The transformation of gene Tech in antibody discovery

Original brick maker Previously, brick said that there are two main routes for antibody discovery: in vivo and in vitro However, after more than 20 years of development, in vitro antibody discovery based on in vitro display technology and synthetic antibody library collided with the drug resistance wall In the past seven or eight years, the whole field has been reflecting on where the problems are and how to seek breakthroughs? At a time when there are more and more questions about the drug-resistance of all synthetic antibodies, the industry leaders are also seizing the time to lay out If you are interested, you will notice that in the past four or five years, gene Tektronix has turned to antibody discovery in vivo based on single cell technology KD wittrup's adimab company (Xinda's technology platform was introduced from adimab company) chose to continue to optimize the total synthetic antibody library (Xu et al, 2013 peds; us20190203305a1; Kelly RL et al, 2018 JMB) and to solve the early prediction of drug-forming (Jain t et al, 2017 PNAs; Lu x et al, 2019 mAbs) Two flowers, one for each Today, we only talk about the transformation of gene Tech in antibody discovery If you know the history of Genentech, you will know that from about 2000 to 2010, Genentech invested a lot in the in vitro display technology platform, and its representative was Sachdev Sidhu The investment in phage display and total synthetic antibody library ended with devsidi's resignation in 2008 Coincidentally, Paul Carter, another legend of Genentech, left his job after joining Genentech's antibody engineering department in devsido, and then returned to Genentech to take charge of antibody engineering department in 2010 What happened to this? Let's look at an article (hotzel I et al, 2013 mAbs) by Paul Carter and Robert Kelley Paul and Bob are both early employees of Genentech The difference is that Paul chose to go out in exile for ten years while Bob has been in Genentech's antibody engineering department for the past three decades When Paul Carter returned to Genentech to take charge of the antibody engineering department, let's see what the first thing these two old guys did There is only one problem to be solved in this article, that is, is the antibody displayed by phage worse than that of humanized antibody in pharmacodynamics? I think after Paul Carter's return, he could not sleep at night in the face of the mess that devsido had left behind On the one hand, we need to solve the problem of the efficiency of new drug research and development, on the other hand, we need to straighten out the technical route In the 1990s, Genentech's antibody engineering was very effective, and it listed antibody drugs such as rituxan and Herceptin However, since 2000, a large amount of high-throughput screening based on phage display technology has been invested, and the efficiency of new drug production has been greatly reduced Where to go (how to turn the tide)? It's the first problem Paul Carter faced after his return With the background of this article clear, let's take a look at a data Fig 1 retrospective study on the clearance rate of 52 antibody drugs and antibodies in development in cynomolgus monkeys (hotzel I et al, 2013 mAbs) The Y-axis of this figure shows the clearance rate of McAb in cynomolgus monkeys (ml / day / kg) The higher the value, the worse the pharmacokinetics X-axis is a list of 52 antibody drugs, including bevacizumab, rituximab, trastuzumab, omazumab and patozumab From the figure, we can see that when the clearance rate is less than 10 ml / day / kg, the pharmacokinetics is very good Therefore, the clearance rate of 10ml / day / kg (blue dotted line in the figure) in cynomolgus monkeys can be used as a boundary value of good and bad pharmacokinetics The median clearance rate of phage displayed antibody in cynomolgus monkeys was 9.0ml/day/kg, and the median clearance rate of humanized antibody in cynomolgus monkeys was 6.5ml/day/kg But what the author didn't tell you is whether there is a statistically significant difference between the phage displayed antibody and humanized antibody if compared with the whole human antibody alone? This comparison has not been made It is likely that the data of whole person antibody is relatively small Therefore, the comparison in statistics is not rigorous scientifically However, from the limited data of all human antibodies, the median value of the clearance rate of all human antibodies in cynomolgus monkeys is about 3.1 ml / day / kg You can make a comparative judgment by yourself (Fig 2a) Although Paul Carter and Bob Kelly didn't specifically compare the pharmacokinetic differences between all human antibodies and other antibodies in this article, I think they were shocked by this discovery Because the layout of gene Tech in antibody discovery technology after 2013 fully illustrates this point From 2013, Genentech began to abandon the discovery of full synthetic antibodies based on phage display, and invested heavily in in vivo antibody discovery based on single cell antibody discovery technology If we only talk about the influence factors, this article has little influence in the academic circle However, the antibody engineering / discovery of Genentech is a link between the past and the future, because it sums up the success or failure of the past two decades and guides the direction for the next two decades Let's see if the next five years will see a significant improvement in the yield of new drugs Fig 2 Effects of antibody humanization and affinity engineering modification on pharmacokinetics (2B selected from hotzel I et al, 2013 mAbs) As for the above data of clearance rate in cynomolgus monkeys, some students will surely say that pharmacokinetics is a very complex thing, involving not only antibodies, but also many factors of the body / patient itself From the perspective of pure pharmacokinetics, your explanation above is not rigorous, because there are many variables, such as the distribution of FcRn and target protein in vivo (Ryman JT et al, 2017 syst Pharmacolo.), which you have ignored This criticism is right Brick family and Paul Carter have a bowl of rice, so they agree with his idea of antibody discovery, which is to solve the problem of drug resistance in the early stage of antibody discovery Under this concept, this paper focuses more on the effect of variable regions of different antibodies on pharmacokinetics Because this is an industrial article and a retrospective study, it is certainly not rigorous in some aspects, but it is tall enough for business decision-making This article also has a very interesting data, which has a great impact on brickers Paul Carter and Bob Kelly chose an antibody (antibody 47) to explain the effect of antibody engineering on pharmacokinetics in vivo It is easier to make a conclusion by using only one antibody and limiting the variable to the sequence of antibody variable region Antibody 47 is an antibody against human fibroblast growth factor discovered by Genentech in the early stage This antibody was obtained by mouse immunization In the early stage of research and development, Genentech's antibody engineering team was tortured (bumbaca D et al, 2011 mAbs), not to mention here As shown in Fig 2B, only one potential glycosylation site was removed from antibody 47c relative to the original antibody sequence, and the clearance rate in cynomolgus monkeys was 4.2 ml / day / kg (n = 4) The affinity of antibody 47b was optimized on the basis of 47c The clearance rates of antibody 47b in cynomolgus monkeys were 9.3 ml / day / kg and 8.2 ml / day / kg, respectively (n = 2) The affinity and chemical stability of antibody 47 were optimized on the basis of 47c The clearance rate of antibody 47 in cynomolgus monkeys was 20.4ml/day/kg (n = 4) This data fully illustrates the problem that the pharmacokinetics of antibodies found in the body's immune system is the best In vitro engineering optimization is not only of little help to the pharmacokinetics of the body, but also worse (sorry, old fellow iron workers who have been working on the antibody, feel shame and then brave.) Let's talk about it today Let's finish class early I think the above two data tell you: (1) The pharmacokinetics of antibodies found in vivo is incomparable After positive and negative selection of the immune system, the specificity of antigen recognition and the physical and chemical characteristics of the antibody are the best In contrast, the antibodies found by in vitro synthesis of antibody library will have more or less multi-specific problem of antigen recognition, and the physical and chemical characteristics of the antibodies are not satisfactory.