Anybody want an anti-boring talk on antibody therapies? You might have heard of a certain figure being given an experimental “antibody cocktail” from Regeneron. You might have heard it was a polyclonal antibody treatment – it isn’t – it’s a mix of 2 monoclonal antibodies. This might sound like some pedantic rant thing, but it’s actually a really great segue into discussing what these terms mean, how they differ, and how antibodies (either ones we generate ourselves in response to infection or vaccination in “active immunity” or ones we are given “pre-made” in “passive immunotherapy”) can play a role in infection control!  

just so I don’t leave you hanging, let’s clear this up up front: Antibodies are little proteins made by B cells (a type of immune cell) as part of the adaptive immune response to an infection. They have unique “variable regions” which specifically bind to parts of different invaders (like viral proteins) and call for backup, etc. Some antibodies (which are called neutralizing antibodies) can bind in such a way that they prevent the virus from infecting cells, so if you administer these neutralizing antibodies as a therapy, you can hopefully prevent the virus from ramping up inside the patient. If all the copies of the antibody you’re injecting are identical, we call this a MONOCLONAL antibody. You can mix a couple monoclonal antibodies together to get an antibody cocktail (which the virus is less likely to have resistance to). This is what Regeneron’s REGN-COV2 is, and it differs from convalescent plasma, which contains POLYCLONAL antibodies – an unpurified mix of antibodies to lots of things, some of which are to the virus of interest (though potentially different parts of it), others of which are to totally different things. I don’t blame people for not knowing this and I don’t want you to feel bad if you didn’t, but I want people be able to understand the difference (if they want to).

note: At the end of the post, I’ll give an update on what we do and don’t know about REGN-COV2 (a look at Regeneron’s press release). But first, let’s talk more about what all these things are. 

All of these “passive immunotherapies” are basically an effort to bypass the sometimes lengthy process involved in the immune system learning to make really good antibodies of its own (by which time the patient might already be super sick) in order to prevent them from getting super sick. It takes your body a while to find “good antibodies” because they have to sift through millions of different ones (each B cell randomly edits its DNA to make a different one), see what works, and then make a ton of them.⠀So people typically don’t develop specific antibodies for an infection (i.e. seroconvert) until late in that infection. Good for preventing re-infection, but not so good at doing the early fighting-off, and definitely not good for preventing someone from getting infected in the first place. But if you could give a person those good antibodies without them having to find them themselves, you could potentially treat or even prevent infection (prophylaxis). ⠀

But how do you get these antibodies? Without having to go through that whole finding process?⠀

One way is with convalescent plasma – these aren’t purified antibodies, though – you just take the blood of people who’ve recovered from COVID-19, remove the non-cell parts to get the plasma, which contains antibodies, and give it to someone who needs it. But this is certainly far from efficient. That person’s plasma doesn’t only contain antibodies for SARS-CoV-2 – it also contains antibodies for a whole bunch of other invaders that person’s body has fought off throughout their lifetime. And – even when it comes to the SARS-CoV-2 antibodies present, there a whole bunch of different ones, because different B-cells experiment differently and can still make ones that work. There’s no one right solution when it comes to antibodies – but there are better ones. So some of the antibodies are stronger than others, either because they bind tighter or they bind in a better blocking position. ⠀

Some companies are looking to find the really good ones from the blood of recovered patients. Other companies are hoping that they can make lab animals do the finding process for them by injecting them with viral protein and isolating antibodies from their blood. Either way, those bodies have done the initial “finding” but you still need to weed out these potential candidates to find the good ones. So, the next step is testing them to see if they can block the virus, isolating the DNA of really good ones, and then sticking that DNA into other cells to make a ton of it. Then purifying it and injecting it into people. ⠀

terminology note: If all the copies of the antibody you’re injecting are identical – they all came from the same “parent” B cell, we call this a MONOCLONAL antibody. This differs from convalescent plasma, which contains POLYCLONAL antibodies – antibodies from lots of different B cell parents. Also, since we’re on this term tangent, the thing the antibody recognizes (e.g. the coronavirus Spike protein) is called an ANTIGEN, the specific part of the antigen the antibody binds to is called the EPITOPE, and the specific part of the antibody it binds to is called the PARATOPE. ⠀

The “getting the lab animals to make antibodies” part is a lot like how some vaccines work, with you being the human and getting to keep the antibodies you make. But there are really important differences between vaccines and antibody therapies. ⠀

Vaccines are designed to induce ACTIVE IMMUNITY – their goal is to introduce someone’s body to parts of the virus or an inactivated or weakened version of the virus in a harmless form and get that person’s body to learn to make its own antibodies against it. This provides long-lasting protection, because your body will keep making it. ⠀

This is different from passive immunity – passive immunotherapies include convalescent plasma and monoclonal antibody treatments – and it’s goal is to just give someone the antibodies (premade). This way you know they’re getting the good stuff, but they only get it when you give it to them – stop giving it and the body won’t have it because it doesn’t know how to make it itself (teach a man to fish thing…)⠀

Both have usefulness, but both are very different and easily confused. I’ve talked a lot about vaccines , and touched on convalescent plasma, but today I want to review monoclonal antibodies (mAbs). Monoclonal antibodies are used to treat cancer & autoimmune diseases (when you see the ending “-ab” on drug names, this is why). But they haven’t been used much for infectious diseases – except a few, most notably Ebola. In fact, in a weird coincidence, it was this Ebola treatment that outcompeted Remdesivir as an Ebola treatment, leading Remdesivir to be sidelined until recently. more on remdesivir here:⠀

That same company (Regeneron) which made that Ebola drug is now working on a coronavirus one and I will tell you more about it – as well as some of the strategies being used by different companies to find & make good ones, but I think it helps to have a little background on antibody making and make-up, so…⠀

When an animal is confronted with something unusual, like a protein from an invading virus or bacterium, it can mount an immune response to destroy the invader, but only if it knows that the thing is unusual, and the way it knows this thing is “foreign” is because it binds to proteins called antibodies that are made by a type of immune cell called B cells, which make antibodies and put them on display (sticking out of their membrane) to “keep watch.” Antibodies are made by a type of immune cell called a B cell as part of the adaptive immune response when you first encounter the thing. So that next time you encounter the thing, alarms will ring!⠀

All antibodies bind to the thing they “recognize” (often a viral protein) but if antibodies happen to bind to a virus part in such a way that they prevent that virus from infecting a cell, we call them neutralizing antibodies (they “neutralize” the viral threat). For SARS-CoV-2 (the novel coronavirus that causes the disease COVID-19) this usually means binding to the virus’ Spike protein (the one that juts out from the viral membrane) in its Receptor Binding Domain (RBD). This binding can be directly in the spot it uses to latch onto the cell’s ACE2 receptor (the Receptor Binding Motif), out-competing it – or it can bind in some way that prevents the Spike protein from undergoing the dramatic shape-shifting (conformational changes) required for the Spike protein to get cleaved by proteases, shoot out its inner parts, latch onto the cell membrane, and fuse it with the viral one. more on that here: ⠀

So if B-cells are making these, how does anyone ever get sick? Problem is, the B-cells don’t know what they’re going to have to face in the future – they can’t predict that someone with COVID-19 is going to sneeze on you. And they don’t even know it’s out there (your cells aren’t tuned in to CNN). So they don’t know what to make. Instead of stocking up on every possible antibody, which is like limitless, they wait until an infection starts and then “experiment” to find what works – different B-cells make different antibodies (they have different variable regions) and then the ones that bind the virus (and not the host) get selected for and more of those get made. This process (called clonal selection because all of the cells made from that original experimenter are clones of it) takes time (which is why antibodies don’t show up until late in infections and your immune system relies on more generic first responses). ⠀

Different B cells make different antibodies (which have different variable regions) but each B cell can only make one type because the random “experimenting” they do to decide what to make involves actually changing their DNA in a process called somatic recombination – don’t worry – the “somatic” tells you that these genetic changes are happening in non-germline-cells – the changes are permanent for this B-cell, and all of the cells that arise from it (clonal cells), but the changes don’t effect any of the other cells in your body, like the ones that would get passed on to your children. ⠀

How it works is that antibodies have a “generic” adapter region (constant region) as well as variable regions which are unique to different antibodies and are where the experimentation occurs. B-cells “randomly” choose what to make by piecing together different options for different parts of the variable regions of antibodies through a process called V(D)J recombination – more on that here:  Immature B cells have a whole menu of options to choose from, but in their maturation process they go through a form of somatic recombination, where they choose from that menu – and remove the other choices so they (and all the cells that come from them) don’t have options anymore – they’re forced to specialize in making that specific antibody. ⠀

Bottom line – different B-cells randomly specialize to make different antibodies. Those antibodies all, by chance, happen to recognize different things. These things can be different molecules all together (e.g. Antibody 1 recognizes Protein A and Antibody 2 recognizes Protein B) OR the things can be different parts of the same molecule (e.g. Antibody 1 & 2 both recognize Protein A, but different parts of it). ⠀

When someone gets infected with a virus, immune cells called “antigen presenting cells” engulf the virus, cut its proteins up into little pieces, and then display those pieces on their surface. Then, B cells that happened to have made an antibody that can bind it do bind it, leading to those B cells being selected for. Those chosen B cells start making more copies of themselves, some of which (memory B cells) continue to display the antibody sticking out from their membrane, while others (effector or plasma B cells) secrete the antibody into the serum. This is why convalescent plasma can be used as a therapy. ⠀

As we mentioned before, this plasma (the cell-less part of blood) contains a lot of stuff. It has antibodies to all sorts of invaders that person has fought off in the past. And, even the antibodies against the virus of interest (e.g. SARS-CoV-2) likely target multiple different viral proteins. In the case of a lab-injected animal, where you’re only injecting a single protein to serve as an antigen (typically the Spike protein here) you’re only getting them to develop antibodies against that antigen (though they still have all their antibodies from previous infections). You can purify out the antibodies that really do bind the antigen, but you will still have a polyclonal mixture of antibodies that bind to different parts of that protein with different “goodnesses.” ⠀

You can test how good they are using things like an ELISA, which measures binding to a protein of interest (e.g. coat the wells of a plate with an antigen (e.g. Spike protein) -> add antibodies to test -> let them potentially bind and wash off non bound stuff -> add labeled antibody that binds to the generic part of the first antibody -> wash off non bound stuff -> see if you see signal -> shows you if there was antibody there that stuck) and neutralization assays to test if the virus can actually infect cells. ⠀

But, even if you find great ones, without the B cells they came from, you have a limited supply. If you want to find the good ones AND make more, you need to go straight to the source – you need those B cells!⠀

What companies are doing is taking bloods’ B cells, splitting them up so that each B cell gets its own well in a chip, letting that B cell secrete antibodies, and seeing if they make good ones (such as with antigen-coated beads in a smaller scale version of an ELISA). AbCellera Biologics, the Canadian company working with Eli Lilly, who started a clinical trial of their mAb LY-CoV555, has a microfluidics chip that can test 200,000 cells at once! ⠀

In the figures, I show a typical workflow for the Berkeley Lights system used by the Vanderbilt team led by James Crowe. You can see a webinar about it here: ⠀

They can then take the “top hits” and sequence their DNA’s antibody instructions and stick those instructions into a piece of DNA called a vector that contains signals and binding sites that will enable expression cells to make the antibody when you stick the vector in them. The
“sticking-in” is called transduction, and, since you “recombined” DNA the antibodies produced are called recombinant antibodies. Scientists express proteins recombinantly all the time, so there are a lot of established protocols and good expression systems in place, but it can be labor/time/cost expensive, which is a downside to antibody therapies. ⠀

For large-scale production, antibody makers often use CHO cells – a cell line that comes originally from Chinese Hampster Ovaries – these are good to use because, as mammalian cells, they’re more similar to human cells than bacteria or yeast, so they’re better at making more complicated proteins that require special folding helpers and modifiers, etc. ⠀

But before you get to large-scale, you want to make sure you only waste all that time/money/etc. making the really good ones. So they start small, doing expressions on the really really small scale, in tiny wells in a dish and cells such as yeast or bacteria and doing further testing before they go big. ⠀

Speaking of cell similarity, antibodies are used a lot in the lab – not to treat people, just to see if things are bound to membranes in a Western blot, etc. and for these we use antibodies made from bunnies, goats, mice – we can do this because we’re not sticking these into people. If we did, those peoples’ immune system would attack them as foreign. So if we want to make antibodies we stick into people, they “need” to be human antibodies. Even though we’re making antibodies in cells from hamsters, the proteins are human because the DNA you put in was. So that’s not a problem. But if you want to get a mouse to make antibodies  through the learning process (e.g. inject it with viral protein and see what it comes up with), you need to use a really special mouse – one whose immune system has been “humanized” so that the antibodies it makes are human ones. This is the strategy being pursued by the company Regeneron (they’re the group that made the Ebola antibody treatment) ⠀

This regeneron treatment is what Donald Trump got and it’s an antibody “cocktail” they call REGN-COV2 – their experimental treatment contains a mix of 2 different monoclonal antibodies (REGN10933 and REGN10987). It has more than one antibody, but it’s different then a conventional “polyclonal” antibody because you only have (lots of copies of) a couple of (really good) antibodies that you choose. This can have benefits including giving you backups in case one epitope of the virus mutates so that the antibody no longer binds.  You get these backups with active immunity as well because you get a polyclonal response – your body stocks up on lots of different antibodies, hopefully strong ones!⠀

The phenomenon whereby a virus is able to evolve to evade antibody recognition is called “viral escape.” You might hear a lot of scary reports on the news about the coronavirus mutating. In reality, mutations are actually pretty common, because the virus has to make tons and tons of copies of itself, and its copying machinery isn’t perfect. But most of the mutations aren’t harmful. In fact, a lot of them are probably harmful for the virus, not for you! The viral RNA contains “recipes” for making viral proteins. Mutations in this RNA can weaken a virus or strengthen a virus, but most mutations don’t really do anything. Some mutations in the viral RNA don’t even change the corresponding protein letter (amino acid) because there’s some redundancy in the genetic code (for example, the RNA sequences GCU & GCC both “spell” the amino acid alanine). ⠀

There are, however, rare mutations in the Spike protein which can prevent the antibody from binding and therefore allow the virus in. The companies that are manufacturing monoclonal antibodies are doing tests where they try to grow the virus in the presence of the antibodies they’re developing. Lots and lots of virus. Virus particles with those rare “escape mutations” will have a growth advantage and thus will outcompete their friends. The scientists can then isolate them and look to see where the mutations are. ⠀

They find that these mutations usually occur in the same couple of amino acids. They can then look to see how frequent those mutations are in actually patient blood and, thankfully, they’re really rare. Therefore, it’s unlikely, although possible, that a person would be resistant to the antibody – at least at first. But there are a lot of viral particles in an infected person. And while they’re mostly identical, there could be copies with random mutations that develop during the copying process within the person. And there is still the possibility that, similarly to the virus in the dishes that the companies were testing, once you apply the “selective pressure” of an antibody, copies of the virus with random mutations that allow them to escape will have a growth advantage within that person’s body which allow them to gain prominence. ⠀

If you only had a single antibody, this could potentially be a problem. However, even if the virus has that one mutation that makes it resistant to one antibody, it’s much less likely that a virus would have mutations that prevent binding of multiple antibodies. So if you include multiple antibodies, you reduce the risk of viral escape. ⠀

We don’t yet have much data on efficacy of the cocktail yet. September 29, Regeneron put out some. The good news: it’s from a “gold standard” type of clinical trial – placebo-controlled so some people got a fake drug (placebo) to compare to; randomized, so that people were randomly assigned to receive placebo, low dose (2.4g) or high dose (8g); and double-blind, so neither doctor nor patient knows who got what. Bad news: the “data” were provided via press release so we haven’t really seen it yet…

What the press release says is:

  • patients on the high dose had a statistically significant reduction in the amount of viral RNA in their blood (viral load) compared to patients on the placebo. for patients in the low dose group, the reduction wasn’t statistically significant

So that seems like a plus for the high dose, but if you look at symptom alleviation, the trend is reversed (though admittedly not that different potentially). An important caveat – at this point they break the patients up into a couple groups, which likely indicates that the effects weren’t striking overall – so who was most likely to benefit? Patients who hadn’t yet* started producing their own antibodies (were seronegative) and patients who had high viral loads to start with. 

  • “Among seronegative patients, median time to symptom alleviation (defined as symptoms becoming mild or absent) was 13 days in placebo, 8 days in high dose (p=0.22), and 6 days in low dose (p=0.09). Patients with increasing viral loads at baseline had correspondingly increasing benefit in time to symptom alleviation.”
  • “15.2% of placebo-treated patients, 7.7% of patients treated with high dose and 4.9% of patients treated with low dose required additional medical visits” So again we see that the low dose did slightly better (though I don’t know if significantly and these are small numbers so we probably can’t make too much of it other to say that viral load might not be a great correlate to symptom severity. Which isn’t that surprising in my opinion because, if by “viral load” they’re just referring to PCR results, and we know that patients can shed harmless inactive viral RNA for months, it seems like a poor primary endpoint to measure (though an easy one to measure) 
  • they don’t give data on symptom reduction in the other patients, so I’m guessing it wasn’t significant (which is a significant problem if the key is to help patients by alleviating their symptoms (things they feel and tell you) and not just to improve their clinical signs (objective measurable things that the patient doesn’t have to tell you)).

They make a lot of the difference between the outcomes of patients who had already started making their own antibodies before treatment started (who saw less benefit) and those who hadn’t, making it seem that these patients were inherently different to start with. In the press release: “The greatest treatment benefit was in patients who had not mounted their own effective immune response, suggesting that REGN-COV2 could provide a therapeutic substitute for the naturally-occurring immune response. These patients were less likely to clear the virus on their own, and were at greater risk for prolonged symptoms”

And in this STAT article, “Nahid Bhadelia, the medical director of the Special Pathogens Unit at Boston Medical Center, said that the results indicate it may be useful to test patients’ antibody response to identify those who may be headed toward worse clinical outcomes” 

But, from an admittedly somewhat naive standpoint, it seems more likely to me that those patients were just at different time courses in the disease – for example, if a patient was infected more recently, they’d be expected to be seronegative. Just because they haven’t yet established a robust antibody response on their own doesn’t mean that they won’t. It would make sense that they’d get more benefit because they don’t have anti-Spike antibodies of their own already on the scene. For patients who do (those who are seropositive already) they already have some – they might not be as great as the ones in the cocktail, but they’re already helping so that boost wouldn’t help as much.

Bottom line, it’s really unclear to me at this point without more data. 

Speaking of more data, what’s next? Regeneron’s hopefully gonna publish this data from the 275 patients soon. And their second phase of the study will be in 1,300 patients who will be followed for 29 days with viral load testing every couple of days and clinical outcomes. So that should give some more data, but  if by “viral load” they’re just referring to PCR results, and we know that patients can shed inactive viral RNA for months, it seems like that’s a kinda bad endpoint to use to claim usefulness? 

The good news is that these trials are being done in non-hospitalized patients, for whom there’s less currently available (remdesivir, for instance, can only be given to hospitalized patients). 

There are also trials underway on hospitalized patients, including the U.K. RECOVERY trial, as well as trials on its use as a prophylactic (can it help prevent exposed people from getting sick?)

A big downside of monoclonal antibodies or cocktails of them is manufacturing capacity & cost. This is one of the reasons why Eli Lily’s treatment, LY-CoV555, only contains a single antibody, gambling that the increased cost & resources for adding a second would outweigh the benefit. ⠀

That may sound like a really crass economics decision, not taking into account patients, but it might actually be able to benefit more patients. It’s really unlikely that even a monoclonal antibody, let alone a cocktail, will be available to patients in developing countries. The monoclonal antibodies currently on the market (for cancer, autoimmune diseases, etc.) are some of the world’s most expensive medications, and 80% of them are sold in the US, Europe, & Canada (despite the fact that those countries only make up 15% of the global population). ⠀

Groups of prominent scientists at the International AIDS Vaccine Initiative (IAVI) and Wellcome are calling for infrastructure, policy, and other practices to be put into place to make monoclonal antibody therapies more widely available to the global population.

That’s, unfortunately, a hard sell in a $-centric world. That’s, unfortunately, a hard sell in a $-centric world. And, in the meantime, people in developing countries do, however, have easier access to convalescent plasma, but doctors and scientists still aren’t sure how well that works, although it at least doesn’t seem harmful and trials are underway 

more Covid-19 resources: ⠀

more on topics mentioned (& others) #365DaysOfScience All (with topics listed) 👉  ⠀

Leave a Reply

Your email address will not be published.