Lab Rat: Protein Factory

A nod to four ZX Spectrum games. Millionaire (John Hunt, Incentive Software, 1984) had you writing and selling games from your bedroom with £500: you shared 20 points between programming, presentation, original ideas and player interest, watched the sales each month, and Honest Harry offered deals that were either bargains or trouble. Cookie (Ultimate Play the Game, 1983) had Charlie the Chef knocking runaway ingredients into a mixing bowl while a dustbin monster threw rubbish in. Here you’re a new PhD student: the 20 points go to the bench stages you’d like help with, the month ends with a report, your corner of a shared bench grows into the best bay in the lab, and the dealers (Al Iquot and family) are Honest Harry’s cousins. The medium is the mixing bowl. And as in Silent Service (Sid Meier, MicroProse, 1985), the submarine game behind Beamline Commander, you move between the lab’s rooms (bench, office, catalogue, library, kitchen) like a captain between the stations of a submarine. The PI run adds Football Manager (Kevin Toms, Addictive Games, 1982): a squad with ratings and energy, changes at the end of each season, buying players and balancing the books. Here the squad is your lab, and nobody is sold or fired.

Finish the PhD (crystals delivered, on any level) and the postdoc opens: a better-equipped lab at a better university (a top institute after a brilliant PhD), a three-year contract and a membrane protein. Its finale is a lectureship interview: crystals and a paper of your own get you in front of the panel (on Medium, after two postdoc proteins; on Hard, three), and the right answers make you Professor Rat. Every finished run stays on your record; another PhD, or the postdoc, starts with the tricks you learnt and the books you read, never the kit. Finish a postdoc and the PI run opens: five years with your own lab at an institute whose name sounds oddly familiar, a start-up package that depends on how the postdoc went (perhaps as head of department), and a team to build. Each person has nine strengths (creativity, new ideas, lab skills, analysis, writing, getting things done, teaching, teamwork, communication), each with a ceiling of its own (nobody becomes a superhero), plus energy, wellbeing, focus and a bond with the lab; you hire from adverts or tempt rising stars from rivals’ labs, put people on projects that suit their strengths, hold regular 1:1s (without them people drift; with them they stay, even when a rival offers more), pair PhD students with mentors whose personalities suit theirs, make promises you’d better keep, talk pay at renewal, read the lab manager’s monthly report, send people on courses, and cheer when they leave for bigger jobs. Rick Ombinant sells “guaranteed” grant writers and cheap maintenance contracts, and Phil D. Post sends students and postdocs at a bargain salary, unseen. Grants pay the salaries, invited talks become keynotes, and better institutes come bidding for you. Each year the institute’s board sets goals (papers, grants, deliveries to the beamlines, a happy team), and at the end of the five years a review panel looks at them: a good enough grade (and a paper) keeps your lab; too few goals met, or three months deep in debt, and it closes. Ideas for new projects come from your team, from people you meet, from old lab members and from new machines, and some come with a rival racing you to the paper. Projects need the right kit (without it the team borrows, slowly), and a confocal microscope or a binding measurement can lift a paper to a top journal. Each week you choose a plan: the bench with the team (a hands-on moment of your own), a week of writing, meetings, or the week’s card (a seminar, a grant panel, the finance office). Once in five years something big goes wrong (a funding cut, a flood, a selfish new head of department, a leak), and sometimes something big goes right. After every talk, as a postdoc or a PI, there’s question time (an easy, a medium and a hard question), then the conference dinner, where a late night at the bar may teach you a secret, cost you one, or end in romance. You choose who your lab rat is: she, he or they.

As in a real lab, a PhD student doesn’t hold the big grants: you help your supervisor write theirs (a share comes to your project) and a postdoc with an equipment bid (the machine it brings is yours to use too), and now and then another student asks for a hand, which can end in a paper with your name on it. Labs are people too: a low patch, a lab crush, a row over bench space, a mistake nobody saw. Talking to someone, asking nicely and owning up all help, and the library has a (silly) book for each.

A PhD student takes posters to meetings (a good one can win the poster prize); a postdoc gives short talks, writes more papers, joins the postdoc association, and can pay a technician for a couple of months when time is shorter than money. Small things run out (precast gels, good competent cells, autoinduction supplements), so keep an eye on the shelf. The harder levels are worth more: reputation comes faster and the final score counts more. A postdoc can also try cryo-EM grids, and learn why a small membrane protein is a job for crystals. Papers gather citations month by month (Millionaire’s sales), a press release can land a newspaper or TV interview, and open days and careers events cost only time.

The science

• The protein. UgpG, from Sphingomonas elodea, the bacterium that makes gellan gum. It joins glucose-1-phosphate to UTP to make UDP-glucose, a building block for the gum. It was David’s PhD protein: cloning, expression, purification and crystallisation (Acta Crystallographica F, 2006) and its structure (Journal of Bacteriology, 2007).
• Cloning. A PCR copies the gene, restriction enzymes cut it and a vector, and the new plasmid goes into E. coli. Colonies growing on ampicillin carry it, except the tiny satellite colonies around them, which live where the big colony has broken the antibiotic down. The plasmid comes back out with a miniprep (once with phenol and chloroform) and is sequenced.
• Growing and expression. LB medium (tryptone, yeast extract, salt) goes into the autoclave; after that, germs can only come in from outside. The growth curve has a lag, a log phase and a plateau, and IPTG switches the gene on through the lac operon. Made too fast, the protein misfolds into inclusion bodies.
• Purification. Columns sort proteins by charge (ion exchange), by how greasy they are (hydrophobic interaction, one of the kinds of chromatography), by size (size exclusion) or by a His-tag on nickel. The Bradford assay shows where protein is, any protein; SDS-PAGE shows which band is yours.
• Before crystals. Dynamic light scattering spots clumps, and a thermal shift assay finds the buffer the protein likes.
• Crystals. Protein crystallisation by vapour diffusion: screens of conditions, hits, optimisation, then freezing with a cryoprotectant for the beamline. Selenomethionine in place of methionine adds a signal that helps solve the structure, and a careful series of concentrations in a matched buffer makes a sample for SAXS.
• The postdoc’s protein. AlgE, the pore in the outer membrane of Pseudomonas aeruginosa that lets out alginate for its biofilms: a beta barrel in the bacterial outer membrane. It’s a membrane protein, greasy where it meets the lipids, and both routes in the game are published ones. Refolding: AlgE made as inclusion bodies, unfolded in urea and refolded into detergent, then crystallised in detergent (the method: Whitney et al., Acta Crystallographica F, 2009; the structure: Whitney et al., PNAS, 2011). Membranes: cells grown cool, membranes pelleted in an ultracentrifuge, AlgE taken out with the mild detergent octyl glucoside, swapped for LDAO on a copper column, then crystallised in the lipidic cubic phase, first in monoolein and best in a shorter lipid, 7.8 MAG (Tan et al., Acta Crystallographica D, 2014; the method: Caffrey and Cherezov, Nature Protocols, 2009). The gene goes in by TOPO cloning. David helped collect and process the data for the 2014 structures, refine them and solve the structure: see AlgE.
• The postdoc’s other protein. DgkA, a diacylglycerol kinase in the inner membrane of E. coli: 121 amino acids, three chains working together, putting a phosphate from ATP onto a lipid. Its route in the game is the published one: the gene made to order in the vector pTrcHisB, stabilising mutations (the thermostable Δ4 and the designed Δ7), the cells’ lysate solubilised in the detergent Empigen BB and swapped on a nickel column into decyl maltoside, then crystals in the lipidic cubic phase with 7.8 MAG at 4 °C, which took weeks to grow (Li et al., Nature, 2013). Molecular replacement failed, so the phases came from selenomethionine. A later soak with a non-hydrolysable ATP analogue and zinc caught it mid-reaction (Li et al., Nature Communications, 2015). David was part of the team behind both papers, on the crystallography side: see DgkA.
• The postdoc’s receptors. Two G-protein-coupled receptors (GPCRs), the family behind the 2012 Nobel Prize in Chemistry for Robert Lefkowitz and Brian Kobilka: seven helices across the membrane, a drug pocket that changes shape, and floppy loops that make crystals hard. Both are made in insect cells (Sf9) infected with a baculovirus carrying the gene. The β2 adrenergic receptor follows the 2011 paper: lysozyme from the T4 phage put into its floppy third loop, the change H93C so a designed agonist holds on for good, an antibody column and an alprenolol column, then tiny crystals in monoolein with cholesterol (Rosenbaum et al., Nature, 2011); with the inverse agonist carazolol instead, you repeat an earlier structure. David is one of the 2011 paper’s authors: see β2AR. The turkey β1 adrenergic receptor is made sturdier first: an alanine scan of 318 mutants in E. coli, each heated and tested for binding a drug, and the best six combined, 21 °C more stable (Serrano-Vega et al., PNAS, 2008); then insect cells and crystals in detergent by vapour diffusion (Warne et al., Nature, 2008), the work of Chris Tate’s and Gebhard Schertler’s groups at the MRC Laboratory of Molecular Biology.
• The PI’s projects. KcsA, the bacterial potassium channel whose structure (Doyle et al., Science, 1998) helped win Roderick MacKinnon the 2003 Nobel Prize in Chemistry; IL-33 held by a gut worm’s protein, HpARI2, made in two hosts and studied by crystals and SAXS (Jamwal et al., Nature Communications, 2024); PTX3, the flagship, by crystals, cryo-EM and SAXS (Shah et al., Matrix Biology, 2025); and a fragment screen, hundreds of soaked crystals for I04.
• The big projects, later in a PI’s five years: venoms from Australia’s dangerous animals, among them a funnel-web spider’s peptide that protects the brain after a stroke in rats (Chassagnon et al., PNAS, 2017); a faster PETase, the plastic-eating enzyme studied on Diamond’s MX beamlines: its structure on I23, I03 and I04 (Austin et al., PNAS, 2018), the PETase–MHETase “super-enzyme” on I03 (Knott et al., PNAS, 2020) and HotPETase, evolved to work at 70 °C, on I03, I04 and I04-1 (Bell et al., Nature Catalysis, 2022); enzymes for oil spills; a drug for the deadly henipaviruses; making antibiotics work again against β-lactamases and pumps; cancer and flu drugs designed from structures, like venetoclax and oseltamivir (Tamiflu); and enzymes that do chemistry in water at room temperature.
• The PI’s kit. Like a membrane protein lab’s: an insect cell suite, an LCP robot and LCP-FRAP (bleach a spot of fluorescent protein in the cubic phase: if it moves back, the protein can crystallise there), FSEC, a plunge freezer and a grid robot for cryo-EM, cell-free expression for proteins that kill their host; and for the paper, a confocal microscope (where the protein goes in the cell), SPR (how tightly it binds) and mass photometry (how many copies stick together).
• Opening the plates. Crystals in the lipidic cubic phase grow sealed between two sheets of glass. To harvest them, two circles are scored in the cover glass round the well, the glass between them broken away and the disc lifted off with tweezers; the crystals are fished out of the sticky mesophase with a small loop and go straight into liquid nitrogen, with no cryoprotectant (Li, Boland, Aragão, Walsh and Caffrey, JoVE, 2012, a video protocol).

Where the game cuts corners

• The tag. You start with the native protein and three columns, and earn the His-tag later. The real UgpG had a His-tag from the start: a nickel column and a desalting column, and Bradford to measure it.
• Restriction sites. The real UgpG was cloned with BamHI and HindIII; the game’s reasons for not using EcoRI or NotI are made up for the game.
• UgpG’s charge. The game’s hint about which ion-exchange column to use comes from a pI made up for the game, not a measured one.
• Prices and times. Kit prices, grant sums and most times are round game numbers. Real kit is dearer, and real grants take longer.
• Polarised light. In the game salt crystals light up and protein crystals hardly do. Real life is less tidy: many protein crystals glow under crossed polarisers too, and cubic salt crystals don’t.
• Optimising. Matching a method to the kind of crystal works far more neatly here than at a real bench, where crystallographers mostly try everything and keep the lab notebook close.
• Autoinduction. Real autoinduction media (F. William Studier’s recipes) are richer than LB with three sugars on top; the game keeps LB and adds the sugars as one jar.
• Grids. Real grid screening looks at hundreds of squares and then the holes inside them; the game shows one atlas and four picks. Small proteins have been solved by cryo-EM since, with tricks (fusions, antibody fragments).
• Refolding. As in the 2009 paper, AlgE goes into 8 M urea and is added to a detergent buffer drop by drop, but the paper also ran a nickel column in urea first, then stirred the drops overnight at 37 °C: real refolding takes a night, not half a minute.
• Detergents. The game offers three and one is right. Real labs screen many, and Triton X-100 has its fans.
• The cubic phase. Real mixing takes about a hundred pushes; the game counts a couple of dozen. Its syringes also warm up much faster than real ones, so that the ice gets a turn. The cold side is real: below about 18 °C monoolein slowly turns waxy, which is one reason DgkA’s cold-room plates used a shorter lipid, 7.8 MAG.
• Opening the plates. Real wells take longer to open than half a minute, and a glass cutter goes blunt after about ten of them.
• DgkA’s genes and time. In the game the genes arrive already in the vector; the paper says they were made to order and cloned into it, without saying by whom. Its crystals took 45 to 60 days in the cold room; the game makes that about six weeks.
• Membranes or inclusion bodies. How much AlgE goes where at each temperature is a game number.
• The receptors. The β2AR paper was a collaboration between several labs, Brian Kobilka’s and Martin Caffrey’s among them; in the game you do every step yourself. The scan has 40 wells rather than 318, and the radioactive drug shows as a glow. The cell counts, the infection and the harvest days are game numbers: the papers’ methods don’t give them. Real Sf9 cells take about a day to double, slower than the game’s.
• The PI’s lab. The institutes, the rival PIs and the team are made up, and their names are jokes. Projects move by weeks of work shared between the people on them, which is a long way from how science really goes, and the grants are far quicker to decide than real ones. The project ideas (a smelly sock’s favourite enzyme, the lab’s immortal sourdough) are mostly for fun, and a machine doesn’t really make a paper twice as famous. Scoops happen, but rarely because of one late night at a conference bar.