A Brief History
From a scroll of Egyptian remedies to molecules that destroy a protein instead of blocking it. 27 moments, in order, each one with what actually changed because of it. Scroll down and the ideas stack up: write it down, test it by rule, purify it, make it, aim it, design it, and finally build the instruction instead of the drug.
From the earliest herbals to the physicians who argued that illness has natural causes and can be treated by rule.
c. 1550 BCE
Egypt
A scroll about twenty metres long listing some seven hundred remedies and the complaints they were for, including willow for pain and fever, opium, castor oil and honey on wounds. It is a catalogue, not an explanation: there is no theory of why any of it should work.
Writing a remedy down is where the whole field starts. Once a recipe survives its author, it can be copied, tested by someone else, and corrected.
c. 400 BCE
Kos, Greece
The Hippocratic writers argued that disease has natural causes rather than divine ones, and should be treated by observation, diet and remedies. Among those remedies was the bark and leaf of the willow, chewed for pain and for the fevers of childbirth.
Two ideas enter medicine here: that illness is a natural process, and that what you do about it should be judged by what you observe. Both are preconditions for everything on this page.
c. 600 BCE to 200 CE
India
The two great compendia of Ayurveda, compiled over centuries. Between them they describe several hundred medicinal plants and minerals, and they go beyond a list: there are preparations (decoctions, powders, oils, fermented liquids), rules for which form suits which patient, and in the Sushruta Samhita a detailed surgical practice.
This is formulation, long before the word existed. The same plant was understood to act differently depending on how it was prepared and given, which is the question every pharmaceutical scientist still works on.
c. 200 CE, from older tradition
China, attributed to the mythical Shennong
The classic Chinese materia medica: 365 entries, one for each day of the year, sorted into three grades by how safe they are to take for a long time. It names mahuang (ephedra) for wheezing, changshan for fevers and dahuang as a purgative, and it is still the root of Chinese herbal practice. A century and a half later Ge Hong added a line about qinghao, sweet wormwood, for the fevers of malaria: soak a handful in water, wring it out, and drink the juice.
Grading substances by how safely they can be taken over time is a therapeutic index in everything but name. And Ge Hong's instruction to use cold water rather than a hot decoction turned out to matter a great deal, sixteen centuries later.
c. 65 CE
Pedanius Dioscorides, a Greek physician in the Roman army
Five books covering around six hundred plants and a thousand drugs, each with what it looks like, where it grows, how to prepare it and what it treats. It was copied, translated into Arabic and Latin, and used as the standard reference for roughly sixteen hundred years.
The first work that is recognisably a pharmacopoeia. It set the expectation that a drug should come with a written identity: this substance, from this source, prepared this way.
Scholars who turned a list of remedies into a method: rules for trying a drug, and the beginnings of dose.
c. 850 to 925 CE
Rhazes and al-Kindi, Baghdad
Al-Razi separated smallpox from measles by careful clinical description, kept case notes, and brought distillation and other chemical operations into the preparation of medicines. Al-Kindi, a generation earlier, tried something no one had: putting a number on the strength of a compound medicine, a first attempt at quantifying dose.
Writing down what you saw in each patient, and attaching a quantity to a drug, are the seeds of the case report and of dose-response.
1025 CE
Ibn Sina (Avicenna), Persia
Five volumes, one of them a materia medica of some eight hundred substances. What sets it apart is a short passage setting out rules for deciding whether a drug works: test it on a simple, uncomplicated disease; use two contrasting cases; watch the time to effect and whether it repeats; and be careful that what you saw was the drug and not the illness running its course. The Canon was taught in European universities into the seventeenth century.
This is the clinical trial in outline, six centuries before anyone ran one. Nearly every rule in it, controls and reproducibility above all, is still in the protocol of a modern study.
c. 1530
Paracelsus, Switzerland
A difficult, combative physician who burned the textbooks he disagreed with and insisted that medicines should be specific chemical substances rather than complicated herbal mixtures. He is remembered for one sentence: every substance is a poison, and only the dose separates a poison from a remedy.
Toxicology begins in that sentence, and so does the central trade-off of drug discovery. A molecule is not safe or unsafe; it has a window between the dose that helps and the dose that harms, and the job is to widen it.
The century in which a remedy stopped being a plant and became a single pure compound that could be weighed, and then made.
1785
William Withering, Birmingham
Withering heard of a Shropshire herbal remedy for dropsy, worked out that the active ingredient was foxglove, and then spent ten years giving it to patients, recording who improved and who was poisoned, and adjusting the preparation and the dose. He published the failures alongside the successes.
The first systematic dose-finding study in print. Digitalis has a famously narrow margin between the helpful dose and the fatal one, which is exactly why careful records were the only way through.
1796
Edward Jenner, Berkeley, Gloucestershire
Dairymaids who caught cowpox did not get smallpox. Jenner put matter from a sore on Sarah Nelmes's hand into the arm of eight-year-old James Phipps, then deliberately exposed the boy to smallpox; nothing happened. Inoculation with smallpox itself was already old, practised for centuries in China, India and the Ottoman lands and brought to England by Mary Wortley Montagu, but it killed one or two patients in a hundred. Using the milder cousin of the disease was the change that made it safe enough for everyone.
A medicine that prevents instead of treating, and the first use of one disease to defend against another. Smallpox is the only human disease ever driven to extinction, and this is where that ends.
1805
Friedrich Sertürner, a young pharmacist in Paderborn
Sertürner isolated a crystalline base from opium, tested it on himself and three friends at a dose that nearly killed them, and named it after Morpheus. It was the first alkaloid pulled out of a plant, and it meant that for the first time a doctor could give a known amount of the thing that actually acted.
A plant extract varies with the soil and the season; a pure compound does not. Everything about standard doses, strengths on a label and batch-to-batch consistency follows from this one separation.
1828
Friedrich Wöhler, Berlin
Heating an inorganic salt, ammonium cyanate, gave Wöhler urea, a substance until then only known from living things. He wrote to his teacher that he could make urea without needing a kidney.
The belief that the molecules of life need a living thing to make them does not survive this experiment. Once chemists accept that, synthesising a medicine becomes thinkable.
1897
Felix Hoffmann and Arthur Eichengrün at Bayer, Germany
Willow bark contains salicin, which the body turns into salicylic acid: it works, and it wrecks the stomach. Bayer acetylated it. Acetylsalicylic acid had been made before, by Charles Gerhardt in 1853, but Bayer made it pure, made it at scale and sold it as a named product from 1899. Who deserves the credit inside Bayer is still argued over.
The first blockbuster, and the first clear case of changing a natural molecule to fix its side effects while keeping its action. That is medicinal chemistry in one step.
The idea that a molecule can be aimed at a disease, and the first medicines that made good on it.
1909
Paul Ehrlich and Sahachiro Hata, Frankfurt
Ehrlich reasoned that if a dye can stain one kind of cell and not another, a molecule could in principle poison a parasite and spare its host. His laboratory made and tested hundreds of arsenic compounds against syphilis; Hata found that number 606 cured infected rabbits. It was the first effective treatment for the disease, and it was toxic and difficult to give.
Two foundations at once: the magic bullet, meaning selective toxicity, and the screening campaign, meaning make many analogues and test them all. Both still describe how most drugs are found.
1921 to 1922
Banting, Best, Macleod and Collip, Toronto
Extracts of dog pancreas lowered blood sugar; Collip worked out how to purify the extract well enough to give to a person. In January 1922 fourteen-year-old Leonard Thompson, dying of type 1 diabetes, was injected. The patent was sold to the University of Toronto for one dollar each so that no one could control the supply.
The first medicine that was a protein rather than a small molecule, and the first time a fatal disease became a managed one. Every biologic since, and the whole business of making medicines in living cells, starts here.
1885, and 1924 in the clinic
Nagai Nagayoshi, Tokyo; Chen Ko Kuei and Carl Schmidt, Peking Union Medical College
Nagai isolated ephedrine from mahuang, the ephedra the Shennong Bencao Jing had listed for wheezing, and almost nothing happened for forty years. Then Chen, who had been told by his uncle, a traditional physician, to look at mahuang, worked out with Schmidt what the pure compound does to blood pressure and airways. Within a few years it was a standard asthma medicine worldwide.
A written herbal tradition can be a screening library if someone bothers to read it. Ephedrine also became the chemical template for a whole family of later drugs, from amphetamines to the inhalers used today.
1932 to 1935
Gerhard Domagk at Bayer; Tréfouël and colleagues at the Pasteur Institute
Domagk found that a red dye cured streptococcal infections in mice, and gave it to his own daughter when she developed a serious infection. The Pasteur group then showed the dye itself does nothing: the body cuts it in half, and the active half is colourless sulfanilamide, a compound already out of patent.
The first broadly useful antibacterial, and the first prodrug: a molecule that is not the drug until the body changes it. It is also a reminder that the thing you tested may not be the thing that acted.
1928, and then 1941
Alexander Fleming, London; Florey, Chain, Heatley and the Oxford team
Fleming noticed in 1928 that a mould had cleared the bacteria around it, and could not purify enough of the substance to do anything with it. Thirteen years later the Oxford group did: Heatley built the extraction, and in 1941 they treated a policeman with blood poisoning, who improved and then died when the supply ran out. Deep-tank fermentation in the United States turned it into a product in time for the war, and in 1945 Dorothy Hodgkin solved the structure by X-ray crystallography.
The gap between the observation and the medicine is thirteen years of chemical engineering. Discovery is rarely the hard part; making enough of it, pure, is.
Chemists stop screening blindly and start building molecules from what is known about the target.
1948 onwards
Gertrude Elion and George Hitchings, Burroughs Wellcome, New York
Instead of screening whatever was on the shelf, Elion and Hitchings looked at what a cell needs to build DNA and made molecules that look almost but not quite like those building blocks. From that one idea came 6-mercaptopurine for leukaemia, azathioprine for transplants, allopurinol for gout, and later acyclovir for herpes. Elion had no doctorate and had been turned away from laboratory jobs as a woman; she shared the 1988 Nobel Prize.
The first sustained programme of designing drugs from knowledge rather than finding them by luck. The antimetabolite idea, a decoy that jams a pathway, is still behind a large share of cancer and antiviral medicines.
1964 and 1976
James Black, ICI and then Smith Kline and French, Britain
Black started from a receptor rather than a disease. If adrenaline acts on the heart through a receptor, a molecule that occupies that receptor without switching it on should quieten the heart: propranolol, 1964, and angina and high blood pressure became treatable. He then did the same thing again for the histamine receptor in the stomach, which gave cimetidine in 1976 and took surgery for ulcers off the table.
Proof that you can reason from a receptor to a medicine, and do it to order. This is the template for the target-based drug discovery that most of the industry runs on.
1972, in the open from 1979
Tu Youyou and Project 523, Beijing
Asked during the Vietnam war to find a malaria drug that still worked, Tu's team went through some two thousand classical Chinese recipes and tested hundreds of extracts. Sweet wormwood looked promising and then failed to repeat, until Tu went back to Ge Hong's wording: wring out the juice, do not boil it. Boiling was destroying the active compound. Extracting at low temperature with ether gave a clean result, and she took the first dose herself. Artemisinin and its derivatives are now the backbone of malaria treatment, and she received the Nobel Prize in 2015.
The molecule carries a peroxide bridge no chemist would have designed and no screen of the day would have kept. It is the strongest argument there is for reading the old literature carefully, and for paying attention to how an extract is made.
1975, reaching patients from 1986
Georges Köhler and César Milstein, Cambridge
Fusing an antibody-producing cell with a cancer cell gave a line that divides forever and makes one antibody, against one target, without end. The first therapeutic antibody was approved in 1986; the ones that changed cancer treatment, rituximab and trastuzumab, arrived in 1997 and 1998.
A way to make a binder for almost any target, with a selectivity no small molecule can match. It is the beginning of the split between the chemistry side of the industry and the biology side.
2001
Nicholas Lydon, Brian Druker, Alex Matter and colleagues; Ciba-Geigy, then Novartis
Chronic myeloid leukaemia is driven by one fused protein, BCR-ABL, made by one chromosomal accident. A molecule built to block exactly that protein turned a disease that killed most patients within five years into one that most people live with. The chemistry and the crystal structures went hand in hand.
The clearest demonstration that if you know the precise molecular cause, you can aim at it and the side effects stay small. It set off the kinase inhibitor era, and the expectation that cancer should be treated by mechanism rather than by organ.
Proteins, antibodies, genetic messages, and molecules that destroy a target instead of blocking it.
2000, and in earnest from 2013
many groups; the modern wave from Genentech, Daiichi Sankyo, Seagen and others
Tie a very toxic small molecule to an antibody with a linker that only lets go inside the target cell, and you can use a payload far too poisonous to give on its own. The first approval came in 2000 and was later withdrawn and reinstated; trastuzumab emtansine in 2013 and trastuzumab deruxtecan in 2019 made the class work.
Ehrlich's magic bullet, built literally: the antibody is the aim and the small molecule is the warhead. Most of the difficulty turned out to be in the linker, which is chemistry, not biology.
2016 and 2020
Ionis and Biogen (nusinersen); Roche, PTC Therapeutics and the SMA Foundation (risdiplam)
Spinal muscular atrophy is caused by a missing protein, but the body carries a near-identical backup gene whose message leaves out one essential exon. Nusinersen, an antisense oligonucleotide given into the spine, makes the cell keep that exon. Risdiplam, approved in 2020, does the same thing as a tablet: a small molecule that changes how a gene is spliced.
The target stops being a protein and becomes the instruction for a protein. And a small molecule turns out to be able to do a job that was assumed to need a nucleic acid drug.
from 2019 in patients
Craig Crews and Ray Deshaies for the idea; Arvinas, Kymera, C4 and others in the clinic
A PROTAC is one molecule with two ends: one holds the protein you want rid of, the other grabs the cell's own disposal machinery, and holding them together gets the protein destroyed. Because the molecule is released afterwards and does it again, a small amount goes a long way. The first ones entered human trials in 2019.
A different question from the one the field has asked since Ehrlich. Not "can I block this protein?" but "can I remove it?", which opens up targets with no pocket to block, long written off as undruggable.
2020
Katalin Karikó and Drew Weissman; BioNTech, Pfizer and Moderna
Karikó and Weissman showed in 2005 that swapping one of the building blocks of RNA stops the body attacking it as an intruder, which was the obstacle that had kept mRNA out of the clinic. With a lipid nanoparticle to carry it into cells, the result was two vaccines designed within days of the virus sequence being published and given to billions of people within a year.
The medicine becomes an instruction rather than a molecule that acts, and the delivery vehicle becomes as important as the active. It also showed what the development timeline can look like when there is no shortage of money or urgency.
The ancient dates are the ranges historians give, not exact years, and where the credit for a discovery is argued over the card says so. If you think something here is wrong or missing, tell us and we will fix it.
