibacterial properties. Alginate is a natural anionic polymer and has been extensively used in many industries due to its high absorbance. Notably, ionic crosslinking occurs between alginate and chitosan, forming a more stable structure with higher absorptive ability. Accordingly, we apply a chitosan and alginate cross-linked complex to absorb the exudate secreted from the wound and maintain a moist healing environment to promote wound healing. We also decided to conduct an experiment to evaluate the best mixing ratio of alginate and chitosan. Click Result for more detail.

    Collagen-promoting wound recovery

    The ionic crosslinking makes alginate and chitosan complex, but may also affect AMPs, which are positively charged. We then tried to identify a protein coating layer that could decrease the ionic influence from the absorptive layer. We found that collagen is the major insoluble fibrous protein in the extracellular matrix, and has low inflammatory properties, good biocompatibility, and low antigenicity. Furthermore, collagen can promote cell proliferation and serve as a sacrificial substrate of wound-secreted excessive matrix metalloproteinases (MMPs, responsible for collagen degradation) to accelerate wound healing.

    Experimental design

    Minimal Inhibitory Concentration (MIC) of AMPs

    MIC analysis is applied to determine the lowest concentration of AMPs that prevents 90% of the growth of microorganisms. This analysis would help us to determine the lowest effective concentration of AMPs on the dressing.


    In brief, E. coli or S. aureus at log phase growth will be incubated with different concentrations of AMPs for 16 hours, and the effect of AMPs on the growing status of bacteria will be determined by OD625 value. The result will be analyzed by statistical software based on the modified Gompertz function to determine the MIC of AMPs [19]. Click Result for more detail.

    Hemolysis assay


    Hemolysis assay is used to determine whether AMPs cause erythrocyte lysis in vitro. The safety standard of hemolysis from ISO-10993 is below 2%, and combining with MIC analysis, these assays will determine the safety range of AMPs concentration on the dressing [20].


    In brief, fresh erythrocytes purified from mouse blood will be incubated with different concentrations of AMPs for 1 hour. The level of hemolysis will be determined by OD414 value, which is the absorbance length of the heme group released by broken erythrocytes. Click Result for more detail.

    Cytotoxicity analysis

    Cytotoxicity analysis is used to examine whether AMPs would damage cells in the open wound. According to the safety standard of ISO-10993 [21], the reduction of cell viability should not be more than 30%.


    In the cytotoxicity assay, we differentiated U937 cells into macrophages and incubated them with different concentrations of AMPs. The cell survival rate is determined by MTS analysis. MTS is metabolized into end products with absorbance by viable cells at OD490 and shows a positive correlation to the viable cell numbers. Click Result for more detail.

    Integrity and swelling test for absorptive layer

    The main function of the absorptive layer in the dressing is to absorb excessive exudate secreted from wounds. Therefore, the absorptive layer should have two basic characteristics: high integrity, and good absorbability. The integrity of alginate and chitosan mixture is dependent on the mixture ratio. Additionally, the mixture ratio of alginate and chitosan also affects absorbability. Therefore, we conducted a swelling test to optimize the mixing ratio of alginate and chitosan to generate an absorptive layer with high integrity and absorbability. Click Result for more detail.

    1. Definition of Integrity:

    To monitor the integrity of alginate and chitosan formed sponge (CA-sponge) without bias, we recorded the light image of the CA-sponge using black paper as a background. The images are recorded with fixed angle, light, and size (the left panel in Figure 5). Because the black background will show in the region with low integrity, we analyzed the black region of each image by Image J (the middle panel of Figure 5). The integrity is defined as the ratio of white areas to all areas. For example, if the ratio of with area is more than 95%, the integrity is higher than 95%.

    2. Definition of absorbability:

    To determine the absorbability, we incubate a dry CA-sponge with the same weight in pH 7.4 of phosphate buffer saline (PBS) at room temperature for 1 hour, and the swelling rate of CA-sponges was determined and calculated as follows (Figure 6):

    Degree of swelling (%) = (Ws-Wd) / Wd × 100%
    (Wd: weight of dried sponge, Ws: weight of swollen sponge.)

    ▲ Figure 5: The first standard---integrity.

    ▲ Figure 6: The second standard--- absorbability.

    Thrombin cleavage experiment:

    Thrombin cleavage experiments are performed to examine whether the modified thrombin cleavage sites are functional. In this experiment, we constructed vectors expressing CBD-eGFP fusion proteins, in which the CBD and eGFP polypeptides are separated by a linker containing modified thrombin cleavage sites. The modified thrombin cleavage sites are GR (TAT48-57), GK (DPK-060), FA (D2A21), YG (TAT47-57), and GR (TAT48-57).

    1. Converting the thrombin unit into physiological value

    To determine the reasonable concentration of thrombin in thrombin cleavage experiments, we undertook a literature search to find the physiological thrombin concentration on wounds. The clinical data indicated that the thrombin activity and concentration are affected by wound condition and the physical condition of patients. Therefore, we reference a research of thrombin concentration in wounds of healthy individual, and select 0.6 μM, the thrombin concentration at early wound stages, to perform our experiment [22].

    From the datasheet of the commercial thrombin, A 1:2000 wt:wt ratio of thrombin to the target protein (equivalent to one unit per milligram of target protein) is generally sufficient for cleavage in 1X Thrombin Cleavage/Capture Buffer at 20°C for 16 hr. The following is our calculation process.

  • Thrombin / Target protein = 1 / 2000 = 5E-07 (g) / E-03 (g)
  • 5E-07 (g) / 37000 (Da) = 1.35E-11 mol
  • Each vial contains 50 units of human thrombin at a concentration of approximately 1 U/µL.

  • 1.35E-11 mol / E-06 L = 13.5 μM = 1 U/μL
  • 0.6 μM = 0.04 U/μL
  • In the end, we used 0.04 U thrombin to conduct the thrombin cleavage assay, a concentration of 0.6 μM.


    To generate and purify the fusion protein for cleavage assay, we designed a protein expressing cassette containing a 6xHis tag at N-terminal for purification, so protein expression can be induced by IPTG treatment. After IPTG induction, the expressed proteins are extracted and the expressed proteins are then purified by a 6xHis affinity column. After eluting purified proteins with the elution buffer, we replace the elution buffer with the thrombin cleavage buffer and determine the final protein concentration through BCA assay. According to the calculations in experimental design, we added 0.04 U thrombin (equal to 0.6 μM, the physiological value) to 10 μg synthetic protein for cleavage, stopped samples at different time points, and ran SDS-PAGE to show the cleavage results. The SDS-PAGE results were analyzed by MATLAB to demonstrate the release rate of the modified thrombin cleavage sites. For more detail of thrombin cleavage assay, click Result and Measurement.

    2. The kinetic model-based estimation of AMP released from dressing

    Apart from the thrombin cleavage assay, we also built a simulation model to estimate the AMP release from the dressing. According to the preference, thrombin may cleave CBD-AMP fusion proteins (defined as “S” in the equation) at two different sites (shown in Figure 7), dividing CBD-AMPs into two or three polypeptides. Through the differential equation and experimental data fitting, we built the kinetic model based on Equations 1~4 using MATLAB. We aim to predict the release rate of antimicrobial agents and take it as a reference to set the effective time for sterilization of our product.

    ▲ Figure 7: The illustration of thrombin cleavage sites and the proposed kinetic models.

    Cell Penetration Ability Test

    This experiment is applied to examine the cell-penetrating ability of TAT48-57. The U937 cells are differentiated into macrophages and applied as target cells in this experiment. To easily monitor the penetration of TAT48-57 in macrophages, we generated constructs expressing TAT48-57-eGFP fusion protein, or expressing eGFP only as a negative control.


    To generate and purify the fusion protein for penetrating assay, we designed a protein expressing cassette containing a 6xHis tag at N-terminal for purification, so the protein expression can be induced by IPTG treatment. After IPTG induction, the expressed proteins are extracted and the expressed proteins are purified using a 6xHis affinity column. After eluting purified proteins with the elution buffer, we replaced the elution buffer with the thrombin cleavage buffer and determined the final protein concentration through BCA assay.

    In brief, 200 nM of purified TAT48-57-eGFP or eGFP were added into the culture medium of U937-differentiated macrophages for 1 hour. The excess TAT48-57-eGFP and eGFP protein were washed out, and the location of TAT48-57-eGFP or eGFP in U937-differentiated macrophages is recorded by a fluorescence microscope.

    Modeling the release activity of CTSS

    To examine whether our designed CTSS cleavage site is reasonable, we docked the tri- or di-peptides from the sequence to the Cathepsin S using AutoDock. We confirmed that the cluster with the lowest binding energy occupies the targeted binding site. The encouraging results suggest that we could validate our designed sequence with simulation to improve the experiments in the future. For more details, click Model.


    1. Lai, Y., & Gallo, R. L. (2009). AMPed up immunity: how antimicrobial peptides have multiple roles in immune defense. Trends in immunology, 30(3), 131–141.

    2. Patrulea, V., Borchard, G., & Jordan, O. (2020). An Update on Antimicrobial Peptides (AMPs) and Their Delivery Strategies for Wound Infections. Pharmaceutics, 12(9), 840.

    3. Koo, HB, Seo, J. (2019)Antimicrobial peptides under clinical investigation. Pept Sci. ; 111:e24122.

    4. Chalekson, C. P., Neumeister, M. W., & Jaynes, J. (2003). Treatment of infected wounds with the antimicrobial peptide D2A21. The Journal of trauma, 54(4), 770–774.

    5. Schmidtchen, A., Pasupuleti, M., Mörgelin, M., Davoudi, M., Alenfall, J., Chalupka, A., & Malmsten, M. (2009). Boosting antimicrobial peptides by hydrophobic oligopeptide end tags. The Journal of biological chemistry, 284(26), 17584–17594.

    6. Addi, C., Murschel, F., & De Crescenzo, G. (2017). Design and Use of Chimeric Proteins Containing a Collagen-Binding Domain for Wound Healing and Bone Regeneration. Tissue engineering. Part B, Reviews, 23(2), 163–182.

    7. Schmidtchen, A., Pasupuleti, M., Mörgelin, M., Davoudi, M., Alenfall, J., Chalupka, A., & Malmsten, M. (2009). Boosting antimicrobial peptides by hydrophobic oligopeptide end tags. The Journal of biological chemistry, 284(26), 17584–17594.

    8. Muchintala, D., Suresh, V., Raju, D., & Sashidhar, R. B. (2020). Synthesis and characterization of cecropin peptide-based silver nanocomposites: Its antibacterial activity and mode of action. Materials Science and Engineering: C, 110, 110712.

    9. Madani, F., Lindberg, S., Langel, U., Futaki, S., & Gräslund, A. (2011). Mechanisms of cellular uptake of cell-penetrating peptides.Journal of biophysics (Hindawi Publishing Corporation : Online), 2011, 414729.

    10. Kalafatovic, D., & Giralt, E. (2017). Cell-Penetrating Peptides: Design Strategies beyond Primary Structure and Amphipathicity. Molecules (Basel, Switzerland), 22(11), 1929.

    11. Futaki, S., Nakase, I., Tadokoro, A., Takeuchi, T., & Jones, A. T. (2007). Arginine-rich peptides and their internalization mechanisms. Biochemical Society transactions, 35(Pt 4), 784–787.

    12. Ceuleers, H., Van Spaendonk, H., Hanning, N., Heirbaut, J., Lambeir, A. M., Joossens, J., Augustyns, K., De Man, J. G., De Meester, I., & De Winter, B. Y. (2016). Visceral hypersensitivity in inflammatory bowel diseases and irritable bowel syndrome: The role of proteases. World journal of gastroenterology, 22(47), 10275–10286.

    13. Sun, H., Lou, X., Shan, Q., Zhang, J., Zhu, X., Zhang, J., Wang, Y., Xie, Y., Xu, N., & Liu, S. (2013). Proteolytic characteristics of cathepsin D related to the recognition and cleavage of its target proteins. PloS one, 8(6), e65733.

    14. Vidak, E., Javoršek, U., Vizovišek, M., & Turk, B. (2019). Cysteine Cathepsins and their Extracellular Roles: Shaping the Microenvironment. Cells, 8(3), 264.

    15. Shi, G. P., Bryant, R. A., Riese, R., Verhelst, S., Driessen, C., Li, Z., Bromme, D., Ploegh, H. L., & Chapman, H. A. (2000). Role for cathepsin F in invariant chain processing and major histocompatibility complex class II peptide loading by macrophages. The Journal of experimental medicine, 191(7), 1177–1186.

    16. Wang, B., Shi, G. P., Yao, P. M., Li, Z., Chapman, H. A., & Brömme, D. (1998). Human cathepsin F. Molecular cloning, functional expression, tissue localization, and enzymatic characterization. The Journal of biological chemistry, 273(48), 32000–32008.

    17. Velasco, G., Ferrando, A. A., Puente, X. S., Sánchez, L. M., & López-Otín, C. (1994). Human cathepsin O. Molecular cloning from a breast carcinoma, production of the active enzyme in Escherichia coli, and expression analysis in human tissues. The Journal of biological chemistry, 269(43), 27136–27142.

    18. Vidmar, R., Vizovišek, M., Turk, D., Turk, B., & Fonović, M. (2017). Protease cleavage site fingerprinting by label-free in-gel degradomics reveals pH-dependent specificity switch of legumain. The EMBO journal, 36(16), 2455–2465.

    19. Lambert, R. J., & Pearson, J. (2000). Susceptibility testing: accurate and reproducible minimum inhibitory concentration (MIC) and non-inhibitory concentration (NIC) values. Journal of applied microbiology, 88(5), 784–790.

    20. Goyal, T., & Schmotzer, C. L. (2015). Validation of hemolysis index thresholds optimizes detection of clinically significant hemolysis. American journal of clinical pathology, 143(4), 579–583.

    21. ISO - International Organization for Standardization. (2017). ISO 10993-5:2009. ISO

    22. Mann, K. G., Brummel, K., & Butenas, S. (2003). What is all that thrombin for?. Journal of thrombosis and haemostasis : JTH, 1(7), 1504–1514.