Monthly Archives: October 2016

Broad-spectrum biotechnologies and their implications for synthetic biosecurity

2 Replies

We live in a rather pleasant time in history where biotechnology is blossoming, and people in general don’t appear to be using it for weapons. If the rest of human existence can carry on like this, that would be great. In case it doesn’t, we’re going to need back-up strategies.

Here, I investigate some up and coming biological innovations with a lot of potential to help us out here. I kept a guiding question in mind: will biosecurity ever be a solved problem?

If today’s meat humans are ever replaced entirely with uploads or cyborg bodies, biosecurity will be solved then. Up until then, it’s unclear. Parasites have existed since the dawn of life – we’re not aware of any organism that doesn’t have them. When considering engineered diseases and engineered defenses, we’ve left the billions-of-years-old arms race for a newer and faster paced one, and we don’t know where an equilibrium will fall yet. Still, since the arrival of germ theory, our species has found a couple broad-spectrum medicines that have significantly reduced threat from disease: antibiotics and vaccines.

What technologies are emerging now that might fill the same role in the future?

Phage therapy

What it is: Viruses that attack and kill bacteria.

What it works against: Bacteria.

How it works: Bacteriophage are bacteria-specific viruses that have been around since, as far as we can tell, the dawn of life. They occur frequently in nature in enormous variety – it’s estimated that for every bacteria on the planet, there are 10 phages. If you get a concentrated stock of bacteriophage specific to a given bacteria, they will precisely target and eliminate that strain, leaving any other bacteria intact. They’re used therapeutically in humans in several countries, and are extremely safe.

Biosecurity applications: It’s hard to imagine even a cleverly engineered bacteria that’s immune to all phage. Maybe if you engineered a bacteria with novel surface proteins, it wouldn’t have phage for a short window at first, but wait a while, and I’m sure they’ll come. No bacteria in nature, as far as we’re aware, is free of phage. Phage have been doing this for a very, very long time. Phage therapy is not approved for wide use in the US, but has been established as being safe and quite effective. A small dose of phage can have powerful impacts on infection.

Current constraints: Lack of research. Very little current precedent for using phage in the US, although this may change as researchers hunt for alternatives to increasingly obsolete antibiotics.

Choosing the correct phage for therapeutics is something of an art form, and phage therapy tends to work better against some kind of infection than others. Also, bacteria will evolve resistance to specific phages over time – but once that happens, you can just find new phages.

DRACO

What it is: Double RNA Activated Capsase Oligomerizer. An RNA-based drug technology recently invented at MIT.

What it works against: Viruses. (Specifically, double-stranded RNA, single-stranded RNA, and double-stranded DNA (dsRNA, ssRNA, and dsDNA), which is most human viruses.)

How it works: DsDNA, dsRNA, and ssRNA virus-infected cells each produce long sequences of double-stranded RNA at some point while the virus replicates. Human cells make dsRNA occasionally, but it’s quickly cleaved into little handy little chunks by the enzyme Dicer. These short dsRNAs then go about, influencing translation of DNA to RNA in the cell. (Dicer also cuts up incoming long dsRNA from viruses.)

DRACO is a fusion of several proteins that, in concert, goes a step further than Dicer. It has two crucial components:

  • P that recognizes/binds viral sequences on dsRNA
  • P that triggers apoptosis when fused

Biosecurity applications: The viral sequences it recognizes are pretty broad, and presumably, it wouldn’t be hard to generate addition recognition sequences for arbitrary sequences found in any target virus.

Current constraints: Delivering engineered proteins intracellularly is a very new technology. We don’t know how well it works in practice.

DRACO, specifically, is extremely new. It hasn’t actually been tested in humans yet, and may encounter major problems in being scaled up. It may be relatively trivial for viruses to evolve a means of evading DRACO. I’m not sure that it would be trivial for a virus to not use long stretches of dsRNA. It could, however, evolve not to use targeted sequences (less concerning, since new targeting sequences could be used), inactivate some part of the protein (more concerning), or modify its RNA in some way to evade the protein. Even if resistance is unlikely to evolve on its own, it’s possible to engineer resistant viruses.

On a meta level, DRACO’s inventor made headlines when his NIH research grant ran out, and he used a kickstarter to fund his research. Lack of funding could end this research in the cradle. On a more meta level, if other institutions aren’t leaping to fund DRACO research, experts in the field may not see much potential in it.

Programmable RNA vaccines

What it is: RNA-based vaccines that are theoretically creatable from just having the genetic code of a pathogen.

What it works against: Just about anything with protein on its outside (virus, bacteria, parasite, potentially tumors.)

How it works: An RNA sequence is made that codes for some viral, bacterial, or other protein. Once the RNA is inside a cell, the cell translates it and expresses the protein. Since it’s not a standard host protein, the immune system recognizes and attacks it, effectively creating a vaccine for that molecule.

The idea for this technology has been around for 30-odd years, but the MIT team that discovered this were the first to package the RNA in a branched, virus-shaped structure called a dendrimer (which can actually enter and function in the cell.)

Biosecurity applications: Sequencing a pathogen’s genome should be quite cheap and quick once you get a sample of it. An associate professor claims that vaccines could be produced “in only seven days.”

Current constraints: Very new technology. May not actually work in practice like it claims to. Might be expensive to produce a lot of it at once, like you would need for a major outbreak.

Chemical antivirals

What it is: Compounds that are especially effective at destroying viruses at some point in their replication process, and can be taken like other drugs.

What it works against: Viruses

How it works: Conventional antivirals are generally tested and targeted against specific viruses.

The class of drugs called thiazolides, particularly nitazoxanide, is effective against not only a variety of viruses, but a variety of parasites, both helminthic (worms) and protozoan (protists like Cryptosporidum and Giardia.) Thiazolides are effective against bacteria, both gram positive and negative (including tuberculosis and Clostridium difficile). And it’s incredibly safe. This apparent wonderdrug appears to disrupt creation of new viral particles within the infected cell.

There are others, too. For instance, beta-defensin P9 is a promising peptide that appears to be active against a variety of respiratory viruses.

Biosecurity applications: Something that could treat a wide variety of viruses is a powerful tool against possible threats. It doesn’t have to be tailored for a particular virus- you can try it out and go.

Current constraints: Discovery of new antibiotics has slowed down. Antivirals are a newer field, but the same trend may hold true.

Also, using a single compound drastically increases the odds that a virus will evolve resistance. In current antiviral treatments, patients are usually hit with a cocktail of antivirals with different mechanisms of action, to reduce the chance of a virus finding resistance of them.

Space for finding new antivirals seems promising, but they won’t solve viruses any more than antibiotics have solved bacterial infections – which is to say, they might help a lot, but will need careful shepherding and combinations with other tactics to avoid a crisis of resistance. Viruses tend to evolve more quickly than bacteria, so resistance will happen much faster.

Gene drives

What it is: Genetically altering organisms to spread a certain gene ridiculously fast – such as a gene that drives the species to extinction, or renders them unable to carry a certain pathogen.

What it works against: Sexually reproducing organisms, vector-borne diseases (with sexually reproducing vectors.)

How it works: See this video.

Biosecurity applications: Gene drives have been in the news lately, and they’re a very exciting technology – not just for treating some of the most deadly diseases in the world. To see their applications for biosecurity, we have to look beyond standard images of viruses and bacteria. One possible class of bioweapon is a fast-reproducing animal – an insect or even a mouse, possibly genetically altered, which is released into agricultural land as a pest, then decimates food resources and causes famine.

Another is release of pre-infected vectors. This has already been used as a biological weapon, including Japan’s infamous Unit 731, which used hollow shells to disperse fleas carrying the bubonic plague into Chinese villages. Once you have an instance of the pest or vector, you can sequence its genome, create a genetic modification, and insert the modification along with the gene drive sequences. This can either wipe the pest out, or make it unable to carry the disease.

Current constraints: A gene drive hasn’t actually been released into the wild yet. It may be relatively easy for organisms to evolve strategies around the gene drive, or for the gene drive genes to spread somehow. Even once a single gene drive, say, for malaria, has been released, it will probably have been under deep study for safety (both directly on humans, and for not catastrophically altering the environment) in that particular case – the idea of a gene drive released on short notice is, well, a little scary. We’ve never done this before.

Additionally, there’s currently a lot of objection and fears around gene drives in society, and the idea of modifying ecosystems and things that might come into contact with people isn’t popular. Due to the enormous potential good of gene drives, we need to be very careful about avoiding public backlash to them.

Finding the right modification to make an organism unable to carry a pathogen may be complicated and take quite a while.

Gene drives act on the pest’s time, not yours. Depending on the generation time of the organism, it may be quite a while before you can A) grow up enough of the modified organism to productively release, and B), wait while the organism replicates and spreads the modified gene to enough of the population to have an effect.

Therapeutic antibodies

What it is: Concentrated stocks of antibodies similar to the ones produced in your own body, specific to a given pathogen.

What it works against: Most pathogens, some toxins, cancers.

How it works: Antibodies are proteins produced by B-cells as part of the adaptive immune system. Part of the protein attaches to a specific molecule that identifies a virus, bacteria, toxin, etc.. The rest of the molecule acts as a ‘tag’ – showing other cells in the adaptive immune system that the tagged thing needs dealt with (lysed, phagocytosed, disposed of, etc.)

Biosecurity applications: Antibodies can be found and used therapeutically against a huge variety of things. The response is effectively the same as your body’s, reacting as though you’d been vaccinated against the toxin in question, but it can be successfully administered after exposure.

Current constraints: Currently, while therapeutic antibodies are used in a few cases like snake venom and tumors, they’re extremely expensive. Snake antivenom is taken from the blood serum of cows and horses, while more finicky monoclonal therapeutics are grown in tissue culture. Raising entire animals for small amounts of serum is pricey, as are the nutrients used for tissue culture.

One possible answer is engineering bacteria or yeast to produce antibodies. These could grow antibodies faster, cheaper, and more reliably than cell culture. This is under investigation – E. coli doesn’t have the ability to glycosylate proteins correctly, but that can be added in with genetic engineering, and anyways, yeasts can already do that. The promise of cheap antibody therapy is very exciting, and more basic research in cell biology will get us there faster.

The bipartisan model of androgynous gender presentation

2 Replies

[Content warning: Talking about ways that people automatically gender other people. If this is a tough topic for you, be careful. Also, a caveat that I’m talking descriptively, not prescriptively, about people’s unconscious and instant ways of determining gender, and not A) what they might actually think about someone’s gender, and certainly not B) what anyone’s gender actually is.

Nonetheless, if I got anything wildly or offensively inaccurate, please do let me know.]

When you try and figure out a stranger’s gender, you don’t just use one physical trait – you observe a variety of traits, mentally assign them all evidence weights, compare them to any prior beliefs you might have on the situation, and then – usually – your brain spits out a “man!” or “woman!” This is mostly unconscious and happens in under a second.

This is called “Bayesian reasoning” and it’s really cool that your brain does it automatically. Most people have some male, some female, and some neutral signals going on. ‘Long hair’ is usually a female signal, but if it’s paired with a strong jawline, heavy brows, and a low voice on someone who’s 6’5”, you’ll probably settle on ‘male’. Likewise, ‘wearing a suit’ is usually a pretty good male signal, but if the person is wearing makeup and is working at a hotel where everyone is wearing suits, you’re more likely to think ‘female’.

Then there are people with androgynous gender presentations – the people who you look at and your brain stumbles, or else does spit out an answer, but with doubt. (As a cis but not-particularly-gender-conforming woman, this is people around me all the time.) When people are read as ‘androgynous’, I think they’re doing three possible things:

  1. Strong male and female signals. Think a dress and a beard, or a high-pitched voice and being 6’4” and muscular, or wearing a suit and eyeliner. Genderfuck is an aesthetic that goes for this.

Left: Drag queen Conchita Wurst. Right: Game of Thrones character Brienne of Tarth.

2) No gender signals. Not giving gender cues, or trying to fall in the middle of any that exist on a spectrum. I think of this one as usually involving de-emphasized secondary sex characteristics – flat chest, no facial hair – which might also mean a youthful, neotenous look. Or maybe a voice or hips or height or whatever that’s sort of in the middle. Some (but not all!) androgynous models have something like this going on.

Left: Model Natacha S. Right: Zara’s Ungendered fashion line.

Fashion-wise, every now and then a company that rolls out a gender-neutral clothing line is criticized because all the clothing is baggy, formless, and vaguely masculine. (See comments below on why this may be.) I think these bland aesthetics are going for ‘No Signals’ – baggy clothing conceals secondary sex characteristics, the plain colors call to mind sort of a blank slate.

3) Signals for Something Else. For a trait that would normally signal gender, signal something else entirely. Long hair is for women, short hair is for men, but a green mohawk isn’t either of those. You might speak in a high-pitched voice, or a low-pitched voice, or in falsetto with an accent. Men wear pants, women wear dresses, but nobody wears this:

Pictured: I don’t know what these people are signalling, but it’s sure not a binary gender. [New York Fashion Week, 2015.]

What does this imply?

I’m not sure.

I expect that people who do No Signals get less shit from bigots (harassment, violence, weird looks) than people in the other two categories (Mixed Signals or Signaling Something Else.) I would imagine that bigots are more likely to figure that No Signals people are clearly a binary gender that they just can’t read, whereas Mixed Signals people are perceived as intentionally going against the grain.

This is unfortunate, because if you want to be read as androgynous, it’s way easier to just do Mixed Signals than to conceal secondary sex characteristics in order to do No Signals. (Especially if your secondary sex characteristics happen to be more pronounced.) Fortunately, society in general seems to be moving away from ‘instant gender reads are your real gender’, and towards ‘there are lots of different ways to do gender and gender presentation’.

Signaling Something Else people probably also get harassment and weird looks, but possibly more because they’re non-conforming in ways that don’t have to do with gender.

Male Bias in Gender Interpretation

Also! There is a known trend that suggests that people are more likely to read ambiguous traits as male than female. This is probably because ‘male’ is seen as ‘the default’, because culture. See: non-pet animals, objects other than cars and ships. This seems to have originally come from Kessler & McKenna (1978), and has held up in a few studies. I’m not sure if this rule is completely generalizable, but here’s a few things it might imply:

You may actually have to have more feminine traits than masculine ones to hit the Confusion Zone. For gender-associated traits that go on a spectrum – chest size, voice pitch, some metric of facial shape, etc., it might look like this:

graph1

Of course, there are also cases where people think a trait is associated with gender when, really, it’s not. That still might mean something like this:

14646614_10210490978858879_848724627_o

(See also.)

One conclusion I’ve heard drawn from this: This explains why it’s often harder for trans women to get automatically gendered correctly, than for trans men. A trans woman has to conceal or remove a lot of ‘male’ traits to get read as female. Trans men, meanwhile, don’t have to go as far to hit ‘male’.

Even gender distribution world

Let’s say there are 100 gendered traits (wearing a dress or pants, long or short hair, facial hair or no facial hair, etc.) Now let’s imagine a population where everybody in this population has the “male” or “female” version of each trait assigned independently and randomly. If the male-bias principle generalizes, you’re likely to read more than half of these people are male.

Regional differences?

Gender presentation, and thus how you read gender, is deeply rooted in culture! If you see someone in garb from a culture you’re not familiar with, and you can’t tell their gender, it’s quite possible that they’re still doing intentional gender signals – just not in a way you can read.

Even for similar cultures, this might be different. When I was in England, people called me ‘sir’ all the time. This doesn’t happen often in Seattle. I have three theories for why:

  1. People in England have different gendered trait distributions for deciding gender. Maybe in England, just seeing ‘tall’ + ‘short hair’ + ‘wearing a collared shirt’ is enough to tip the scale to ‘man.’
  2. Where I was in England was just more culturally conservative than Seattle, and if I spent more time in, say, small towns in Southern or Midwest US, I’d also be ‘sir’d’ more.
  3. People in England are more likely to say ‘sir’ or ‘m’am’ at all. So if you were to ask a bunch of Seattle and England strangers if I was a man or a woman, the same percent would say ‘man’, but I wouldn’t notice in Seattle.

I think 2 or 3 are more likely, but 1 would be interesting as well.

Post Notes

  • Ben Hoffman pointed out that this maps to classifications for people who don’t consistently vote for a major political party. Mixed Signals people are like swing voters or nonpartisan voters. No Signals people are political moderates or don’t vote at all. Signaling Something Else people are, like, anarchists. Or Pirate Party members.
  • The Bayesian Evidence model of gender identification doesn’t only apply when the result is inconclusive – often your brain will, say, match someone as ‘man’, but also observe that they’re doing some non-masculine things.

(The first thing to consider in this case is that your brain may be wrong, and they may not actually be a man at all.)

  • Anyways, what gender people are and what they signal to the world is more complex than an instantaneous read, and this is an important distinction. For instance, even when people look at me and think ‘woman’, they can tell that I’m not doing standard femininity either.
  • If you’re trying to cultivate auto-gendering people less often, I suspect that training your subconscious to quickly separate whatever traits from gender would be useful. Finding efficient ways to do this is left as an exercise to the reader.
  • It’s obviously possible to train your brain to look at someone and mentally assign them a gender other than the instantaneous response. I’ve also heard stories of people looking at people and automatically going “nonbinary”. I suspect that if you grew up in binary-gendered society, as so many of us tragically did, this is a thing you developed later in life. Maybe you learned this as a possible answer to the “confusion on gendering androgynous people” brain-state.