How does penicillin do its bacteria killin’? And how did a couple superhero female crystallographers help us harness its power to defeat those villains? If asked to name a scientist associated with the antibiotic penicillin, most people would probably say “Alexander Fleming” – and indeed, he played a key role in recognizing its antibiotic potential when his bacterial plate got contaminated by a mold that secreted something that prevented bacteria from taking hold. But if you ask a structural biologist, they might answer Dorothy Hodgkins and (hopefully) Barbara Low. We have those heroines to thank for figuring out the structure of that mold’s molecular weapon, making possible the penicillin derivatives including ampicillin, which doctors (MD kind) use to kill harmful bacteria and doctors (PhD kind) and pre-doctors and pre-pre-doctors use to “select for” bacteria containing things we want in molecular cloning.
Over the past couple of weeks, I’ve introduced you to some “structural biology” techniques including x-ray crystallography and cryo-electron microscopy (cryo-EM) that help us “look” at molecules close up – hopefully at the atomic scale, where we can make out individual carbons, hydrogens, nitrogens, oxygens, etc. Structural biology, a subfield of biology that seeks to connect molecular form and function (e.g. spoon shape/scoop soup), can seem kinda arcane, so I often find myself in a position where I need to explain. And it just so happens that a great example of the power of structural biology to help us understand and manipulate biological phenomena comes from one of the first “complex molecules” to have its structure solved by x-ray crystallography – the solving of the structure of penicillin by Dorothy Crowfoot Hodgkin and her graduate student Barbara Low in 1945.
I will explain more about this discovery and its importance for both biomedicine and crystallography later on in the post, but first I want to tell you about how penicillin works, because, as you’ll hopefully see, its structure plays a huge part!
The cells in our bodies are filled with watery stuff that’s kept separate from the other watery stuff, airy stuff, other cell-y stuff, etc. around it through being surrounded by phospholipid bilayer membranes. These are basically molecular sandwiches where the “bread” is phosphate-containing (and thus negatively-charged and water-loving) “heads” of the individual phospholipid molecules and the “peanut butter” is their fatty tails. Their tail-to-tail arrangement provides a lipid barrier while also giving the water on either side something to hang out with.
This is good enough for our needs, but bacteria are adapted for environments where everyone’s out to kill them, so they shield themselves even further by building cell walls to reinforce their fatty cell membranes. In addition to keeping stuff *out,* the wall’s important for keeping stuff *in.* It’s kinda like a water balloon – if you let too much water in, the pressure of the water pushing against the balloon skin gets too high & the balloon will pop. If this happens in cells we call it lysis. By interfering with the wall-building, beta-lactam antibiotics can cause bacterial cells to burst (good for disease-fighting). And by interfering with the interfering, we can selectively prevent that bursting to help us put bacteria to work (good for biochemistry research).
In molecular cloning, we stick a gene we’re interested in into a circular piece of DNA called a plasmid, then stick that into host cells (often harmless bacteria) to make more copies of that gene &/or the protein it codes for. Later we’ll *want* to lyse these cells to get our protein out. But, while they’re still growing, we want to keep all the protein & DNA-making machinery in the cells so the cells can make the DNA & protein for us. BUT we only want the “right” cells to stay intact – those that actually have our plasmid. We can select for just those cells by putting an antibiotic resistance gene into our plasmid alongside our gene, then spiking the bacteria food w/the corresponding antibiotic so that only cells with the plasmid can survive & grow. More here: http://bit.ly/2IxPOM8
For example, we can design a plasmid to hold our gene & the bla gene, which has the instructions for making Β-lactamase, which protects cells that have it from beta-lactam antibiotics like penicillin (Pen), or more commonly, one of its many derivatives, ampicillin (Amp). Beta-lactam refers to these antibiotics’ “odd” structure – and it was this structure that Hodgkins and Low showed was present in penicillin, paving the way for the design of derivatives with enhanced properties (like better solubility or irresistibility).
So, beta-lactam – what is it? Like many things in chemistry and biochemistry, the answer’s in the name. But also like many things in chemistry, that only helps if you know the jargon of the name. And, for most people, “it’s all Greek to me” – well some of it is *actually* Greek. Greek letters like Beta (β) are commonly used to tell us about the position of “functional groups” – basically most biochemicals are made up of hydrocarbon “skeletons” – chains of carbons bound to hydrogens – and those hydrogens can get “swapped out” for more reactive things, like hydroxyl (-OH) groups – or nitrogens – and the Greek numbering helps us know where such swapping occurs relative to some reference point.
position #1: alpha (α), #2: beta (β), #3: gamma (γ), #4: delta (δ), #5: epsilon (ε), etc.
So the beta in beta-lactam tells us something is going on in the #2 position of something (here relative to a carbonyl (C=O). But what’s that something? What the heck is a lactam?
“Lactam” comes from joining the terms lactone & amide – and it’s a mashup both in name and in structure. A lactone is a cyclic ester – a ring with a C double-bonded to an O and next to another O. An amide is a C double-bonded to an O and next to an N. And a lactam is a when that’s ringified like in a lactone (just with an N instead of an O). The β here refers to how many carbons away from the carbonyl carbon would be if it was split between the carbonyl (C=O) and the nitrogen. So a beta-lactam has nitrogen (N) in a 4-sided ring (uh, wouldn’t that be a square? or, if they aren’t all 90° angles, at least a polygon? – well, we call ‘em all rings). Anyways, N likes being in *some* rings. But it’s less happy to be in other rings. Its relative happiness depends in part on the bond angles & the potential for resonance.
First, the resonance part – The N in imidazole rings, like those in the amino acid (protein letter) Histidine’s side chain, are happy to be there because they can contribute to (and benefit from) something called resonance stabilization (aka electron delocalization). Basically, atoms are made up of 3 main subatomic particles – positively-charged protons & neutral neutrons hang out in a dense central core called the atomic nucleus. And they work to reign in energetic negatively-charged electrons that whizz around them in an “electron cloud.” Within those clouds are places the electrons most like to hang out – called orbitals – and atoms in molecules connect to one another through strong covalent bonds by merging some of their orbitals to share pairs of electrons.
Share one pair of electrons for a single bond, 2 for a stronger, shorter, stabler, double bond, etc. Sometimes molecules have groups of atoms with more than enough electrons for single bonds, but not enough for each pair to have a double bond, so they do a kind of communal sharing of those “extra” electrons (e⁻) that makes all the sharers happy because they get a bond that’s in-between the single and the “deluxe”-ness of a double. But, it comes at a cost – the atoms involved in the sharing have to lie in the same plane in order to get their orbitals to align, so if you want a ring to resonate, it’s gotta be flat – and if you have awkward bond angles, good luck with that…
6-sided ring makes for great bond angles, so you see these a lot in organic (hydrocarbon-based) molecules like the “classic” resonance-stabilized benzene ring. 5-sided rings work too, but 4 (like you have in beta-lactams) is pushing it… The “corners” have to be “sharper” than the bonds would like, leading to RING STRAIN.
And, to make things worse, Pen’s 4-sided B-lactam ring is fused to another, 5-sided thiazolidine ring. This puts the N in a physically awkward position – to get them to join at all you have to make them “pucker” and this pushes the N out of line with the neighboring C=O, so it can’t resonate with it like it would in a “normal” amide (an N next to a carbonyl group (C=O)). So, it can’t get stabilization from e⁻ delocalization
Amide bonds are really common – and we’ve talked about them a lot – but when we’ve discussed them, I’ve usually referred to them as “peptide bonds” which is just a name we give to the amide bonds that connect the amino acid building blocks of proteins to one another. Those protein backbone amide bonds are planar – the N, C, & O are in line – which limits the protein’s flexibility so that, instead of being all spaghetti-noodley, they adopt distinct structures containing backbone motifs like alpha helices & beta strands.
In the early days of biochemistry, some scientists didn’t think that proteins could have such intricate structures – but, by showing that an x-ray diffraction pattern could come from a crystal of the protein pepsin in 1934, J. D. Bernal & Dorothy Crowfoot Hodgkin showed that proteins have orderly structures or else they wouldn’t have been able to give a clear pattern. While they got a distinct pattern, they couldn’t interpret it at the time. They just get some rough dimensions that didn’t tell them all that much about pepsin in particular, but it opened the door for protein crystallography. I’m going to quote the ending to their paper because I love it…
“At this stage, such ideas are merely speculative, but now that a crystalline protein has been made to give X-ray photographs, it is clear that we have the means of checking them and, by examining the structure of all crystalline proteins, arriving at far more detailed conclusions about protein structure than previous physical or chemical methods have been able to give.” https://www.nature.com/articles/133794b0.pdf
But back to our antibiotic story (more on that historical story at the end).
So, the resonance in the amides in the protein backbone are really important – in addition to physically stiffening the protein backbone, they chemically protect the protein chain because the resonance means that *these* amides are NOT very reactive, which is good or else our proteins would get broken apart easily!
BUT the non-planarity of the “weird” amide in a Β-lactam prevents this resonance. Combine this w/ring strain & the amide in the lactam *is* reactive. Why does all this matter? It gives beta-lactams a way to interfere w/bacterial cell wall synthesis. And it gives us an “Achille’s heel” to target Amp w/antibiotics! Here’s why…
We started off by talking about how bacteria have those strong walls – well, they have to build them. And they build them by linking sugar chains w/amino acid linkers. If you put a lot of amino acids together you get a protein. But you can also link together just a few to get a peptide. And you can link them to sugars to get peptidoglycans. It’s like an expansion pack for your sugar LEGO set. It gives you more options.
Bacteria use these peptidoglycans (aka murein) to build their cell walls. Their sugar (glycan) part is a polysaccharide (long chain of sugars) made up of alternating N-acetylglucosamine (NAG aka GlcNAc) & N-acetylmuramic acid (NAM aka MurNAC) & the peptide part’s a short peptide (typically 4-5 amino acids long, sometimes with a “branch”). Different types of bacteria can have different peptide sequences, but they share the same NAG-NAM-NAM-NAM sugar part. The peptides can bind each other to “crosslink” the sugar strands & make a strong multi-layer cell wall. BUT they need help doing this…
A transpeptidase coordinates this linking. It’s an enzyme meaning that it gets *used* but NOT *used up* – enzymes coordinate reactions & might get transiently modified in the process, BUT then when it’s finished it’s back to its GO state – any changes that were made are “erased”
The transpeptidase is one of those enzymes that gets transiently modified along the way – how it works is that the transpeptidase has a hydroxyl (-OH) group sticking out from a Serine (Ser) amino acid in its active site (the place where the action occurs). This acts as a nucleophile (something that has “extra” e⁻ it wants to share) to latch onto one chain’s peptide. This “holds it still” so that the end of another peptide from a neighboring chain can grab it. This works like a “pass-off” – In the middle of the process, the enzyme *has* been modified (it’s bound to the 1st peptide). BUT it then hands that peptide over to the 2nd & is freed again. More on such nucleophilic substitution reactions here: http://bit.ly/2kyxh9T
A beta-lactam antibiotic like Amp looks like the end of the peptides, so it “tricks” the transpeptidase into binding it similarly to how the peptide would bind. BUT when Amp binds the transpeptidase, it DOES NOT get “passed off.” Instead it gets stuck because it forms a “permanent” bond w/it so it can’t get back to “GO” & can’t do its job of strengthening the cell wall. So the balloon weakens & pops, killing the cell
Because human cells don’t have cell walls (just a fatty cell membrane), cell walls are a great target for antibiotics – we don’t have to worry about them disrupting our cell walls since we don’t have any to disrupt! Another way Amp’s attack is bacteria-specific takes advantage of the bacteria-specific nature of those peptide ends.
Stereochemistry refers to the relative 3-D orientation of bonded atoms. Amino acids can be in the “L” form or the “D” form – the same atoms are bonded together, but their 3-D orientation or “handedness” is different. More here: http://bit.ly/2YftOMP
Normally (in proteins) amino acids are in the “L” form. BUT the end amino acids in these peptide linkers are a pair of “D” alanines (D-Ala-D-Ala). This protects the bacteria from normal protease enzymes (protein/peptide chewers) because, like a right hand & a left-handed baseball glove, they don’t fit. BUT the D-Ala-D-Ala *does* fit the active site of the transpeptidase. AND so does Pen because, as we know from its x-ray structure, it mimics this “unusual” amino acid pair
Transpeptidase attacks the 1st of these D-Ala & kicks out the last one as a “leaving group” (a carbon can only form a limited number of bonds so it has to let something go to make new friends). But when it attacks Pen, the leaving group can’t leave – it’s physically tethered there because cutting a ring once only “un-ring-ifies” it, so you still have a chain. And it’s not going anywhere – so you’ve irreversibly inhibited the enzyme.
Cells w/the bla-containing plasmid are protected from this because Β-lactamase inactivates Pen by attacking its Β-lactam ring & popping it open. In covalent bonds like those linking the atoms in the ring, neighboring atoms bond by sharing electrons (e⁻). On its own, neither atom has as many e⁻ as it wants so they share. BUT because the ring’s under strain (kinda like a compressed spring), it’s eager to pop open if there’s something to give it the e⁻ it needs. The C of the carbonyl (C=O) is electrophilic (it *wants* e⁻). This is a perfect match for the nucleophilic active site residues of Β-lactamase. So it attacks -> the ring breaks open -> no longer can bind the transpeptidase -> transpeptidase goes about its business as usual.
The transpeptidase is aka Penicillin Binding Protein (PBP) and, just like it binds Pen, it binds ampicillin. Amp is actually a modified version of penicillin. Penicillin can be modified in different ways to change its properties – just make sure you keep that beta-lactam part intact. By changing the other chemical groups you can optimize antibiotics to work better on different types of bacteria and/or avoid B-lactamase inhibition. Other penicillin derivatives include methicillin, oxacillin, amoxicillin, & ticarcillin. And other Β-lactam antibiotics include cephalosporins (Ceph), monobactams, carbapenems
Note: A problem with bla/Amp as opposed to other antibiotic resistance gene/antibiotic pairs is that B-lactamase gets secreted – the cells ship some out to destroy Amp in their food before it even gets to them. In addition to protecting itself, this protects the cells around it, some of which might not have your plasmid. This can lead to growth of plasmid-less cells. So if you’re growing on a plate these will show up as satellite colonies (clusters of colonies popping up around the good colonies). more here: http://bit.ly/2tcIAVT
You might have heard of “Gram negative” & “Gram positive” bacteria. “Gram” is a type of stain discovered by a guy named Gram & it takes advantage of peptidoglycans to help classify bacteria. Bacteria that have strong, thick, cell walls (coming from up to 40 layers of peptidoglycan) retain the stain so show up dark purple & are “Gram POSITIVE.” An example is Staph aureus. Gram NEGATIVE bacteria have weaker cell walls with only a couple peptidoglycan layers (but they have other sources of additional protection including an “outer membrane”) so the stain leaks out and the cells appear lighter. An example is E. coli. Penicillin works better on Gram positive bacteria which depend more on peptidoglycan & don’t have an outer membrane making it harder for the drug to get to where that peptidoglycan’s at.
So, I hope you can now see the importance of structural information – by knowing what Penicillin looks like, we can better understand how bacteria build their walls, how Pen can prevent this wall building, and where we can and can’t tweak Pen’s basic structure to make more powerful versions. So how do we know what it looks like? The motivation behind today’s post – One of the first atomic resolution x-ray crystallography structures!
It’s really hard to look at really small things – likes atoms separated by tiny distances. You can’t use visible light microscopes because the wavelengths of visible light are thousands of times longer than bond lengths and light’s only useful for resolving (telling apart) things that are separated by distances about half a wavelength or more. X-rays have short enough wavelengths, but short wavelengths come from having high energy – and this makes it so that we can’t focus them with lenses. So x-ray crystallography, while able to provide super valuable information gets tricky. Really tricky. And it’s only possible because of contributions by Dorothy Crowfoot Hodgkin and other early pioneers.
Much more complicated situation grossly simplified it works like this: you get molecules to crystallize (arrange themselves into an orderly 3D lattice) -> beam x-rays at them -> x-rays get scattered by the molecules -> scattered x-rays interfere with one another, some “cancelling out” while others strengthen one another depending on their relative phases (where in their peak-trough-peak-trough… cycle they are) -> these “diffracted” x-rays hit a detector, leaving a pattern of spots called a diffraction pattern -> you work backwards from those spots to figure out where they scattered from.
Even for expert crystallographers these days, it isn’t easy. Even if they have really powerful computer software helping them out. But in the early days of x-ray crystallography, they were working with pen and paper (and wire – they built physical models)
And speaking of models – when you see a crystallographic “structure” what you’re usually seeing is an atomic model showing the position of the various atoms in the molecule. They’re sometimes represented as balls for atoms and sticks for bonds, but this can get “cluttered” for big proteins, so other times they can be “cartoonized” into representations like ribbon diagrams (which you have Jane Richardson to thank for) that emphasize important backbone features.
However you draw it, these aren’t what the x-rays give you directly. Because the x-rays get scattered by the electrons in the atoms (thanks to their negative charge interacting with the ELECTROmagnetic radiation). And the electrons are “on the outskirts” of the atoms – so what you get when you work backwards from the diffraction pattern is the location of the electrons, not the atomic nuclei. And it’s not like you get a nice Battleship game-like readout of – you have an electron at this coordinate, and an electron at this coordinate. Instead, since electrons are constantly whizzing around in their electron clouds, and there are a lot a lot of them, what you get is a meshy thing called an electron density map. Our computers show it as a 3D mesh that we can rotate around and stuff. But Hodgkins was working way before those times. So what she had to do was literally draw (with her sister’s help) “topography lines” like in a mountain map, on lots of pieces of transparency paper that she stuck one on top of another. In the pics I show some of these which are housed in Museum of History of Science, University of Oxford.
A lot of lot of work – so it isn’t surprising that something “small” like penicillin (which “only” has 41 atoms) would be the first whole molecular structure determined by x-ray crystallography. These days, when scientists are routinely solving structures of big ‘ole proteins, this “little” molecule may seem “simple” – but this was crystallography at ints infancy, before computer help. At the time it was the largest molecule ever solved by x-ray crystallography. So how’d it come about?
Firstly – the motivation for trying in the first place – Fleming discovered its antibiotic properties in 1928 but couldn’t make enough of it for medical use. But the onset of WWII led to scientists putting some more serious effort into it. They figured out how to make enough of it to give soldiers – and it worked. But they still didn’t know what it was or what it looked like. In 1943, scientists figured out it had a sulfur atom. But they still didn’t know what it looked like. There were 2 main ideas about its structure – one camp, led by British chemist (and another future Nobelist) Robert Robinson thought there were 2 5-membered rings connected by a single bond (a thiazolidine-oxazolone structure). The other camp, led by Edward P. Abraham and Ernst B. Chain at the University of Oxford and Robert Burns Woodward at Harvard, thought it had the right idea – a 4-membered β-lactam ring stuck to a 5-membered ring. They needed a crystal structure to settle the debate – and for that they needed crystallographers that were first-rate.
So, in 1945, Dorothy Crowfoot Hodgkin solved the structure with the help of her graduate student, Barbara Low. and its discovery came at a time it was greatly needed. It was in the midst of World War II, when battlefield wounds and infections were rampant, and knowing the structure helped chemists develop modified versions of penicillin to treat a wider range of infections. (the structure was solved in May, but wasn’t published until December because it was considered secret wartime work). Even “penicillin” has variants – Hodgkin & Low solved the structure of penicillin G, AKA penicillin II, or benzylpenicillin.
Barbara Low went on to become a big-wig crystallographer in her own right, discovering among many things, the “pi helix” – a special type of shape (conformation) that parts of proteins often form (for the geeky among us, this is where N-H groups in the protein backbone hydrogen-bond with C=O groups 5 letters upstream (instead of the “usual” 4 you find in the more common alpha helix). She died January 10, 2019, at the age of 98, and you can learn more about her here: http://cshlwise.org/wise-wednesdays/barbara-low/
And stay tuned for a WiSE Wednesday profile of Dorothy Hodgkin, who went on to solve the structure of vitamin B12 (for which she won the 1964 Nobel Prize) and later the hormone insulin.
more on x-ray crystallography: http://bit.ly/2QASc8h