Our main goal is to create a sensitive, non-invasive, and accessible CRC screening method by applying RNA switch (Loop Initiated RNA Activator-LIRA) on-off system production and degradation according to either presence or absence of miR21 and miR92a. This page displays six engineering cycles explaining the development process of biodevice based EcN-1917 and optimized design of RNA switch. These cycles show how we tackled challenges during our project, what we learned from each engineering iteration, and what we did to improve the design.
CYCLE 1: Test aeBlue expression in EcN

Design: Determining vector (EcN) and reporter (aeBlue chromoprotein)

Escherichia coli Nissle 1917 (EcN) is a probiotic nonpathogenic that can specifically be colonized in colorectal tumor tissue compared than others bacteria (Alizadeh et al., 2020). The parts of that bacteria specifically colonized is K5 capsule and F1C fimbriae (Chiang & Huang, 2021) that can interact with Toll-like receptor 4 (TLR-4) and Toll-like receptor 5 (TLR-5) in colorectal tumor cell (Hafez et al., 2010). EcN can be modified for producing a protein by transforming a plasmid vector that bring a specific gene (Danino et al., 2015). Therefore, EcN is potentially probiotic that can detect colorectal tumor and can be developed by protein synthesize modifying. Finally, it can be observed in feces.
aeBlue chromoprotein plasmid vector is a plasmid that brings an aeBlue gene that can synthesize aeBlue protein (Liljeruhm et al., 2018). This protein came from Actinia equina, coloring blue in absorbance 597 nm (Shkrob et al., 2005). This chromoprotein is reacted with oxygen to form chromophore pigment with maturation time in 24 minutes (Bao et al., 2020 and Liljeruhm et al., 2018). aeBlue chromoprotein plasmid consists of constitutive promoter medium BBa_J23110 and ribosome binding site (RBS) BBa_B0034 in pSB1C3 plasmid that bring CampR (chloramphenicol resistance) gene. After the plasmid successfully transformed in E.coli, the colonies will be in blue color under the proper light with proper concentration of oxygen (Liljeruhm et al., 2018). Therefore, EcN modified with aeBlue chromoprotein plasmid can be easily observe within feces. It can be a new method of more sensitive, easy, cheap, and comfortable colorectal cancer screening.


  1. Escherichia coli Nissle 1917 (EcN) Isolation
    Escherichia coli 1917 is isolated from Mutaflor capsule then is cultivated in LB media shows in Figure 1. Because of EcN containing pMUT1 and pMUT2, it is confirmed by Colony PCR with selected each pair primers. The product of pMUT1 shows in 361 bp while pMUT2 shows in 429 bp (Kan et al., 2021) explained in Figure 2.
    Figure 1. EcN isolate
    Figure 2. Colony PCR Result of pMUT1 and pMUT2 in EcN

  2. Plasmid Curing
    The curing method is used to remove pMUT1 by transforming pFREE and pMUT2 by transforming pCryptDel from EcN sequentially. pMUT1 is successfully removed by pFREE transformation and confirmed with colony PCR in >250 bp. That result is shown in Figure 3. However, the next step of removing pMUT2 by pCryptDel transformation has not been successfully worked.
    Figure 3. Colony PCR of pFREE transformation M: DNA Ladder 1 kb 1,2,3,5: Colonies with pMUT1 detected 4: Colonies without pMUT1 detected

  3. aeBlue Chromoprotein Isolation
    aeBlue chromoprotein plasmid is consisted of aeBlue gene which the codon is already optimized for E. coli expression, BBa_J23110 constitutive medium promoter, Ribosome Binding Site (RBS) BBa_B0034 in pSB1C3 plasmid that brings CampR (chloramphenicol resistance) gene (Liljeruhm et al., 2018). Because of the pSB1C3 plasmid is a BioBrick plasmid backbone, the isolated plasmid is being amplified with primer VF2 and VR and resulted in 1074 bp, showing in Figure 4.
    Figure 4. Isolation of aeBlue chromoprotein plasmid. K= control.

  4. Transformation
    aeBlue chromoprotein plasmid is transformed into EcN competent cell. The EcN-aeBlue colonies are shown in Figure 5. The colonies grow in blue color in LB agar with chloramphenicol resistance.
    Figure 5. EcN-aeBlue colonies


  1. Colony PCR
    The colonies are confirmed with colony PCR and electrophoresis showing in Figure 6 with length 1000 bp.
    Figure 6. Colony PCR of EcN-aeBlue. K= control, M= DNA Ladder

  2. Plasmid Stability Assay
    Stability assay of aeBlue chromoprotein plasmid is used to observe the expression of chomoprotein cycle and repetition. The stability decrease when the expression of chromoprotein is disappear before reaching the second cycle in each repetitions. The result of stability assay shows that EcN-aeBlue is keep expressing the chromoprotein in every cycles in each repetitions, eventhough the expression is decreasing in third cycle and so on. It shows in Table 1.
    Table 1. Plasmid Stability Assay of EcN-aeBlue interpretation
    ++ = strong coloured in blue
    + = weak coloured in blue
    - = no coloured in blue

  3. Modified EcN Growth Rate Assay
    The growth rate is observed by measuring the absorbance of EcN-aeBlue compared with EcN as control. The growth rate curve shows a decrase about 0,06 shown in Figure 7.
    Figure 7. Growth rate curve of EcN-aeBlue and EcN

  4. Modified EcN Toxicity Assay
    Toxicity Assay was done by inoculating EcN-aeBlue cells and EcN control cells into CRC cell culture inside 96-well plates. The amount of inoculated bacteria was varied to test for viability on various values of MOI (multiplicity of infection), i.e. the ratio of bacterial cells to cancer cells. Cell viability were then estimated using MTT assay.
    The results of the toxicity assay is shown in Table 2. The differences between EcN-aeBlue and EcN inoculated CRC cell culture growth are not significant (α=0,05), even with increasing MOI. Therefore, both EcN-aeBlue and EcN are not toxic for colorectal cancer cell line.
    Table 2. Toxicity Assay of EcN-aeBlue and EcN in colorectal cancer cell line


From our first cycle of experiment, we gained invaluable knowledge that could help us in subsequent cycles, such as:
  1. E. coli Nissle 1917 could be modified by transforming aeBlue chromoprotein-carrying plasmid and was able to synthesize blue-colored chromoproteins

  2. aeBlue chromoprotein containing plasmid is stable within EcN cells

  3. There isn’t any significant decrease in growth rate between control and transformed EcN cells

  4. Unmodified EcN is non-toxic towards colorectal cancer (CRC) cell line.

CYCLE 2: Early RNA Switch Consideration Design


In research by Wang et al. (2019), the expression of specific miRNAs in mammalian cells can be monitored using a single-stranded nucleic acid molecular detector called RNA-toehold switch (RTS). RTS is a posttranscriptional riboregulator added to the 5' UTR of mRNA. RTS is composed of several parts (Figure 1).
Figure 1. Arrangement and RTS secondary structure. Modified image from (Baabu et al., 2022)
Figure 2. RTS working mechanism. Modified image from (Heo et al., 2021).
In this research, a chromoprotein-sensitive miR21 biodevice design can be developed using RTS which can detect miR21 (CRC biomarker) (Yu et al., 2015) and then initiate the synthesis of blue chromoprotein.
Figure 3. Working mechanism of the biodevice chromoprotein-sensitive miR21 on CRC
However, through further analysis, problems were found in the working mechanism of RTS in detecting miR21. This problem is caused by the complementary sequence of miR21, which acts as the ascending bottom stem, containing a stop codon that is likely to interfere with the translation process. The sequence in RTS can be changed so that no stop codon is found. However, there will be nucleotide pairs that do not comply with the Watson-Crick Rule which will likely affect the stability of the RTS secondary structure.
In the research of Baabu et al. (2022), a new RTS called RTS second generation was developed with a different working mechanism from the previous RTS. In detecting inducer molecules (miRNA), RTS second generation uses the help of a co-inducer molecule (anti miRNA). The working mechanism of RTS second generation in detecting miRNA begins with the formation of a miRNA-antimiRNA complex (Figure 4a). The miRNA-antimiRNA complex will then bind to the toehold domain and ascending bottom stem thereby opening the hairpin structure (Figure 4b) and allowing the ribosome to bind to the RBS (Figure 4c) and initiating mRNA translation.
Unlike previous RTS, in RTS second generation the ascending bottom stem sequence is designed to be complementary to some antimiRNA sequences. Thus, the use of RTS second generation can overcome the problems of previous RTS by designing the complementary sequence of antimiRNA, which acts as an ascending bottom stem, so that it does not contain stop codons it will not interfere with the translation process.
Figure 4. The mechanism of action of RTS second generation involves RTS, miRNA, and antimiRNA (Baabu et al., 2022).
Figure 5. Mechanism of action of a chromoprotein-sensitive miR21 biodevice using second-generation RTS under normal conditions.
Figure 6. Mechanism of action of a chromoprotein-sensitive miR21 biodevice using second generation RTS under CRC conditions.


Building the plasmid constructs carrying chromoprotein-sensitive miR21 biodevices including RTS first generation (Figure 5) and RTS second generation (Figure 6) begins with collecting and designing biodevice parts and vector plasmids. Next, the biodevice parts and vector plasmids were combined with Gibson Assembly using Benchling software. Additionally, the assembly simulation also resulted in a pair of primer designs that could be used to assemble the biodevice parts and plasmid vector in vivo. For further information on our part, vector plasmid, and primer design can be seen in Part Page.
We also built an Ordinary Differential Equation (ODE) that begins by creating a chemical equation for each reaction from the working mechanism scheme for the chromoprotein-sensitive miR21 biodevice design. Next, the chemical reaction equation is translated into ODE. For, chemical equation, and ODE can be seen in Model Page.


The miR21 chromoprotein-sensitive biodevice was developed based on the RTS first generation and RTS second generation basic framework. The structural stability of both designs was tested and analyzed using the NUPACK software. Analysis results of the RTS first generation and second generation can be seen on Model Page. Based on the test results, RTS that has an ideal structure (RBS and start codon opened when there is any biomarker) is RTS first generation.
In this experiment, also, four kinetic modeling simulations were also carried out, namely for RTS first generation with an initial extracellular miR21 concentration of 2 M and 0.2 M and RTS second generation with an initial extracellular miR21 concentration of 2 M and 0.2 M. Based on the concentration fluctuation data in the appendix, the duration of aeBlue chromoprotein can be clearly seen for approximately 2 hours, starting from 46 seconds after miR21 first enters the EcN with the biodevice. Meanwhile, the maximum concentration of aeBlue chromoprotein in RTS second generation was reached at a value of 0.041281 M and at 1098.75763 seconds or around 18 minutes after extracellular miR21 with an initial concentration of 2 M first diffused into the EcN with the biodevice. Based on the concentration fluctuation data in the appendix, the duration of aeBlue chromoprotein can be clearly seen for approximately 7.6 hours, starting from 2.8 seconds after miR21 first enters the EcN with the biodevice. The complete result of kinetic modeling can be seen on Model Page.


Although RTS second generation doesn’t have a stop codon in the toehold domain (detector segment) and the longer duration of aeBlue chromoprotein visibility. But, from the in silico test, namely Analysis Job NUPACK, It doesn’t have an ideal structure when there is any miR21. Because RBS and the start codon of RTS second generation are still in a close state. To ensure this, we need an in vitro test whether RTS second generation will work or won’t work, or we need to re-design the sequence(s) of RTS to get the ideal structure in silico test. After we consider, There is also the possibility that antimiR21 complements with RTS before a hairpin structure is formed or before antimiR21 complements with miR21 resulting false positive.
CYCLE 3: LiRA OR Gate Consideration Design


In this cycle, we redesigned our biodevice so can detect two biomarkers that are upregulated in different CRC phases and work as OR Gate logic to increase the sensitivity of our biodevice. This is based on the literature that CRC shows the upregulated expression of miR21 (advanced adenoma) and miR92a (early adenoma) (Okugawa et al. 2014). The reason why we chose LIRA is that it can be designed to detect two biomarkers in an OR Gate logic manner (Ma et al. 2022) but only use one transcription unit so it can minimize the burden of the cell. The miR21 detector domain, which has a stop codon, also can be put far enough from RBS and start codon compared to RTS first generation so it is expected will not disturb the translation process. miR21 detector domain will be located in the upstream region and continued by the second detector domain.


The design was carried out by first re-simulating the LIRA OR gate from the reference (Ma et al. 2022) using an Analysis Job from NUPACK Software So we can determine the parts that makeup LIRA, including the Ribosomal Binding Site (RBS), start codon, and detector part. Then, substitute the detector part so that it can detect CRC biomarkers (miR21 and miR92a) by making a complement of the two miRNAs.


For testing, we re-simulate the LIRA job analysis where the detector part has been replaced. Modify the nucleotides that makeup LIRA to obtain the expected structure. The expected LIRA structure is a LIRA structure where the RBS and start codon remain closed when there is no miR21 or miR92a, or neither. And the RBS section and start codon will open when there is miR21, miR92a, or both. The complete results of resimulation can be seen on the Model Page.


After getting the structures and constraints that we need, there are known subsequences that can be randomized. But, we know that it's hard to try every possibility of sub-sequence and these would be impractical for wetlab. So, we are considering making an automated pipeline to reconsiderate toehold design in order to optimize toehold performance as a switch.
CYCLE 4: LiRA Optimized Design by Drylab Approach Metrics


In designing screening tools based on biomarkers, it's important to know how to optimize switch ON-OFF logic performance as it would affect specificity and sensitivity performance of screening tools. To optimize LiRA switch OR gate, sequence customization needs to be evaluated. To do that, we sampled design 10 LiRA sequences (sequence gathering) using NUPACK based on our constraints then calculated ON OFF analogically approach metrics to all sequences. In this cycle, we use limited data of LiRAs such as parameters (pair probabilities and rank concentration results) and its secondary structure from design and analysis job NUPACK. From its secondary structure, we made an analogical approach to ON OFF logic and got several metrics to measure our goals of design. From that approached metrics, we choose the best sequence LIRA as our main toehold switch. Analogical approach scheme made by following scheme at Figure 1 then produces off level and on level metrics for one complex at Equation (1) and (2).
Figure 1. Analogical Approach Metrics of ON and OFF Level
off-level=i=1nProbicLRBStostartcodon(1)\text{off-level} = \frac{\sum_{i=1}^n Prob^c_i}{L_{RBS-to-startcodon}}\cdot\cdot\cdot\cdot(1)
on-level=i=1nProbioLRBStostartcodon(2)\text{on-level} = \frac{\sum_{i=1}^n Prob^o_i}{L_{RBS-to-startcodon}}\cdot\cdot\cdot\cdot(2)


LIRA biodevice sequences and its corresponding primer pairs from the dry lab team was subsequently submitted to sequencing company for synthesis. Synthesized LIRA fragments were then prepared for insertion by adding BioBrick prefix and suffix on both ends respectively using Hi-Fi PCR. Backbone plasmids (pMUT1) were isolated from EcN cells and were also prepared for assembly the same way. Assembly were done using Gibson Assembly protocol. Lastly, assembly products including controls were transformed into competent EcN cells.


Transformed bacterial colonies were then checked for transformation success using colony PCR. PCR results shows that no insert were detected inside the tested colony. To confirm whether or not our prepared assembly fragments were indeed correct (i.e. flanked by BioBrick prefix and suffix, thus allowing assembly by Gibson Assembly), we submitted our purified PCR results for sequencing. The sequencing results shows multiple deletion (red) and point mutation along the length of the sample for both the insert (LIRAreg2 or LIRA-OR-A2aeB + GFP sekuens + biobrick) at Figure 3and backbone (pMUT1) at Figure 2. These results offer an explanation for the poor assembly performance.
Figure 2. Alignment result between sequenced pMUT1 PCR product vs pMUT1 design
Figure 3. Alignment result between sequenced LIRAreg2 (LIRA-OR-A2aeB + GFP sekuens + biobrick) PCR product vs LIRAreg2 (LIRA-OR-A2aeB + GFP sekuens + biobrick) design


From the design process, we have several metrics from the definition approach of drylab. These metrics can be measured by wetlab too, in wetlab definition, and can be compared to each other sequences. Developing metrics is just an early step to have more optimized design of LiRA OR gate sequences. We are curious to know what and how parameters could be affect much into our metrics approach values. Those process would be covered in next cycle to gain more insight on our metrics approach and made pipeline that can be substitute to wetlab data if there is any.
During the assembly process and the subsequent wet lab experiment, we learned that a more rigorous selection scheme is required to ensure that the recombinant plasmid is successfully transformed and reliably maintained within the bacterial cell. Furthermore, better quality control and checks are needed to ensure that the DNA fragments prepared during the preparation steps are of high quality and does contain the desired overhangs. Lastly, better logistics and preparation steps are paramount in ensuring steady supply of reagents and DNA molecules throughout the lengthy and oftentimes repeated cycles of experiments.
CYCLE 5: LiRA optimized design pipeline based regression


To improve the design process of LiRA OR gate sequence, we made a regression pipeline by exploring parameters that can be known from sequence (parameter gathering) and done regression on drylab approach metrics. Before building the pipeline, we design the flowchart process at Figure 1 to know step-by-step how to build the pipeline. Then we also define each parameter and its correlation into either ON OFF performances or secondary structure.
Figure 1. Toehold Optimization Design Pipeline

Build and Test

To build this cycle, we used complex, sequence, and sub-sequence (sequence split into 3 sub-sequences) as input that got from the previous process (Design and Analysis Job NUPACK in sequence gathering). We built this pipeline using some library based Python especially Pandas, Scikit-Learn, AutoGluon, NUPACK, and ViennaRNA. The build process itself consists of parameter gathering and regression (preprocessing until model training process) while test process are evaluation process of regression model (model metrics interpretation) but there are some sub-cycle under regression process due to our evaluation about R-squared result that not fulfill our hypothesis and our hypothesis target on RMSE. The regression process built as 3 repeated sub-cycle consists of ‘Built’ and ‘Test’. ‘Built’ by constructing algorithms code and ‘Test’ by evaluating the metrics of regression performance such as R^2 and RMSE (Root Mean Squared Error). The process by regression performance metrics in Figure 2 explained below:
Figure 2. Regression Performance Metrics Results
  • Sub cycle 1: Linear regression
    There might be several reasons why we got pretty high RMSE, it might be because the data is too complex/not enough to show the pattern of data. Since we only use 10 sampled sequences.

  • Sub cycle 2: Linear regression with selected features
    As it might be that the data are too complex, we try to use linear regression with selected features and get lower RMSE and MSE for weighted metrics. But, we still got negative R-squared. The negative R-squared might happen because the parameters are not linearly correlated to the approached metrics we made. To know whether from our 10 sequences data we can get a model that can be predicting a regression use case with positive R-squared, we use AutoML.

  • Sub cycle 3: Auto Gluon regression
    Auto Gluon is known as one of the best ways to get easy comparisons of several models and known by its algorithm that also tries to use the stack and ensemble of several machine learning models. So, we run auto gluon regression on kaggle. From each metric, we got at least one model that got positive R-squared in each approached metric. These efforts show that with very small data, there also needs some effort on model building to get expected model metrics.


From the development of the regression pipeline process to optimize LiRA metrics performance, we can conclude that our models and metrics approach can be developed and customized for anyone who wants to optimize their biological switch. On other way, there are still chance to future development such as add more sequence data, add wet lab data to bring model closer to real case, and expand knowledge to understands context of sequences into algorithm with NLP approach such as N-Grams and so on.
CYCLE 6: LiRA expression system : kinetic modeling 2nd cycle


In the design phase of this 6th cycle, we designed a simulation scheme by understanding the mechanism of the LIRA expression system in EcN in the colonic lumen environment, the complete mechanism can be seen on the Page Model


In the build phase, based on the mechanism we understand, we decided to create an ODE scheme involving two types of miRNAs and 1 type of miRNA with initial extracellular miRNA value input in the range of important constituents on the human extracellular fluid from the lowest normal value until the highest value of approximate short-term nonlethal limit, that is 1.2x10-3 M - 175x10-3 M (Hall and Hall 2020). For complete equations, see the Page Model


Based on the kinetic simulation test results, modeling can be seen that our biodevice, both the two-miRNA and one-miRNA schemes still has the possibility of being able to detect the presence of CRC because at the input miRNA with the lowest concentration, namely the lowest normal value (1.2x10-3 M), the visible color of aeBlue can last for approximately 12.15 hours. Then, when it is increased 10x from the normal concentration, the duration increases to 14.53 hours, based on the data, this duration has entered the range of 12-24 hours (Sensoy 2021) and enters the transit time range in the sigmoid-rectum (12.7 ± 2.1 hours) (Tomita et al. al. 2011). The complete result of the simulation test can be seen on the Page Model


From this cycle, we learned that for future development of the DBTL model and cycle, it is necessary to consider the dynamic parameters of the EcN population and adjust the dose based on each person's circadian defecation rhythm. Apart from that, adjustments can also be made in terms of the circadian rhythm of defecation, for example by making probiotic preparations which can also facilitate the defecation process, or other alternative adjustments as simple as advising EcN biodevice consumers to eat healthy foods rich in fiber which can facilitate digestion, we are sure This improvement can be carried out in stages until a biodevice design that is close to ideal is obtained. We also need a value of miRNA concentration range of CRC patients that is more reliable.
Alizadeh, S., Barzegari, A., Esmaeili, A., and Omidi, Y. 2020. Designing a light-activated recombinant alpha hemolysin for colorectal cancer targeting. BioImpacts: BI. 10(3): 187-193.
Bao, L., Menon, P., Liljeruhm, J., and Forster, A. C. 2020. Overcoming chomoprotein limitations by engineering a red fluorescent protein. Analytical biochemistry. 611: 113936.
Baabu, P.R.S., Srinivasan, S., Nagarajan, S., Muthamilselvan, S., Selvi, T., Suresh, R.R. and Palaniappan, A., 2022. End-to-end computational approach to the design of RNA biosensors for detecting miRNA biomarkers of cervical cancer.Synthetic and Systems Biotechnology, 7(2), pp.802-814.
Kan, A., Gelfat, I., Emani, S., Praveschotinunt, P., and Joshi, N. S. 2021. Plasmid Vectors for in Vivo Selection-Free Use with the Probiotic E. coli Nissle 1917. ACS synthetic biology,10(1): 94-106.
Chiang, C. J. and Huang, P. H. 2021. Metabolic engineering of probiotic Escherichia coli for cytolytic therapy of tumors. Scientific reports. 11(1): 5853.
Danino, T., Prindle, A., Kwong, G. A., Skalak, M., Li, H., Allen, K., Hasty, J., and Bhatia, S. N. 2015. Programmable probiotics for detection of cancer in urine. Science translational medicine. 7(289): 289ra84.
Hall, J.E. and Hall, M.E., 2020. Guyton and Hall textbook of medical physiology e-Book. Elsevier Health Sciences.
Hafez, M., Hayes, K., Goldrick, M., Grencis, R.K., and Roberts, I. S. 2010. The K5 capsule of Escherichia coli strain Nissle 1917 is important in stimulating expression of Toll-like receptor 5, CD14, MyD88, and TRIF together with the induction of interleukin-8 expression via the mitogen-activated protein kinase pathway in epithelial cells. Infection and immunity. 78(5): 2153–2162.
Heo, T., Kang, H., Choi, S., and Kim, J. 2021. Detection of pks Island mRNAs Using Toehold Sensors in Escherichia coli. Life. 11(11): 1280.
Liljeruhm, J., Funk, S. K., Tietscher, S., Edlund, A. D., Jamal, S., Wistrand-Yuen, P., Dyrhage, K., Gynna, A., Ivermark, K., Lovgren, J., Tornblom, V., Virtanen, A., Lundin, E. R., Wistrand-Yuen, E., and Forster, A. C. 2018. Engineering a palette of eukaryotic chromoproteins for bacterial synthetic biology. Journal of biological engineering. 12: 8.
Ma, D., Li, Y., Wu, K., Yan, Z., Tang, A.A., Chaudhary, S., Ticktin, Z.M., Alcantar-Fernandez, J., Moreno-Camacho, J.L., Campos-Romero, A. and Green, A.A., 2022. Multi-arm RNA junctions encoding molecular logic unconstrained by input sequence for versatile cell-free diagnostics. Nature biomedical engineering, 6(3), pp.298-309.
Okugawa, Y., Toiyama, Y. and Goel, A., 2014. An update on microRNAs as colorectal cancer biomarkers: where are we and what’s next?. Expert review of molecular diagnostics, 14(8), pp.999-1021.
Sensoy, I., 2021. A review on the food digestion in the digestive tract and the used in vitro models. Current research in food science, 4, pp.308-319.
Tomita, R., Igarashi, S., Ikeda, T., Sugito, K., Sakurai, K., Fujisaki, S., Koshinaga, T. and Shibata, M., 2011. Study of segmental colonic transit time in healthy men. Hepato-gastroenterology, 58(110-111), pp.1519-1522.
Skhrob, M. A., Yanushevich, Y. G., Chudakov, D. M., Gurskaya, N. G., Labas, Y. A., Poponov, S. Y., Mudrik, N. N., Lukyanov, S., and Lukyanov, K. A. 2005. Far-red fluorescent proteins evolved from a blue chromoprotein from Actinia equina. The Biochemical Journal. 392(Pt 3): 694-654.
Wang, S., Emery, N.J. and Liu, A.P., 2019. A novel synthetic toehold switch for microRNA detection in mammalian cells. ACS synthetic biology, 8(5), pp.1079-1088.
Yu, Y., Nangia-Makker, P., Farhana, L. et al. miR-21 and miR-145 cooperation in regulation of colon cancer stem cells. Mol Cancer 14, 98 (2015).
Angkasa PuraLiew EnzymLiew StrategicPLNKopi BundaRIPanin GroupSatria Budi Dharma SetiaToyotaNutrilabNutrilab


Bulaksumur F11, Caturtunggal, Kecamatan Depok, Kabupaten Sleman, Daerah Istimewa Yogyakarta, Indonesia 55281

Get in Touch!

All content on this wiki is available under the Creative Commons Attribution 4.0 license (or any later version).