Antibodies, vaccination and the four kinds of immunity

What an antibody is made of

An antibody is a protein, an immunoglobulin, with a quaternary structure of four polypeptide chains: two identical long heavy chains and two identical short light chains, held together by disulfide bridges into a Y.

Describe the drawing

A Y-shaped antibody drawn with two long heavy chains running from the arm tips down through the stem and two short light chains lying along the outside of each arm. Labelled features include the two identical antigen-binding sites at the tips of the arms, the variable regions at the top of each arm, the constant regions below, three disulfide bridges, and the flexible hinge where the arms meet the stem.

The tips of the two arms are the variable regions. Their amino acid sequence differs between one antibody and the next, so the tertiary structure there differs, so the shape of the binding site differs. Each arm carries one antigen-binding site, formed where a heavy chain and a light chain meet, and the two sites on any one antibody are identical to each other. Two sites, one shape.

Everything else is the constant region, and its sequence is the same, or nearly the same, in every antibody of a given class. That is what makes it useful. Phagocytes have receptors for the constant region, so they can bind any antibody regardless of what it is attached to, which is how one receptor deals with millions of different pathogens.

Antibody

A protein with a quaternary structure of four polypeptide chains, produced by plasma cells, with variable regions forming two binding sites complementary to one antigen.

Variable region

The part of an antibody whose amino acid sequence differs between antibodies, forming the antigen-binding site.

Constant region

The part of an antibody that is the same in all antibodies of a class, and that binds to receptors on phagocytes.

Antigen-antibody complex

The structure formed when an antigen binds to the complementary binding site of an antibody.

The hinge between the arms and the stem is flexible, so the arms can move to reach antigens at different spacings on a pathogen's surface. In a structure-and-function answer, name the feature and then state what it allows the molecule to do.

What an antibody does, and what it does not

An antibody binds an antigen, forming an antigen-antibody complex. Everything else follows from that, and there are three consequences worth naming.

Agglutination. Because each antibody has two binding sites, one antibody can bind two pathogens at once. Repeat that across thousands of antibodies and the bacteria are clumped into a mass. The clump cannot move about the body, and a single phagocyte can engulf many cells at once rather than one at a time. Agglutination requires at least two binding sites, which is the functional reason for the two arms. Antibodies acting in this way are called agglutinins.

Neutralisation. Antibodies bound over a bacterial toxin cover the part of it that would otherwise fit a host receptor, so it cannot act; antibodies that neutralise toxins are called anti-toxins. The same works on viruses: an antibody covering the attachment proteins of an influenza virus stops it binding to and entering a host cell, and a virus that cannot get into a cell can do nothing at all.

Marking for phagocytosis. With antibody bound, the constant regions stick outwards from the pathogen's surface, and phagocytes have receptors that bind them. The pathogen is far more readily engulfed than it would be otherwise. Antibodies acting as these markers are opsonins.

In none of those does the antibody destroy the pathogen. Hydrolysis occurs inside a phagocyte's phagosome. State that the antibody binds, agglutinates, neutralises or marks for phagocytosis, not that it destroys.

Four ways to end up immune

Immunity is classified on two independent criteria: whether the individual's own lymphocytes produced the response, and whether the antigen was encountered naturally or administered. The two criteria give four combinations.

Describe the drawing

A two-by-two grid. The columns are natural and artificial, the rows active and passive. Active natural is catching measles and recovering. Active artificial is the MMR vaccine. Passive natural is antibody crossing the placenta and arriving in breast milk. Passive artificial is anti-tetanus immunoglobulin injected after a dirty wound. A note underneath says passive immunity is fast and short-lived because no memory cells are made.

Active immunity means your own lymphocytes met the antigen, underwent clonal selection and expansion, and left memory cells behind. It takes days or weeks to establish and then lasts for years. Passive immunity means ready-made antibody was introduced from outside. It works immediately, because the antibody is already there, and it fades within weeks to a few months, because antibodies are proteins that get broken down and nothing was made to replace them.

Both passive examples rely on the same property: ready-made antibody acts at once. A newborn cannot mount a full specific response for some months, so maternal antibodies supply protection, which fades over roughly six months. Someone bitten by a possibly rabid animal cannot wait the eight days a primary response takes, so they are given antibodies directly and vaccinated at the same time: passive protection now, active protection later.

Vaccination and herd immunity

A vaccine puts antigen into the body in a form that provokes a primary response without causing the disease. Depending on the vaccine, that antigen may be a weakened (attenuated) strain, an inactivated pathogen, a purified surface protein, a harmless toxin derivative, or the mRNA instructions for making one viral protein. What all of them have in common is antigen; what none of them contains is a full dose of virulent pathogen.

A vaccine is sometimes described as producing a mild form of the disease. For most vaccines that is not what happens. A sore arm and a day of malaise after vaccination are effects of the primary immune response, meaning inflammation, cytokines and a slight fever, rather than the disease. A tetanus vaccine contains no live tetanus bacteria and cannot cause tetanus.

The response to a first dose is a primary response: slow, small, and leaving memory cells. A booster is deliberately arranged to be a secondary response, producing much more antibody and many more memory cells, which is why schedules involve several doses rather than one.

Describe the drawing

Two five-by-five grids of twenty-five people. In the left grid eighty per cent are immune, and arrows trace an infection running from the first case through four susceptible people who happen to be next to one another. In the right grid ninety-six per cent are immune, the single susceptible person is the first case, and no arrows leave them because every neighbour is already immune.

Herd immunity is the protection unvaccinated people get from being surrounded by vaccinated ones. An infected person can only pass the pathogen to someone susceptible, and if nearly everyone they meet is immune, the chain of transmission breaks before an outbreak can build. This matters because some people cannot be vaccinated: babies below the recommended age, people receiving chemotherapy or immunosuppressants, and anyone with a genuine allergy to a component.

There is a threshold because transmission is a multiplication. If each infected person would on average infect R₀ others in a fully susceptible population, then a proportion of them being immune reduces that number proportionally; once the average falls below one, each case leads to fewer than one new case and the outbreak shrinks. That happens when the immune proportion exceeds 1 − 1/R₀. Measles is extremely transmissible, with R₀ around 15 to 18. Put the two ends in and do the arithmetic: 1 − 1/15 is 93.3%, and 1 − 1/18 is 94.4%, so the threshold is 93 to 94%. Coverage of 90% sounds high and is not high enough, which is why measles returns in communities where uptake slips by a few per cent.

The figure quoted in public health is 95%, and it is a different number, not a rounding of the same one. The threshold is the bare point at which transmission stops on average, in a population where everyone mixes with everyone and every immunised person is genuinely immune. Real populations are not like that: the unvaccinated cluster in particular schools and neighbourhoods, a small fraction of those vaccinated never respond properly, and R₀ is higher in a crowded city than on a farm. The 95% coverage target is what the bare threshold becomes once those idealisations are corrected: an operational figure allowing for imperfect vaccine effectiveness, waning immunity and uneven mixing. If a question gives you R₀ and asks for the threshold, calculate it and quote 93 to 94%; if it asks what health services aim for, that is the 95%.

Vaccine

A preparation containing antigens from a pathogen, given to produce a primary immune response and memory cells without causing the disease.

Herd immunity

The protection of unvaccinated individuals that results when a large enough proportion of a population is immune for transmission not to be sustained.

Antigenic variation

Change in the antigens on a pathogen's surface, so that memory cells and antibodies produced against the previous form are no longer complementary to it.

Why the flu vaccine changes and the measles one does not

Immunity is immunity to a shape. Memory cells specific to one antigen recognise that antigen and nothing else, so a pathogen that changes its surface antigens is effectively a new pathogen to an immune system that has met the old one.

Influenza does this constantly. Its surface proteins, haemagglutinin and neuraminidase, accumulate mutations, and occasionally two strains infecting the same animal exchange whole segments of genome. Your memory cells from last year carry receptors complementary to last year's haemagglutinin. Against this year's they bind poorly or not at all, so the response to infection is essentially primary again. That is why the vaccine is reformulated annually against the strains predicted to circulate, and why the prediction is sometimes wrong.

Measles virus has stable surface antigens, so antibodies and memory cells raised against it in 1985 still fit the virus circulating today. Two doses in childhood give protection that generally lasts for life. The difference between the two vaccines is nothing to do with vaccine quality; it is a property of the viruses.

HIV goes further still, mutating rapidly within a single infected person, so antibodies effective at one stage of the infection are useless months later. That, and the fact that it infects the very cells that would co-ordinate the response, is why an HIV vaccine has proved so much harder to make than a measles one.

Monoclonal antibodies

A monoclonal antibody is a population of identical antibody molecules produced by a clone of identical cells, and therefore specific to one antigen. A molecule that binds one shape and ignores everything else in a sample of blood is a very precise tool.

An immunised mouse produces B lymphocytes specific to the injected antigen. B lymphocytes isolated from the spleen are fused with myeloma cells. Selective medium removes unfused cells, and screening identifies the hybridoma that secretes the required antibody. That clone is cultured to produce genetically identical monoclonal antibodies. The steps are set out below.

Start with the antigen the antibody is required against and inject it into a mouse. The mouse mounts an ordinary specific immune response: B cells with complementary receptors are selected, they divide by clonal expansion, and some differentiate into plasma cells secreting antibody against that antigen. Its spleen is then removed and the antibody-producing B lymphocytes are isolated from it. Those cells make the required antibody but cannot be cultured on their own, because a B lymphocyte removed from the body divides only a few times in culture and then dies.

The second ingredient supplies what is missing. A myeloma cell is a cancer cell of the bone marrow, and like other cancer cells it divides indefinitely in culture. Cell lines are chosen that make no antibody of their own, so the product is not contaminated.

The two cell types are then fused. The B lymphocytes and the myeloma cells are mixed and treated with polyethylene glycol, or given an electric pulse, which makes the membranes of cells in contact merge. A cell formed from one of each is a hybridoma, and it has inherited both properties: it secretes the B lymphocyte's antibody and it divides indefinitely like the myeloma cell.

Two separate sorting steps follow, and both should be named. The fusion mixture contains unfused B lymphocytes, unfused myeloma cells and hybridomas, so it is grown on a selective medium on which only fused cells can survive: the unfused B lymphocytes die of their own accord, and the medium is chosen so that the unfused myeloma cells cannot grow on it either. That is selection, and it leaves hybridomas. Then comes screening, which is a different question. The mouse made B lymphocytes against everything it had ever met, not only the injected antigen, so the surviving hybridomas are a mixture of clones making a mixture of antibodies. Each is grown separately in its own well and its antibody tested against the antigen, and the wells that come up positive are kept.

One hybridoma is then chosen and cultured, either in large vessels of medium or in a fermenter, and because every cell in that culture descends from a single cell, every antibody molecule it secretes is identical. That is what monoclonal means. The antibody is harvested from the medium and purified, and the cell line is frozen so the same antibody can be made again years later.

Monoclonal antibody

A population of identical antibody molecules secreted by a clone of identical cells, and therefore all specific to the same antigen.

Hybridoma

A cell made by fusing an antibody-producing B lymphocyte from the spleen of an immunised mouse with a myeloma cell, so that it both makes the wanted antibody and divides indefinitely in culture.

Myeloma cell

A cancer cell of the bone marrow that divides indefinitely in culture; the lines used for fusion make no antibody of their own.

Two engineered forms of monoclonal antibody are named on drug labels. An antibody made in a mouse is a mouse protein, so a human immune system treats it as non-self and mounts a response against it, which both clears the drug and can make the patient ill. A chimeric antibody combines mouse variable regions with human constant regions. A humanised antibody retains only the mouse sequences needed to form the antigen-binding site, while most of the remaining sequence is human. This reduces the likelihood of an immune response against the treatment. Both designs are possible because the constant region is the same in every antibody of a class, so it can be replaced without altering the binding site.

Targeted cancer treatment. Roughly one breast cancer in five over-produces a receptor protein called HER2 on its cell surfaces. Trastuzumab is a monoclonal antibody with binding sites complementary to HER2: it blocks the growth signal those receptors carry and marks the cells for destruction. Healthy cells carry far less HER2, so the drug affects them much less than a conventional cytotoxic agent would, and the side effects are correspondingly narrower, though not absent. A monoclonal antibody can also carry a drug to its target. A cytotoxic drug attached to an antibody against a cancer-cell antigen is delivered to the cells that display that antigen, so the drug acts mainly on them and much less of it reaches healthy cells. Trastuzumab emtansine is trastuzumab joined to such a drug.

Pregnancy testing. The placenta produces the hormone hCG, which appears in urine. A test strip carries mobile anti-hCG antibodies attached to coloured beads, and a second set fixed in a line further up. Urine carrying hCG picks up the mobile antibodies and carries them to the fixed line, where they are trapped and the colour concentrates. No hCG, no trapping, no line. A control line further up carries antibodies that bind the mobile antibody itself, so it colours whether or not hCG is present and shows that the test has worked.

ELISA (enzyme-linked immunosorbent assay) uses the same logic in a laboratory. In a test for HIV infection, HIV antigen is fixed to the bottom of a well and the patient's plasma is added, so any antibody against HIV binds to it. The well is washed to remove unbound antibody. A second antibody, which binds to human antibody and has an enzyme attached, is added, and the well is washed again, so any enzyme left behind is held there by the patient's antibody. Finally a substrate is added, and a colour change shows a positive result. Without the washes, unbound enzyme would colour a negative sample. A control well given everything except the patient's plasma should stay colourless, which shows that the second antibody does not stick to the well by itself. Note what is detected: antibodies against HIV, which is why a test taken very soon after exposure can be negative in an infected person. Modern fourth-generation tests shorten that window by also detecting p24, a protein from the virus itself, which appears in the blood before the antibodies do.

From foxglove to the double-blind trial

A new drug reaches patients only after testing that asks two questions separately: is it safe, and does it work. The history of that protocol usually starts with William Withering, a physician in Birmingham. In 1775 he was asked his opinion of a family recipe for dropsy, the swelling of the legs and body with fluid that is often the result of heart failure. The recipe was a herbal brew, often called digitalis soup, made from twenty or more plants. Withering reasoned that only one of them was active, identified it as the foxglove, Digitalis purpurea, and spent about ten years giving it to his patients. In 1785 he published An Account of the Foxglove, describing 163 cases, including the ones in which the drug failed.

He changed from a boiled brew to dried, powdered leaf, because the strength of a brew varied. He found the dose by trial, starting low and increasing it until either the swelling went down or side effects appeared: nausea, vomiting, a slowed pulse and disturbed vision. The working dose is close to the poisonous one. Digoxin, a compound from a closely related foxglove, is still prescribed for some heart conditions, now as measured tablets.

His method lacked what a modern trial is built around. There was no control group, so he could not say how many patients would have improved without treatment. The patients knew they were being treated and Withering knew whom he had treated, so expectation could affect what patients reported and how he judged them. He chose his patients rather than allocating them by chance. Safety and effectiveness were tested at the same time, in people who were already ill.

Placebo

An inactive substance made to look and be taken exactly like the drug being tested, given to the control group of a trial.

Double-blind trial

A trial in which neither the patients nor the doctors and researchers assessing them know who is receiving the drug and who the placebo or comparison treatment.

Phase I trial

The first test of a drug in humans, on a small number of healthy volunteers, to find whether it is safe, what dose is tolerated and how the body handles it.

Phase II trial

A test of a drug on a few hundred patients with the condition, to find whether it works, at what dose, and its side effects.

Phase III trial

A large randomised, usually double-blind trial on thousands of patients, comparing the drug with a placebo or the best existing treatment.

Patients often improve because they believe they are being treated, the placebo effect, so a new drug has to do better than a placebo, not merely better than nothing. In a double-blind trial neither the patients nor the doctors assessing them know who received the drug until the results are analysed. Blinding the patients controls for the placebo effect; blinding the researchers stops their expectations influencing how they treat, examine or record a patient. Patients are allocated to the groups at random, so that differences in age or severity of illness are spread evenly between them. Where an effective treatment already exists, the control group is given that instead of a placebo, because withholding it would harm them.

The three phases test safety in small numbers before effectiveness in large ones, after preclinical tests on cells, tissues and animals.

Withering's method and a modern protocol compare point by point.

Untested drugs during an epidemic

Taking a drug through three phases takes years. During an epidemic of a disease with no effective treatment, that timetable sets the people who are ill now against the evidence that would protect people later. The usual case is the Ebola outbreak in West Africa from 2014 to 2016, in which more than 11,000 people died.

No drug or vaccine for Ebola had been licensed. ZMapp, a mixture of three monoclonal antibodies against the virus, had protected monkeys in laboratory tests but had never been given to a human. In mid-2014 it was given to a few patients, two American aid workers among the first, and the small supply soon ran out. In August 2014 a panel convened by the World Health Organization concluded that, in the particular circumstances of that outbreak, it was ethical to offer unproven treatments whose effectiveness and side effects were not yet known, provided that patients gave informed consent and were free to refuse, and that the results were collected and shared.

The case for offering such a drug starts with the patient. Ebola killed a large proportion of the people it infected, so a small chance of benefit can outweigh an unknown risk, and an adult who understands that a drug is unproven can choose to take that risk. Trials run at the normal pace finish too late for anyone ill during the epidemic.

The case against has several parts. Without a phase I trial the drug's toxicity is unknown, and harm it does to a very sick person can look like the disease. Consent from a frightened, very ill person with no alternative is weaker than the consent a trial normally requires. Supply is small, so someone decides who receives it: the first doses of ZMapp going to foreign aid workers rather than to West Africans, who made up almost all the cases, were criticised on that ground. If a drug is handed out without a comparison group, the epidemic can end with no one knowing whether it worked. An unproven drug can also draw money and attention from measures known to work: isolation, rehydration, safe burial and contact tracing.

The answer that emerged was to run trials inside the epidemic. The vaccine rVSV-ZEBOV was tested in Guinea in 2015 by ring vaccination: the contacts of each new case were vaccinated either at once or three weeks later, decided at random. Nobody received a placebo, and the comparison showed that the vaccine was highly effective. In the outbreak in the Democratic Republic of the Congo that began in 2018, a randomised trial compared four experimental treatments, and two monoclonal antibody treatments, mAb114 and REGN-EB3, gave better survival than ZMapp.

The ethics, stated evenly

An ethical discussion sets out the considerations on both sides and distinguishes factual claims from value judgements. The reasoning matters more than the conclusion.

Testing. A new vaccine is tested on animals before humans, raising the question of whether harm to the animals is justified by benefit to people, and on humans before licensing, raising questions about consent and about who bears the risk. Trials often run in the countries where the disease is common, which are frequently poorer countries: an argument that the trial goes where the disease is, and an argument that consent means less where the alternative is no medical care at all.

Control groups. A trial needs a comparison group, and giving that group a placebo means deliberately leaving people without protection during an epidemic. Using the existing vaccine as the comparison avoids that but makes the trial larger, slower and more expensive, which delays the new vaccine for everyone.

Individual and population. Vaccination carries a small risk of side effects to the individual and a large benefit to the population through herd immunity. Someone who declines is protected by others' uptake without accepting the risk themselves. Some countries make certain vaccinations a condition of school entry; others do not, and the disagreement is about how far a state may compel a medical procedure for a collective benefit rather than about the immunology.

Monoclonal antibodies. Producing them uses animals: mice are immunised and then killed for their spleen cells, which raises the same question as animal testing of vaccines, weighed against the diseases the antibodies diagnose and treat. Testing a new antibody in people carries its own risk: in 2006 six healthy volunteers in a London trial of TGN1412, an antibody acting on T cells, became critically ill within hours of receiving it. The drugs are also expensive, so a health service has to decide who receives them.

Set these out as claims and counter-claims. An argument made wholly on one side, and a list of facts with no weighing of them, are both incomplete discussions. The biology does not settle any of these questions, and saying so belongs in the answer.