Antibodies as a therapy
Let’s have a look at what is (in my opinion) probably our best shot at a reasonably short-term targeted therapy against the COVID-19 epidemic: the possibility of using monoclonal antibodies. These can be developed more quickly than vaccines, and a lot more quickly than a new targeted small-molecule antiviral.
Like everything else, though, there are some tradeoffs, which will become apparent as we go into the details. So, a quick antibody refresher: recall that if you successfully fight off an infection (from the coronavirus or most anything else) then you have probably raised antibodies to that pathogen. I have a post here with some basics; like all of immunology, though, it’s a subject that just keeps on getting more complicated the further you go into it. As a quick example of one of those complications, it’s certainly possible to clear out an infectious agent without a big antibody response, if your broad-spectrum innate immune system is up to the job. In fact, it appears that some younger people exposed to the coronavirus might be doing just that.
But most survivors will end up with antibodies, picked from the insanely huge number of different ones that every human carries around and amplified in their bloodstream because they proved effective against the virus. These are produced by B cells that became primed during the immune response, and memory B cells can survive for decades, ready to kick up production if the same infection (or something close enough to it) shows up again. Now, some pathogens can dodge this defense: malaria parasites are famously good at it, and influenza viruses mutate theirs constantly (so the flu virus you encounter this year is never really the same as the one you ran into last year – thus the need for a yearly flu vaccine that tries to predict This Year’s Model). But there are many diseases that confer lifelong immunity, or at least many decades of it. Ideally we’d want to suddenly flip everyone’s immune system over to a full-coronavirus-seek-and-destroy state, and that, folks, is what a good vaccine could do for you.
Convalescent Plasma
In the interim, though, what about finding some antibodies from people who have been infected and who then recovered? As people have been hearing about, one thing you could do is get donated blood plasma from such people and administer it directly to a sick patient. The antibodies in the convalescent plasma should do exactly what they did in the original patient: bind to the coronavirus particles whenever they encounter them. Getting this to work on scale as a therapy has some real challenges, it should be noted, but a six-company consortium is trying to get it to work (Takeda, Biotest, CSL Behring, BPL, LFB, and Octapharma).
Neutralizing Antibodies and Not-So-Neutralizing Ones
This mechanism brings up another subtlety, though: antibodies can bind to all sorts of things and to all sorts of protein surfaces (although it’s for sure that some proteins set them off a lot more readily than others, and when you yourself have a particular one like this that you’re sensitive to, you have an allergy!) What parts of the coronavirus will they bind to, and with what effect?
The antibodies might stick on to a part of the viral particle that still leaves it operational and infectious, or they might bind in a way that shuts it down right where it stands (probably by covering up the parts, like the Spike protein, that are needed to recognize and infect human cells). Those two classes are called “binding antibodies” and “neutralizing antibodies“, and if you are offered the choice you should probably take the neutralizing ones. Your body will raise both kinds, most likely: several different antibodies will turn out to recognize a pathogen, and your immune system will expand different candidates (a polyclonal response, in the lingo). Like baking powder in biscuits, the neutralizing ones are double-acting: they shut down the virus’s function by binding to it, and at the same time alert other immune system cells to come destroy the particle they’ve latched on to. The plain binding antibodies can still do the latter job, but (as has been mentioned here and many other places), they can also have an extremely unwanted effect, antibody-dependent enhancement, where their binding to the virus actually makes it easier for it to infect human cells. It’s been documented in the closely related MERS coronavirus and many others. You want to avoid that at all costs; safer to go with the neutralizing ones.
A Temporary Vaccine?
You would certainly imagine giving such neutralizing antibodies to people who are sick with the coronavirus, of course. But a particularly interesting possibility is giving them to people who aren’t infected yet, as a preventative. If the antibodies are potent enough and long-lasting enough in circulation, they could provide sufficient protection for weeks or possibly up to several months (and this would take effect immediately upon injection, as opposed to the immunological time lag seen in vaccination). This will take some hard work and close observation to realize, and there could well be variability in the patient population that you’re dosing (i.e., some people might lose such protection faster than others). But until a reliable vaccine is deployed, it’s a very appealing idea – and if the various vaccine efforts underway run into trouble, it might be our only option for a while.
The behavior of mAbs once they get into the bloodstream is an interesting subject if you’re into pharmacokinetics. Antibodies in general are supposed to circulate for quite a while – they’re too large to be filtered out by the kidneys, and the liver doesn’t break them down. One way they can get eliminated is by binding to the surface of a cell and getting taken in by endocytosis and broken down by lysosomes. If you’re giving an antibody to some cell receptor, that’s naturally going to be a slow but steady means of disposal (target-mediated drug disposition or TMDD). And an antibody to a pathogen like the coronavirus will be depleted in the same way as it binds to its target – how much it goes down and how much protection will be left afterwards are things that I’m just not sure anyone knows yet. There’s another effect that you have to watch out for: anti-therapeutic antibodies (ATA), which are – yep – antibodies that get raised against the antibodies you’ve giving to the patient, should they get recognized as foreign substances. That, naturally, can increase over time and with repeated exposure.
Monoclonal antibodies
Convalescent plasma has the whole suite of a person’s antibodies in it, of course. Some of them are neutralizing, some of them aren’t, and a bunch more are antibodies to all sorts of things that the donor might have encountered during their life so far, mixed in with the usual heap of random antibodies that just wander around for years looking for something to bind to. What, though, if you could reach in and select out the ones that are just what you want, and only administer those? Now we have (at last) climbed up to the topic of monoclonal antibodies (mAbs): one lineage only, hand-picked. So how do you find these things, and how do you scale them up to be real-world therapies?
Both those questions have had massive amounts of money and brainpower thrown at them even before this epidemic, which is very good news indeed. If you look at the list of best-selling drugs in the US these days, you will see the list is well-stocked with mAbs. Back when I started in the industry, these things were just barely moving out of the fairly-crazy-idea category. The first mAb approved as a drug (1986, for kidney transplant rejection) was a mouse antibody, which is not ideal. Technology improved, first to make chimeric part-mouse-part-human antibodies like Rituxumab (approved in 1997, and a big success), and then to make fully human ones (panitumumab was the first of those approved, in 2006).
Now in these coronavirus days, there’s a huge push to apply the mAb current technologies (here’s an excellent article at Technology Review about this). There are several platforms used now, and I’ll go through some of the major ones individually:
Regeneron: For example, Regeneron (in Tarrytown, NY) has what they call their “Velocimab” technique, which as I understand it basically uses mice that have been engineered to have a humanized immune system. Exposing these animals to a pathogen or antigen of interest (multiple times) gives you human or nearly human antibodies without having to find recovered human patients to extract them from. As that article details, this technique worked to produce a mAb therapy against the Ebola virus, which really helped change the course of the epidemic. Regeneron is heading towards trials with a combination of their two best antibodies, and working to increase production (more on that issue below).
AbCellera/Lilly: Then you have AbCellera, a Vancouver company that has a technique for isolating individual B cells from patient plasma into single nanofluidic compartments. The cells continue to produce antibodies inside these things, to levels that can then be used for assays. They claim to be able to run through huge numbers of candidates very quickly, and Eli Lilly is convinced. They’ve signed a development deal with AbCellera and have narrowed the company’s list of antibody candidates still further. Lilly’s CEO stated in their recent earnings call that the current top candidate shows “potent neutralization” of the virus, and that it’s now being scaled up in GMP manufacturing for clinical trials. They plan to file an IND in May and go into human patients in June. Meanwhile AbCellera is continuing to screen patient plasma samples for more candidates.
Vir/GSK: Vir Biotechnology (San Francisco) has been working with Biogen, among others, and in April they signed a big development deal with GlaxoSmithKline, one that also involves GSK’s vaccine efforts. They have two antibody candidates from what I can see (VIR-7831 and VIR-7832), and they seem to be bringing a couple of variations to the field that could help with the “temporary vaccine” approach. I believe that one of these is selected to be particularly long-lasting in the bloodstream, while another has been chosen because of its ability to set off a notably robust and long-lasting immune response. They’ve also signed a deal with Samsung Biologics for scale-up starting this fall.
AstraZeneca: AZ also has expertise in developing therapeutic antibodies, and they’ve announced that they’re working in the area. Among their antibody sources are candidates from the Vanderbilt Vaccine Center. That got some headlines because VU listed one of their sources of funding for the effort as Dolly Parton, who has indeed made some large donations to the university over the years, and this work in particular. AZ is working with the Pandemic Prevention Platform (P3) at DARPA (as is AbCellera, above), which is hoping that sufficiently long-lasting and potent neutralizing antibodies could be given preventatively, as mentioned earlier in this post.
Amgen/Adaptive: Amgen has a huge amount of expertise in the antibody field, and in April they signed an agreement with Adaptive Biotechnologies (in Seattle) for work on the coronavirus. Adaptive has also said that they’re hoping to develop preventative antibodies. From what I can see, though, there have been no recent announcements, so it will be interesting to see what’s happening when details finally surface.
Manufacturing mAbs
Like any other large proteins, you would rather have cells make these things for you than synthesize them yourself. The Nobel-winning breakthrough, many years ago, was the idea of fusing a plasma cell (the immune cells that are the actual source of antibody production) with a tumor cell to make a “hybridoma” immortal antibody-producer line. You can do it in several other kinds of cells as well, if you engineer the sequence in as you do for protein production in general. Like any cell-culture production system, this has to be watched carefully. You want to make sure that the production levels stay within range (if it drops, that’s a sign of trouble, such as some sort of infection in your bioreactor), and that the impurity profile is pretty constant (so that your carefully-optimized purification procedures give you the same product every time), and so on. Obviously, these problems are solvable and are solved every time a new monoclonal antibody is brought to market, but they are each their own problem and need careful optimization. You’ll notice that the companies working in this area are all already thinking about manufacturing and scale-up, as well they should. Clearly we’re going to end up with a “good enough” process with rapid deployment rather than something that’s been as tuned-up as the existing mAb production lines. But that might well be good enough indeed!