once claimed to be able to quickly detect the new crown, and the SHERLOCK system, which is nearly 9 times faster than fluorescent PCR (polymerase chain reaction), has been silent in the public opinion environment since it was approved by the FDA for emergency testing in May.
Until this Tuesday, in an online sharing session, Zhang Feng, the main developer of the SHERLOCK system and one of the pioneers of the CRISPR application field, appeared again and shared the development history and technical solution paths behind this new crown detection tool in nearly 40 minutes. The
SHERLOCK system is based on the gene editing tool CRISPR-Cas13. It initially discovered RNA sequences of viruses such as Zika and dengue in a variety of human samples (such as saliva), and even some sequences related to cancer mutations. Its detection sensitivity for RNA and DNA reaches the single-molecular level.
This test method is similar to a pregnancy test stick. It only takes one test strip to display the results of the virus infection test. In the detection of the new coronavirus, this technology can determine whether the new coronavirus exists in just 1 hour.
"Our product finally solved two key problems through two steps of transformation," Zhang Feng said during the sharing process.
Shenghui recorded Zhang Feng's sharing process throughout the process. The following is the sharing record based on the original intention:
I have been developing a CRISPR auxiliary system for molecular diagnosis in the laboratory of MIT . A few years ago, we developed SHERLOCK for detection of SARS viruses and Zika viruses, among others. At the beginning of this year, I read about a new virus in " New York Times ", which is the current novel coronavirus. At that time, many chain reactions broke out where the virus appeared. I think maybe SHERLOCK can be used to detect this new virus. So, our laboratory began to redesign different primers and Guide-RNAs (guiding RNAs), hoping to find out some of the sequences of the new coronavirus.
But a few weeks later, the virus sequence of COVID-19 was officially announced. In late January, I began to receive a lot of emails from my colleagues in China. One of my colleagues in Wuhan sent me this photo of him standing near the Wuhan Railway Station in early February. He said that there were crowds of people in the past, but because of the spread of the new coronavirus, we had to block the entire city. Can you cooperate with us to develop a method to achieve rapid diagnosis in the early stages of the spread of the virus.
So we started this project. The working principle of
SHERLOCK is as follows. It requires two steps of reaction, first of all, amplification, but the amplification process of SHERLOCK is very simple and does not require a PCR process, but isothermal amplification method. This means that after the virus is amplified isothermal, a Guide-RNA can be added to the CRISPR protein to accurately identify the virus sequence. In the process of identifying the virus sequence, the Cas protein can not only cleave the virus sequence, but also cleave the probe and then read the results.
The probe we use is a shorter allele that carries molecules at both ends, which allows it to display the results on the test strip. If the virus exists, you will see two lines. If the virus does not exist, you can only see one line. This is very similar to the way the pregnancy test strips work. We then made simple assay kits with developed primers and Guide-RNA for use by scientists and hospitals around the world, and began collecting test results.
The first set of data we received was from the Keith Jerome and Greninger group at the University of Washington.
In March, Seattle was one of the most rampant areas of the coronavirus, so the University of Washington research team visited different patient samples and carried out tests with the SHERLOCK system. They found that after extracting saliva samples from nasopharyngeal swabs, SHERLOCK was able to accurately detect the virus. The test strip (below) shows that the horizontal line appears above represents the existence of the virus, and the horizontal line below is a positive control. We can see that the detection line appears above all the positive samples.
In addition to working with Keith's group, another collaborator also tested SHERLOCK with the above team.They found from the sampling test that the sensitivity was quite good, able to reach 93%, and at the same time it was 100% specific, and did not show false positive signal. Based on this data, they started applying for permission and used SHERLOCK on a wide range of scale in local hospitals in Thailand, and we also received effective feedback (pictured below).
Based on feedback from different collaborators, we learned that SHERLOCK can effectively detect the new coronavirus, , but it still has a disadvantage that cannot be ignored. SHERLOCK detection is a two-step reaction, which involves running amplification, opening the amplification tube and moving the reagent into the second reaction tube, hiding the risk of reagent contamination, which may even induce false positives.
So we tried to solve this problem and developed a new reaction system, which we call STOPCovid 1, which is part of the SHERLOCK detection. We integrate the amplification process and the CRISPR detection process into a single reaction. It works like this: collects saliva samples through a nasopharyngeal swab, then extracts the RNA in it, lets it enter a copy reaction and incubates it in an environment of 60°C for an hour, and finally puts the test strip directly into the reaction tube to read the virus test results.
But some problems must be solved before implementing the above process.
first, simplifies the extraction of viral RNA. As early as March or April, RNA extraction kits and extraction columns were very short, so we developed the Extract buffer , called QuickExtract, which is commonly used to analyze the RNA genome of the virus. It turns out that QuickExtract can quickly break down the saliva of nasopharyngeal swabs and release viral RNA without the need for processes such as extraction column purification. At the same time, we found that QuickExtract performed similarly with the sensitivity levels detected by extraction columns (below).
The second question is reaction temperature. The reaction temperature of STOPCovid 1 is 60°C, while most of the CRISPR proteins we selected before are unstable, and their operating temperature is at 37°C. At 60°C, the protein will become inactive. So we looked for the right protein from thermophilic bacteria and found two suitable samples: proteins from Aac bacteria and Aap bacteria can survive stably at 60°, so that we can achieve the amplification and detection process. From the figure, we can see that the protein (blue line) of Aap bacteria has better effect and stronger stability. We used it to detect about 200 samples to obtain the following data graph.
But we still hope to make the detection faster and recognize more signals. To achieve this, we tested different reaction additives to enhance the reaction process. Therefore, we ordered a variety of different compounds to improve reaction efficiency. Through systematic analysis, found that glycine (Glycine) and taurine (Taurine) are the best additives, which can increase the reaction speed, signal strength and reaction capacity at the same time. You may have heard of taurine, which can enhance human body functions and allow us to stay up late to read papers (laughs).
above is part of the work we have done. We have combined each of the above transformation processes to develop the STOPCovid 1 system. We also compared him with the LAMP detection method (loop-mediated isothermal amplification). As you can see from the left image, LAMP will have false positive signals over time. The three lines on the right in the figure show that their actual signal output is below the detection threshold. The data in the figure on the right shows that if CRISPR protein and the above additives are added, the false positive signal can be removed and 100% detection specificity can be achieved. This reaction can even be carried out in a simple water bath environment, and the low-cost investment can obtain accurate test results.
However, compared with the SHERLOCK system, the STOPCovid 1 system is simple and easy, but the detection sensitivity is still not at the same level. Therefore, we have developed the STOPCovid 2 system, and uses magnetic beads (BEAD) to optimize the RNA extraction and enrichment process to improve specificity and sensitivity.The main working method of
STOPCovid 2 is to add nasopharyngeal swab samples to the reaction tube, use magnetic control to achieve cleavage of viral RNA, and then use BEAD to "capture" RNA to extract it. At this point, the lysis buffer of RNA remains on BEAD, but has been concentrated to a smaller volume. Add STOP Covid reaction reagent, and after sufficient dissolution, it can be used for detection by fluorescence or test strips.
This is the comparison between (below) STOPCovid 1 and 2, according to two different patient samples. It is not difficult to find that the detection sensitivity of STOPCovid 1 is equivalent to the RT-qPCR CT value of 30. The RT-qPCR CT value that STOP Covid2 can achieve is 40. So based on the input statistics of Covid-19 patients, it can be found that if the CT value reaches 40, it proves that it can successfully detect almost 100% of patients, and if the CT value is 30, it proves that only 60% of patients can be successfully captured. So it is very important to increase the CT value to around 40-50. It is worth noting that patients with CT value reaching 40 may have a weaker infectivity because the virus’s ability to spread to the environment is significantly reduced.
Then, our collaborators launched a systematic test, and they tested more than 400 samples, including 202 positive samples and 200 negative samples. The results showed that STOPCovid 2 was able to obtain a sensitivity of 93.1% and a specificity of 98.5%, and the detection results were almost consistent with the sensitivity and specificity of RT-qPCR detection.
This is the test result of 200 of the samples. It can be seen that some false positives have occurred at the remote end.
Then, we broke up the patient and tested again. The left side shows that in the negative test, the patients cannot detect the virus, and among the positive patients on the right side, positive signals can be detected. The CT range detected ranges between 20-35.7.
At present, we are developing and designing a small detection instrument that can be used to achieve detection without pipetting and other equipment. The way we designed the work is that the instrument will contain an enrichment tube, and the operator can insert a cotton swab dipped in saliva directly into the enrichment tube that dissolves the virus, and after extraction, the virus will be dissolved in the extraction buffer, and finally inserted into the device and run it for 30 minutes to complete the test. I hope this helps people detect COVID-19 in a very simple way, especially in nursing facilities and homes.
My sharing ends here, we will continue to work hard. I would like to thank my lab members and collaborators, as well as the funding agencies that support this work, thank you.
In the subsequent question session, Zhang Feng also answered some questions from the audience, and the content was sorted out as follows:
asked: How big is the viral RNA fragment detected?
Zhang Feng: Guide-RNA in STOP Covid is a 24-base paired sequence that we use to identify some viruses, but it is just one of the primers. We actually used six different primers to expand the detection fragment, which contains 2 rounds of detection, which is why STOP Covid can get better sensitivity.
Asked : How much is the estimated cost of testing?
Zhang Feng: The cost of detection depends on the scale of the test. Enzymes, Guide-RNA, primers, etc. all need to be produced. The price of each test may be around one dollar, or slightly more than one dollar. If large-scale manufacturing can be achieved, the price of test may be cheaper.
asked: What is the working principle of glycine and taurine?
Zhang Feng: We are not very clear about how glycine and taurine work at this time, they may play a role in maintaining protein stability to ensure they are more stable in the reaction. The additives we tested are commercially available, with a total of 96 types, including lyophilized reagents, biochemical reagents that optimize specific reactions, and even crystals.
Question: There is a Swiss research team using nano-gold particle biosensors for monitoring the content of the new coronavirus in the environment. What are the necessary steps to complete before achieving this?
Zhang Feng: All these environmental monitoring projects can be accomplished using molecular detection methods, whether it is PCR, isothermal amplification or even CRISPR detection based on isothermal amplification. The initial step is to collect samples from the environment and concentrate them. There are many devices that can implement this process. For example, the RNA extraction column can concentrate larger inputs into smaller volumes. If these methods can increase the sensitivity of the test, we can all try it.
Of course, researchers have used this method to test wastewater, and a professor at MIT has been using PCR assays to find the coronavirus in wastewater samples. This study is also published online on the Massachusetts Sewage Treatment Bureau website, you can check it out. And as Covid-19 spreads in the United States can see an increase and decrease in the amount of virus in sewage, this data can be used to monitor the number of patients with positive cases, which are very useful applications.
asked: COVID-19 is not the last infectious disease we encountered. Is this testing effective for other viruses we encounter in daily life?
Zhang Feng: The reaction in the detection is a general molecular amplification reaction process, so assuming that the virus's DNA or RNA can be released from the virus particles, it should be possible to read it. But one challenge is that the virus or sample you collect from a patient may contain the inhibitor , which will prevent amplification or prevent detection of fragile viruses. Therefore, there are different ways to degrade inhibitors. The first type is to perform replication detection after adding protease to lyse the protein. The second type is that small molecule compounds may exist during the detection process, so you may need to use filters and other equipment to reduce the content of inhibitors.
asked: Is the detected positive signal valuable for determining the virus content? Is it in linear dynamic range?
Zhang Feng: There is this possibility. We have found that combining LAMP and CRISPR reactions results in a binary-like signal—negative or positive. However, when the viral load becomes larger, the signal with high viral load will appear slowly, and the signal with low viral load will appear longer. I think it is possible to generate a standard curve with different input concentrations as variables. We are figuring out the rise time of the detection signal, and maybe we can use it to infer the viral load.
At present, we do not have enough patient sample data to confirm this situation. The most important variable is the number of inhibitors that may be contained in the patient sample, so I think a standard curve can be established for purified RNA samples, but for the patient sample, we need to do more work.
asked: Can multiple viruses be detected at one time?
Zhang Feng: This is possible. Multiple CRISPR proteins can be used, multiplexed RNA or DNA. For example, one can be used for influenza, one for Covid-19, one for respiratory fusion virus RSV, etc. Each virus will emit a different color as a signal, using this to achieve multiplexing and detect multiple viruses.
asked: What are the main functional differences compared with CRISPR Cas12a detection?
Zhangfeng: In the reaction we used Cas12b as the CRISPR protein because some Cas12b comes from thermophilic bacteria, which allows it to work at 60°C. Cas12a can work at 37°C, but not at 60°C. This is also the reason why we use Cas12b.
