True Facts about Antibodies — Deplatform Disease

I recently made a post in one of my science Facebook groups trying to explain some of the basics about antibodies to people and was asked if it could be shared. The group, however, is private so sharing its content is not permitted but I realized I could just reproduce the post below, so this is that (also I can include hyperlinks here which is nice). It’s more informal than the content of this blog usually is. I have modified it slightly to give some additional details that I thought were important and were not included in the first version.

Also, yes, this title is a reference to the True Facts series by zefrank on youtube, each of which is a true work of art.

I wanted to have a short post telling you all about antibodies, because they’re amazing. Let’s go from the basics and work our way up. I won’t make this fully comprehensive but I do want you all to have a better idea of how antibodies do and don’t work given how much they appear in the news lately. Also the principal reference for most of this post is this as well as everything hyperlinked or listed in the captions.

Antibodies are glycoproteins (proteins that are decorated with sugars) that are made by antibody-secreting cells (ASCs- very creative name, right?), which are branches from the family of B cells (B lymphocytes). More precisely, antibodies refer to the secreted form of the B cell receptor (BCR) which is present on the membrane of B cells and their descendants. A more general term that I like is “immunoglobulin” which doesn’t specify whether the protein is attached to the membrane or secreted (and it’s the same protein). In other words: immunoglobulin, antibody, and B cell receptor are basically synonyms so don’t get intimidated if you see the terminology get changed up. Antibodies are Y-shaped molecules (mostly), which are made of two proteins- a heavy chain and a light chain. Structurally, antibodies have a Fab region (fragment antigen-binding) which is the tines of the “Y” and Fc region (fragment crystallizable) which is the stalk of the “Y.” These are connected by a hinge region.

B cells themselves do not secrete antibody but differentiate into plasma cells that do. Initially in an immune response you make a lot of plasmablasts, short-lived plasma cells. A portion of these may receive cues to migrate in the bone marrow and establish survival niches to become long-lived plasma cells. These can persist for life. Some plasma cells can also populate your mucosal tissues and persist for decades there too, wherein they become a source of persistent antibody secretion. You also have memory B cells which don’t secrete antibody but can rapidly differentiate into antibody-secreting cells, some of which may go on to become long-lived plasma cells. For details on B cell memory, I strongly recommend this review.

For simplicity, I won’t discuss how antibodies are made in detail, but it’s really fascinating and you can read about it here (also here if you prefer a more comprehensive context of the process as it relates to the development of a B cell- though the reference is a tad dated now). It’s been estimated that humans can make as many as 10¹⁸ (that is- a quintillion) distinct immunoglobulins, which is an insane number. If you know a bit about molecular biology and human genetics though, that shouldn’t make any sense because the human genome encodes only about 23,000 proteins. There are 3 regions in the genome that encode antibodies. One encodes the heavy chain. Two encode the light chains, either κ or λ (for the most part, these aren’t thought to be functionally different (though some recent studies may challenge that) but the fact that there are 2 gives the B cell 2 chances to make an immunoglobulin in case it fails the first time; if it fails both times there’s also receptor editing but I won’t go into those details here). The B cell will literally mutate its genome to construct the antibody in pieces from variable, diversity, joining, and constant regions (in the case of the heavy chain) or variable, joining, and constant regions in the case of the light chain. The B cells can also activate a program where they increase their mutation rate 10⁶ times through an enzyme called AID (activation-induced cytidine deaminase) which will replace the C’s in the DNA with U’s at the region where antibodies are encoded. U is not supposed to appear in DNA (they do appear in RNA though), so this triggers the B cell’s DNA repair machinery, resulting in the inadvertent introduction of mutations. Incidentally, U and C can spontaneously interconvert through this deamination reaction, which is why DNA contains T instead of U- this way if U appears, it triggers DNA repair machinery. This process is called somatic hypermutation (SHM). Somatic hypermutation works together with another process called affinity maturation which allows you to test your antibodies against a target antigen over and over and evolve antibodies that progressively bind more and more tightly. Sometimes this can lead to antibodies that cross-protect against many different strains of a pathogen by binding tightly to a well-conserved site on it. You heard correctly:

Your immune system evolves to deal with pathogens.

*boom* Did I blow your mind? I know I did. You don’t have to say anything.

Okay so, I should give you a bit of an overview of the different types of antibodies (not all 10¹⁸- don’t worry). Broadly, antibodies can be grouped in terms of isotype which describes which constant region the antibody’s heavy chain has. Humans have the following isotypes:

  • Immunoglobulin M (IgM): All B cell antibodies start out with the IgM isotype and then undergo class switching (aka isotype switching) to become one of the other classes. The specific one switched to depends on the cytokines the B cell is exposed to which depends on the threat the immune system is responding to. IgM antibodies typically exist as pentamers (5 antibodies held together in one) linked together with a J-chain, which means they have a total of 10 binding sites. They generally are not very high-affinity (don’t bind very tightly) but they compensate for it by binding with many many binding sites. IgM may also appear as secretory IgM (see the discussion of secretory IgA). We also have a population of B cells that make so-called “natural IgM” which binds weakly to a bunch of targets and helps you to tread water while you’re trying to ramp up your humoral immunity in the course of an infection.
  • IgD: No one is sure what IgD does. It’s a bit different from the other isotypes in that it exists as an alternative spliceform of the initial IgM (meaning that the RNA encoding the antibody heavy chain has the μ and δ constant regions and splices one of them out to make either an IgM or IgD) and B cells have to make it as a quality control checkpoint in their development. It also shows up in mucosal secretions so it’s thought to have some role there in contributing to protection but tbh we’re all kinda just like ¯\_(ツ)_/¯
  • IgE: I am sure some of you are not fans of IgE. IgE is the antibody isotype responsible for most allergies. However, IgE is also critically important in immune responses against parasites (especially helminths). IgE has the shortest half-life in serum of just 2 days. Humans actually have a second IgE constant region for the heavy chain but it’s now a pseudogene and doesn’t make a functional product. IgE is also thought to play a role in defense against venoms and indeed, envenomation is thought to be how we evolved defense against anaphylaxis. A comprehensive discussion of all the things IgE does is beyond the scope of this current post.
  • IgA1 and IgA2: Humans have 2 IgA subclasses that differ from each other a bit in structure, which together are the most abundant antibodies at the mucosal surfaces. Mucosal surfaces are basically the parts of the body that have mucosal secretions. They are all in contact with the outside world and thus constantly exposed to a sea of pathogens. These would be your urogenital tract, your digestive tract, your respiratory tract, your eyes, etc. There are some exceptions to IgA being the major isotype at mucosal surfaces though. IgA1 is a bit more T-shaped (the other antibodies have kind of a Y-shape) and has a longer hinge region that actually makes it more susceptible to cleavage by proteases- like those expressed by pathogens (e.g. meningococcus). IgA1 is more abundant in plasma than at the mucosal surfaces compared with IgA2. IgA2 has a shorter hinge region, which is thought to make it protease resistant. IgA at the mucosa exists typically in the form of secretory IgA (sIgA), which comprises 2 linked molecules of IgA, a J chain, and the secretory component (this is the cleaved form of the polymeric immunoglobulin receptor, pIgR). pIgR recognizes the J chain present in IgA dimers and IgM pentamers. Like IgM, IgA can oligomerize to make tetrameric or even pentameric structures and because of its abundance at the mucosa is thought to be the most important antibody for protection there (for instance anti-spike IgAs are very potent neutralizers of SARS-CoV-2; more on what that means in a minute). Importantly, IgA in plasma does NOT become IgA in the mucosa even though it is the second most abundant antibody isotype in plasma. Mucosal IgA has to be made locally by IgA-secreting plasma cells in the lamina propria.
  • IgG1/2/3/4: We have a bunch of IgG antibodies. IgG is the most abundant antibody isotype in the plasma, and it is a marker of matured immune responses. Most of the cytokines responsible for class switching will direct B cells to make one of the IgG subclasses in addition to IgA/E. The subclasses are named based on their relative abundance in the serum with IgG1 being the most abundant. IgG antibodies have the longest half-life in serum, with IgG3 being an outlier of just 7 days but most of them last 20+ days which is really neat. The reason for this is there’s another receptor called FcRN (the neonatal Fc receptor- I’ll tell you all about Fc Receptors next; it’s so named because it’s responsible for the transfer of antibody across the placenta to the fetus but it’s expressed in a bunch of tissues. IgG is the only isotype known for sure to be able to move across the placenta. Most sources will say other isotypes can’t do it but there’s some controversy). FcRN can pick up IgG floating around and basically hide it in the cell for a little while and then release it later. To give you some more details of how the different subclasses differ:
  • IgG1 is made basically in response to any soluble protein antigen.
  • IgG2 is the major antibody isotype made against bacterial capsular polysaccharide (sugar) antigens.
  • IgG3 is an extremely potent inducer of inflammation (will explain why shortly).
  • IgG4 is also induced by allergens along with IgE and has a role in promoting tolerance. It is also unique in that it can split in half vertically down the middle part of the “Y.” There is a group of autoimmune diseases called IgG4-related diseases which seem to involve inappropriate secretion of IgG4 by plasma cells.

Okay, okay- but how do antibodies… do? Great question!

Most immunology books will tell you antibodies have 3 major functions (and then they will tell you that was a lie and introduce 50 more like I am about to do):

  1. Neutralization: All antibody isotypes are capable of neutralization (even IgD). This occurs via the Fab region of the antibodies (the paratope), which sticks onto a target on the antigen (the epitope). Specifically, neutralization means that the antibody binds and the antigen can no longer function as appropriate for that antigen. For example, a neutralizing antibody against SARS-CoV-2’s spike protein is one that blocks the interaction with the viral receptor (usually ACE2). It can do this by covering the receptor binding domain (RBD) which the spike protein uses to interact with ACE2 or it can bind to another site of the spike protein that in turn makes the key parts of the RBD inaccessible to ACE2 to prevent secure attachment long enough to enter the cell either by fusion at the membrane or through endocytosis. Neutralization is also critical when you’re talking about toxins. For instance, vaccines against diphtheria, tetanus, and pertussis all contain toxoids -chemically inactivated forms of toxins each of those bacteria make that structurally resemble the toxin- with a bit of aluminum salts as an adjuvant (stimulator of the immune system) to elicit neutralizing antibodies against those toxins. This way if you encounter the toxin, you are not affected by it because it is rapidly cleared by the immune system before they have a chance to act. This is also how antivenoms get made- it’s an antibody cocktail against the venom. Antibodies that are not neutralizing are sometimes called binding antibodies. Whether or not an antibody is neutralizing depends on the specific epitopes it binds and how tightly. Binding antibodies are generally NOT useless or harmful because they can have other really important functions.
  2. Neutralization is much more complex than it may seem at first blush. Most viral immunologists would regard any antibody-mediated processes that stop the spread of a virus from cell to cell to be a form of neutralization. Thus for instance, antibodies targeting influenza’s neuraminidase protein, which helps to permit its exit from infected cells by cleaving sialic acid residues, can also be neutralizing (if they stop this process) even if they failed to block entry of the virus into the cell. It’s even more complex than that though. For example, TRIM21 is a protein in the cytosol of cells that is used to carry out intracellular antibody-mediated degradation. Complexes of antibodies bound to their target protein antigen can be taken up by cells and TRIM21 will recognize antibodies (it’s not really specific to any particular isotype) to activate the proteolytic machinery of the cell to degrade that protein. This pathway seems to be particularly important in the control of some viruses, though on its own doesn’t seem to be sufficient for controlling most of them. Note however that if the antibody induces the TRIM21 effector mechanism cannot be detected through standard neutralization assays. Additionally, neutralization generally is not something that should be assessed one antibody at a time because sometimes it is combinations of antibodies that become potently neutralizing whereas no individual antibody is together (cooperativity), but this does depend on the question being asked. Furthermore, when evaluating virus neutralization in particular, it’s critical to emulate the conditions in the cell that’s infected in the relevant host as closely as possible. To give a current relevant example, this study found that the S2P6 antibody neutralizes SARS-CoV-2 spike protein very well in cells that express TMPRSS2 and less well in those that didn’t. Why? Probably because TMPRSS2 allows SARS-CoV-2 to enter through the cell membrane rather than the endocytic pathway, which involves exposure to acidic environments that likely disrupt antibody binding to the spike protein. The TMPRSS2 pathway dominates the means by which SARS-CoV-2 enters cells by a wide margin (this is thought to be a major reason why hydroxychloroquine didn’t show effectiveness in lung cell lines expressing it).
  3. Opsonization: Antibodies are also capable of making the antigen easier to recognize by the other machinery of the immune system for consumption by phagocytes (opsonization), but different isotypes have different levels of skill with this. IgG1 and IgG3 are the best at it, and the other isotypes are kind of meh about it. Opsonization is driven not by the paratope-epitope interaction like neutralization but by the interaction of the Fc region of the antibody with Fc receptors (told you we would get there). There are a bunch of these (you can see them in table 1), and you can tell which antibody isotype they are specific to based on the Greek letter they use (e.g. ε for IgE, γ for IgG, etc.). DC-SIGN is a weirdo though (kind of). It’s a C-type lectin meaning it actually recognizes sugars, and it helps dendritic cells to pull IgA across the intestinal epithelium for example but it’s not itself directly recognizing antibodies (at least the protein part). Opsonization allows antibodies to carry out antibody-dependent cellular phagocytosis (wherein a phagocyte like a macrophage can swallow and digest the antibody and whatever is attached to it).
  4. Antibodies also have a number of other effector functions through their Fc region. Antibody-dependent cellular cytotoxicity (wherein a cytotoxic cell like a natural killer (NK) cell can kill the cell the antibody has bound- this is mostly relevant for viral infections as the virus is budding out of the infected cell), as well as promote some specialized forms of cell death (e.g. NETosis wherein neutrophils release their DNA and a bunch of antimicrobial peptides), cytokine release (including some chemokines), release of reactive oxygen species, etc. Basically antibodies can do A LOT beyond just binding and neutralizing.
  5. Complement activation: Complement refers to a network of proteins in the blood that are basically like little bombs. They are made in an inactive form and can be triggered to activate a cascade of reactions (I won’t overwhelm you with the details) that ultimately result in formation of the membrane attack complex (MAC) which makes a pore in the membrane of the cell (e.g. bacteria) that the complement is attached to, killing it. IgM is incredibly good at complement fixation, with IgG3 being second best because its giant hinge region lets C1q bind to initiate the classic pathway of complement activation. Complement activation can also be used to inactivate enveloped viruses, but is mostly thought of in the context of bacterial infections. If I also discussed the complement cascade in detail on this post I am pretty sure some people would riot so I’m just going to leave it there for now.

Many of these processes are actually extremely destructive and promote tissue damage and inflammation. The immune system works very hard to control that inflammation (a major reason why even theoretically trying to boost your immune system is not a great idea- even though you can’t). There are some exceptions though. Antibodies can absolutely be anti-inflammatory. For example, it’s not fully understood how, but high dose intravenous immunoglobulin (IVIg) is known to have a strong anti-inflammatory effect. IgA is somewhat unique because it is thought to neutralize in a non-inflammatory manner which is known as immune exclusion, which is partially mediated by the FcαR family. The details of this process are incompletely understood, however, it is known to play an especially important role in the gut because it helps to suppress bacterial and food antigens from triggering inflammation.

In sum, antibodies have a remarkable synergy with basically every component of your immune system and your immune system is incredibly skilled at taking full advantage of an expansive repertoire of antibodies to protect you from disease. Note however that there is an entire additional arm of immunity- cell-mediated immunity (CMI)- which antibodies depend on and interact with that I didn’t discuss here because… well that’s a semester of immunology right there.

Okay, so, now let me tell you about my ulterior motives in making this post. Firstly, I was really frustrated by this false perception that antibodies can be either binding or neutralizing and nothing else. There is an entire spectrum to antibody function that is massive and goes far beyond the ability of an antibody to neutralize its target antigen. We are relying really extensively lately on neutralization assays to give us status reports of how recovered patients and vaccinees will fare against the new variants. Neutralizing antibodies are clearly a really central component to our protection against SARS-CoV-2/COVID-19, but they are not the whole story. In fact as I have said before, these assays don’t even capture the full spectrum of neutralization possible. They don’t, for instance, show you anything on the other effector functions of the immune system work with antibodies -even if they are not neutralizing- to clear the infection (in this case I am referring specifically to TRIM21 as a neutralization mechanism). These assays are also imperfect at capturing some other important virological realities. For example, there’s accumulating evidence that the reason the delta variant spreads so well is related to the incredible speed at which it replicates, which presumably means that individuals are exposed to larger doses of virus over shorter periods of time, but these neutralization assays can’t really capture the difference in viral inoculum size, which may lead them to understate how effectively the virus evades antibodies. Most assays suggest that the Delta variant isn’t particularly skilled in evading neutralization compared with some other variants like Beta which has extensive ability to evade neutralization in pseudovirus assays and against which several vaccines performed markedly worse (Pfizer and Novavax both had significant declines in their effectiveness against the Beta variant). However, clearly, this alone does not predict whether or not a given variant will be successful as Delta seems to presently be outcompeting everything. When I read the comparisons by the CDC to chickenpox in terms of transmissibility (R0 specifically), I initially thought that it was overblown, until I saw the CDC MMWR of the Marin County outbreak. To be sure, the exposure was extensive as the teacher was in contact with her students for several days (and was certainly contagious well before the onset of her symptoms) and this was a superspreading event. However, the school had taken virtually every precaution I could think of short of the teacher wearing her mask while reading to the students (which to be fair is among the most important things she could have done, second only perhaps to being vaccinated). Basically- virus neutralization studies are incredibly important and valuable as they can certainly flag concerning variants, but they are also very reductive models for how the immune system handles an infection. It’s important to be mindful of the limitations of these studies and not to base vaccination policy solely on them and consider the broader picture.

That said, the picture with vaccines is far from calamitous even with the delta variant, in spite of what headlines may lead you to believe. For details on that, I refer you to Dr. Katherine Wu’s latest in The Atlantic.

Originally published at on September 3, 2021.

I write about vaccines here. You can find me on Twitter @enirenberg and at (where I publish the same content without a paywall)