GEN BiotechnologyVol. 2, No. 3 Asked & AnsweredFree AccessAntibody There? Mapping the Human Proteome with Mathias UhlénMathias Uhlén and Fay LinMathias Uhlén*Address correspondence to: Mathias Uhlén, Department of Protein Science, Science for Life Laboratory, KTH-Royal Institute of Technology, SE-10044, Stockholm, Sweden, E-mail Address: mathias.uhlen@scilifelab.seDepartment of Protein Science, Science for Life Laboratory, KTH-Royal Institute of Technology, and Department of Neuroscience, Karolinska institutet, Stockholm, Sweden.Search for more papers by this author and Fay LinSenior Editor, GEN Biotechnology.Search for more papers by this authorPublished Online:19 Jun 2023https://doi.org/10.1089/genbio.2023.29099.muhAboutSectionsPDF/EPUB Permissions & CitationsPermissionsDownload CitationsTrack CitationsAdd to favorites Back To Publication ShareShare onFacebookTwitterLinked InRedditEmail Mathias Uhlén, Professor of Microbiology at KTH Royal Institute of Technology.Mathias Uhlén became professor of microbiology at KTH Royal Institute of Technology in 1988. His group has become one of the leaders in the field of proteomics, including affinity-based protein engineering, next-generation sequencing by synthesis, and mapping the human proteome in a project known as the Human Protein Atlas (Fig. 1).1 Remarkably, his research has resulted in >750 publications and the creation of 20 biotech companies.FIG. 1. Since the launch of the Human Protein Atlas in 2003, the consortium has achieved key milestones over 20 years. More details are outlined in the Atlas portal.In this interview, conducted by GEN Biotechnology Senior Editor Fay Lin, Uhlén describes his professional journey into proteomics, goes behind the scenes of the Human Protein Atlas, and offers insights for academics who strive to enter the biotech industry.(This interview has been edited for length and clarity).GEN Biotechnology: Mathias, what were some of the pivotal moments early in your career and what led you to where you are today?Uhlén: I started as a chemical engineer. My PhD focused on genetic engineering back in the 1980s. It was actually a collaboration with Genentech and a fantastic scientist, Jim Wells. In the early days, we were developing protein production for human growth factors. I stayed in the protein field and moved toward protein engineering. My goal was to engineer proteins to be perfect biotechnology molecules. At that time, we were trying to improve the methods for purifying therapeutic antibodies and developed what would later become the MabSelect SuRe sold today by Cytiva (Sweden)—most of the antibodies in the clinic today are purified using this ligand, and it is fantastic to think that our group has made a contribution to saving thousands of patients lives.In the 1990s, I became quite involved in the Human Genome Project (HGP) through Craig Venter. We actually organized some conferences together! I was working on the technology development side. In the early 1990s, we described in a few articles a new concept called real-time sequencing-by-synthesis, which later became pyrosequencing. This ended up becoming massive parallel sequencing or next-generation sequencing (NGS). When the HGP was almost completed in 2001, we realized that it would be nice to look back at the proteins. That is how I transitioned into focusing on the proteome.I recently attended a fireside chat with Craig Venter in which he said, “without measuring the comprehensive phenotype, the genome was virtually worthless!” Along that vein, there are so many omics these days, whether it be transcriptomics, metabolomics, and proteomics. How would you define the value of proteomics?The proteome is ∼20,000 protein-coding genes that make us get up in the morning, drink coffee, and so on. Proteins are the molecules of life and also the targets of drugs. The HGP gave us the blueprint, which is nice when you want to build a house, but to actually complete a house, you have to furnish it and put all the bricks in place. That is proteins!Proteins are a much harder molecule to study technically. Mass spectrometry is one method but [my group] chose another method, which is the use of antibodies. Most people think that antibodies are a drug, but for us they are reagents to study target proteins. We said back in the early 2000s, “let's just make antibodies to all human proteins.” So we did! We produced 60,000 antibodies in a factory-style manner. We then used these antibodies to look into cells, organs, and diseases to understand what the proteins are doing.With these antibodies, we also produced ∼15 million bioimages and created 5 million webpages, thereby providing a resource for researchers who want to know everything about their favorite protein and drug (Fig. 2). This was the background for the Human Protein Atlas. We were very pleased that our funders, the Knut and Alice Wallenberg Foundation, decided that this project was essential for human disease and biology and [motivated the project's] transition to open access with no restrictions. We have several million visits every year from pharmaceutical companies and universities.FIG. 2. The Human Protein Atlas contains more than 10 million images generated by antibody-based high-resolution microscopy.These images are created using both classic immunopathology and fluorescence-based technology.The Human Protein Atlas project started in 2003. You mentioned a handful of events, such as the HGP, that pushed the science forward. Can you elaborate on the launch of the project?As mentioned, I was involved with the HGP with Craig Venter. We also collaborated with a fantastic scientist at Baylor College of Medicine, Richard Gibbs. We started a pilot project to produce the proteins discovered through the HGP, which led to us producing antibodies to use as hooks to go into tissues and [investigate spatial distribution]. We presented this study at a conference in San Diego in 2001. A person from the Wellcome Trust approached me after my talk and said, “you have to come to the UK and run this project!”We ended up putting together a consortium. I worked with a fantastic scientist at University of Oxford, John Bell, to start the Human Protein Atlas in Oxford. At the last minute, the Wallenberg Foundation stepped in and said, “let's do this in Sweden instead.” That is when the project officially started in 2003 in Sweden. The Wallenberg Foundation has now supported us for 21 years and they just gave us funding for another 7 years. We are very pleased to receive ∼30 years of funding, which is of course amazing. With the new funding, we are now moving into the next decade of the protein atlas.What are the goals for the next 10 years of the Human Protein Atlas?During the first 20 years, we have seen the rise of new technologies, such as transcriptomics and single-cell genomics. We use these technologies to complement the Human Protein Atlas. We also have this fantastic effort with artificial intelligence and deep learning called Alpha Fold. With all the proteins that are being studied, we also investigate the three-dimensional structure. [It has been a huge movement] for many people in the community and has been a fantastic breakthrough.Our focus right now is on blood analysis and disease characterizations, moving into what some people call precision medicine. Blood is a fantastic mural of what is going on in the human body when it comes to diseases. By taking these tools, we hope to develop next-generation diagnostics. This is a huge effort involving many clinicians to get human samples. We also work with ethics committees because we want the data to be open access and protect the individuals. Disease and cancer diagnostics are big areas of focus for the protein atlas.In genomics, we have seen the cost of sequencing the whole genome drop to ∼$200. How would you compare the pace of advancement for the field of proteomics?Until very recently, I would describe the pace as a bit slower than genomics. Mass spectrometry, which is the main tool for analyzing proteomics, is getting better and better, but it is still a cumbersome technology compared with genomics. When we started on the antibody side, there were limited antibodies for human proteins. Now we have a portal called Antibodypedia with 5 million antibodies for human proteins. This is an increase from ∼100,000 compared with 10 years ago. Companies such as Olink from Sweden and SomaLogic in the United States are pushing to get these tools to study proteins. These two companies are moving in a similar way as Illumina has done on the DNA sequencing side.In California, we recently had the Theranos case, where the company founder, Elizabeth Holmes, was convicted and is facing prison. She claimed she could do about a hundred protein assays from a small drop of blood. This was ∼10 years ago and, of course, not true. No one could do it! Today, we are analyzing 3,000 proteins from just a fraction of a blood drop. The pace is quite unbelievable.With this new technology, we decided to follow 100 individuals, 50 males and 50 females, initially for 2 years (since extended to 4 years). We take samples every 3 months and throw every analysis on these people. We then perform longitudinal studies to analyze what happens when they get sick and compare the genome with the proteome. We have published several articles on that front.You are one of the founding directors of SciLifeLab. What is that organization? Are there huge tie-ins between SciLifeLab and the Human Protein Atlas?Around 2008–2009, we had a lot of discussions with the Swedish government and the ministry of research. My colleagues and I had the vision that the biological field is moving into big data and technology-driven infrastructures, which are expensive. We convinced the government to create a budget for a national infrastructure. I was employee number one and the founding director of [SciLifeLab]. More than 10 years later, we have ∼1,500 people at the Stockholm site and another 500 in Uppsala.The focus is data-driven life science using machine learning, deep learning, and artificial intelligence. [We want to] generate data and analyze it. It is very exciting but also very demanding because you have to be very multidisciplinary. In one institute, we have all the multiomics, such as metabolomics, genomics, proteomics, and more.Are there any specific projects at the top of your head right now?One of our current preprints looked at 12 of the most frequent cancers, such as lung, prostate, colorectal, and breast cancer.2 We took small blood samples and used proximity extension assays, then we asked a machine learning algorithm, “what is the best predictors of [these cancers]?” We produced ∼2 million data points and condensed it to something that is feasible in the diagnostic setting (Fig. 3). Another collaboration, with the company Capitainer, focuses on home tests, where you take a drop of blood from your finger at home and send it to the laboratory by mail for analysis for different diseases. This is very exciting, although it will be years before these tests become routine.FIG. 3. Toward next-generation cancer prediction medicine.The example shows elevated levels of the protein GFAP in the blood of patients with brain tumors.As you mentioned, there are many advances in machine learning and precision medicine. What are some of the key bottlenecks preventing this technology from reaching its full potential?When talking about cancer screening and diagnostics, a big issue is false positives. There is natural variation among blood fingerprints that leads some people to be scored as positive when they are completely healthy. This is very dangerous because it causes a lot of grief for patients and costs for society. It is important to have logistics where you do screening but then you confirm in another setting whether you are a true positive. We are addressing this issue during screenings.You have been highly involved in the biotech industry and have founded 20 companies. Do you have advice for academics interested in entering the industry?We are a bit spoiled here in Sweden because we have something called “Teacher's Exemption,” which means that all inventions, innovations, patents, and intellectual property rights, belong to the [individual] researcher. That means that we can develop an idea in the academic setting for quite a while before we take it into the harsh environment outside the university.There are several things I have learned after working with 20 start-ups. First, there is no formula. Every journey is different. There are always new situations as you move into different stages where growth was once important and now profit becomes more important. Second, in life sciences and probably in general technology as well, you have to be number one in your niche to make money. If it is a small niche, you make small money but you still make money! If it is a big niche, you make a lot of money. It could be a geographical niche, but usually in my field, it is a technological niche where you have to strive to be number one. This is extremely challenging, as there are so many talented and hardworking researchers around the globe.It is also important to not start companies too early. When you are taking in venture capital, you have to be sure what you want to deliver with that funding and bring value for the people who have invested in the company.Are there any particular start-ups that you would like to highlight?I am very fond of Affibody Medical, which focuses on an alternative to antibodies. They recently made a very nice deal with Aceleryn in the United States. They are currently ∼100 people and are now moving into Phase III for psoriasis with fantastic results in the clinic. There is also a 20-year-old company called Pyrosequencing, which was based on the NGS techniques developed in our group. That company has since changed its name to Biotage, which is now a unicorn company moving into protein and small molecule purification.I am also working with a company in South Korea called Abclon. This company has spectacular results on therapeutic antibodies for stomach cancer in a clinical trial in China and results in CAR-T therapy. Abclon is on the South Korean stock market (KOSDAQ) and is doing really well.How would you sum up the life science field in the past 20 years?It is a really exciting time for life science and biotechnology. We have had incredible development of technologies ranging from sequencing-by-synthesis to NGS. The HGP published in 2001 cost $2–3 billion. Now you can sequence the genome for $200. That is a factor of a million cheaper! Proteomics is also moving in that direction. When things are getting cheaper and more multiplexed, we are producing a lot of data, which makes machine learning and deep learning very attractive.I think the next 10 years in life science will provide the most exploding knowledge base in history, which will hopefully translate into new drugs and other applications. I wish I was 30 years younger to fully take in the advances that are to come!