Life of a Lab Rat

I can clone! Yay!

After a long (too bloody long) time geeking around with perl, preparing posters and talks, and doing some hardcore microarray data mangling, I went into the lab this morning and filled my ice bucket. I got out my cut vectors and inserts and found the water and ligase buffer. It's all on ice, thawing, so for the first time in about six weeks I'm going to do a real experiment (keeping cells alive doesn't count).

It feels good. I'm all set. I'm ready, and now I'm going...

to have a cup of coffee. Yeah.

So you spend months doing knockdowns and testing antibodies and optimizing the Western protocol::

and then your (very talented, admittedly) student comes along, does her first real-life Western with your antibody and protocol and it's bloody perfect.

There is no justice, there's only me.

You can spot a protein chemist/molecular biologist who is also a parent by his or her reaction to the smell of bugs grown in minimal media.

Your non-parent will probably say "Ew, stinky" with a side-order of an attempt at description. The parent, on the other hand, will immediately file the smell alongside breast-fed baby's nappy. The nappies that look like they have mustard or sesame seeds in them.

Aren't you glad this weblog isn't scratch 'n' sniff?

If you're a scientist then you've probably heard of Nature Precedings,

a place for researchers to share pre-publication research, unpublished manuscripts, presentations, posters, white papers, technical papers, supplementary findings, and other scientific documents. Submissions are screened by our professional curation team for relevance and quality, but are not subjected to peer review. We welcome high-quality contributions from biology, medicine (except clinical trials), chemistry and the earth sciences.

Turns out, from a comment made by Hilary Spencer over at Jennifer Rohn's weblog, that pre-publication research includes stuff that didn't work.

I am eyeing my thesis, curiously and thoughtfully.

The hot water in the Cage was turned off for what I'm guessing was the Annual Flow Test.

I must have been the first person to use the gents after the water came back on, because when I turned on the taps to wash my hands there was an airlock: the water hit the handbasin with tremendous force and splashed all over my trousers.

Two hours later I went to thaw some cells. I washed my hands with Hibiscrub. When I turned the water on there was an airlock: the water hit the handbasin with tremendous force and splashed all over my trousers.

(Never mind that I know other people had used the hood since the water came back on...)

And why was I thawing cells on a Friday? Because the bloody RNA prep from the KO experiment failed, and I need more cells:

Fortunately it's the lab advance starting this evening. And breathe two three four...

It might not look much to you, but that rather messy blot up there helped to tell me three things this afternoon:

1. The original antibody is crap and does not reliably and uniquely detect my protein. I suspected this already
2. The detection reagents give a better signal to noise ratio when you leave them for over an hour before exposing to film (data not shown. Hopefully I'll get a nicer picture of that tomorrow)
3. The knockouts are working again.

It worked. It bloody well worked!

How is it that one little gel can make a post-doc so happy?

BOUNCE

(Shame the Westerns are still a bit crap, but they seem to be corroborating what's going on in lanes 3, 9 and 10 of the lower strip).

My babies! My poor, starved, over-crowded babies!

I got in this morning, went to Stores, saw the stack of autoclave bags and realized that in the heat of microarray analysis and genomic mining yesterday I'd forgotten to look after my cell culture.

They think I don't love them anymore. Sniff

I tell you the truth, there is more rejoicing over one colony that is positive than over 19 that aren't.

That's all very pretty, but this

is a result.

Yay!

Tidying up the office, and came across this gel:

Whatever it meant, I didn't seem happy about it.

Once, just once, it would be nice to have a simple project.

Image credit

I have been using siRNA to knock down the expression of my favourite protein (FP), in an attempt to figure out just what the heck it does. And the experiments have been looking pretty good — the message (RNA) goes 'phut' and the protein itself disappears within two days. We're assaying what affect this has on the cells by running exon microarrays (because we have reason to believe that FP is a splicing factor), which are expensive but potentially very powerful.

The first experiment worked well; we saw lots of changes when compared with an 'irrelevant' control (a lamin knockout) and got really excited. And then I did some n >1 and some more controls, and it all went a bit squiffy.

This is a graph showing the expression of lamin exons, on 12 microarrays representing triplicates of four conditions:

The three experiments on the very left are arrays of the lamin knockout experiment; the other 9 are (in triplicate, left to right) two different controls and the FP knockout. Note how the signals disappear almost completely from the lamin KO, but are pretty constant across the others. This is good, and right, and hopeful.

But then we look at the FP exons, and Oddness Occurs:

Look at the three right-most arrays (numbers 10, 11, 12). Number 12 disappears completely, which is what we'd hope for and expect. Ten is a bit meh (although it does go down a bit), but there is a huge uptick in four or 5 of the exons in experiment 11 (look at the orange line, for example).

I spent a goodly time renormalizing the data and trying to see if there was any systematic error in that particular experiment, but all the control exons (and indeed, the lamins) look good. Puzzling. Thinking about the experiment this morning, and what to do about it, I happened to read Jenny's ponderings on scepticism. And it struck me that what I am seeing is tantalizingly similar to that which Li and co. describe.

In other words, ignoring for the moment the obvious question of why supposedly the same cell line would behave completely differently in two presumably identical experiments, the squiffiness could be interesting.

This is Science.

You try. You push, you fight, you struggle. You take tiny, baby steps, and all the time you feel like you're running to stand still. Everyone else seems to be successful, and your plugging away only draws attention to to the void that waits where your next paper should be.

But still you try, hoping against hope, long ago going through the place where any sane person would have given up because deep in your heart you know that this is the only thing you can do; the only thing worth doing.

And maybe you look at the papers in the field and realize that the experiments that set you off on this wild goose chase were complete crap anyway, and the mechanistic interpretation, if there is one, is deeply, fundamentally flawed. You present a poster with your ideas, which, despite — or maybe because of — your lack of results, is very pretty and even enjoys a brief moment of glory on display alongside the prize-winners of this year.

Others need convincing, so you perform more experiments, and with tragic inevitability any data you generate are variable, standards don't and negative controls aren't. Little hints here and there suggest you might not be completely crazed, but you wonder if you've given your boss any reason at all to believe in you.

Then one afternoon you sit down to look at some very preliminary data: incomplete, waiting on the proper controls and still shy of the experimental nirvana that comes from n = 3; and you really don't know what you should be doing with this program but you fight it because by God it's not in you to give up, and you realize that you're reading the wrong strand of the chromosome but when you finally get the numbers to match six bases SHOUT at you from the Ensemble web site and you echo the shout to the office as you realize that here, indeed, is an Answer.

All your heartache and disappointments are forgotten in that sweetest of brief moments. You are the only person in the entire world to know what you do now.

You savour the exultation while your pulse recovers, then you grab your scribbled notes and a pencil and hotfoot it to the boss's office, where you try to keep the shaking out of your voice while you explain what you've just found. His reaction stuns you, as he leaps from his chair and calls in other members of the lab who have a vested interest in this project and whose own work has just been vindicated. You have to explain the result three times while phrases like "this is the best result" and "this is so fucking cool" are bandied around carelessly. The uninitiated look on, somewhat bemused.

Then comes the inquest, the 'whatifs' and the 'yeahbuts' and you have to explain how your model appears be right, pending further investigations and appeals and peer review. It's dark outside, it's late and you still need to set up a PCR before you can leave.

Nonetheless, they can not take it away from you:

For a Day, you were King.

This has got to be in the running for the coolest cloning experiment ever.

Last Tuesday a grad student in the reciprocal space cadet lab, let's call him Fu Manchu, asked me if I had any GFP. 'GFP' expands to 'green fluorescent protein', which is a protein that is green, and, um, fluorescent. It's naturally found in a certain jellyfish and glows green when you shine a certain wavelength of blue light at it. This is really useful for studying where proteins go inside cells or whole animals, because you can join the DNA that codes for GFP to the bit of DNA that codes for the protein you're interested in, and put that into your experimental model. Much like I did here, there and elsewhere, in fact.

But Fu Manchu didn't want to localize a protein in a cell (because as I hinted above, he's a scatter-brain and wouldn't recognize a whole cell if it mitosed); he wanted the protein itself to use as a crowding agent in some unspeakable experiment. I had to tell him that no, it wasn't the sort of thing I kept in my fridge, but I did have the DNA that codes for GFP in a plasmid vector and he was quite welcome to take it out and make the protein in a system of his choice. It would take a few days to design and make the primers and do the PCR but it would be simple enough.

I went to have a cup of tea and realized that actually, GFP in a protein expression vector, that is, in plasmid that we use to make vast quantities of purified protein rather than in a different sort of plasmid that we use to make relatively small amounts of protein in situ, might be a useful reagent and I was a damned fool for leaving a very similar reagent in Cambridge and not getting it shipped over with my other useful bits and bobs (and I couldn't get it shipped in the time frame that Fu Manchu wanted it). So I trooped back to the computer and had a look at restriction sites, and realized that if I was only semi-clever I could cut the GFP gene out of the one plasmid and into the other sort.

So on Wednesday I set up a couple of enzyme digests, purified the appropriate bits of linearized DNA and set up the reaction to stick that gene for GFP into this expression vector, where it would, if I did nothing else, make GFP and GFP alone (and left a couple of restriction sites so that I could, at a future date, drop in the gene for some other protein after the gene for GFP and make green some-other-protein).

To retrieve the construct made in this way I had to transform some bacteria — in other words, persuade them to take up this new construct and propagate it. The bacteria can be persuaded to do this because the vector you make has a gene for resistance to some antibiotic on it, and you select only the bugs that contain the plasmid you want by growing the bugs on plates that contain that antibiotic. Thursday morning, then, I hoped to see colonies on my plates; each colony having grown from a single antibiotic-resistant bacterium.

That's the theory: in practice you always get 'background'; bacteria that grow because they have taken up some plasmid that doesn't have the gene you really want, or one of a myriad other excuses. And you then have to make DNA from several of these colonies and cut them up in special ways and all sorts of tedious stuff. Knowing this, I chose the bacteria that I would transform to be the sort that cannot help but make protein, even when you don't want them to. And I reasoned that the bacteria that had the right plasmid, that is with the GFP in it, would be green.

On Thursday morning then, I took my plates upstairs — which indeed had colonies — and asked to borrow Tiffany Case's fluorescent microscope. I told her what the plan was, and we looked at the plates together, and this is what the butler saw:

That's two photographs of the plate. On the left, normal light, and you can see all the colonies. On the right, we've illuminated with blue light and the only colonies you can see are the ones that are making GFP, and are therefore glowing green. Ignore the black pen marks — they're just from when I counted colonies. Here's a closer look:

.

See those two really bright suckers? They're making GFP from the DNA I gave them. That third colony isn't, and therefore glows not. So on Thursday afternoon, forty-eight hours after conceiving the experiment, I was able to hand Fu Manchu a fresh plate containing bugs that make the protein he was after.

I got to play on the confocal and multiphoton microscopes yesterday. I was even allowed to drive the multiphoton myself, which has cool green laser tubes and all sorts of funky stuff and is basically more fun than a postdoc should be allowed to have (so don't tell my boss, 'mkay?).

It was basically a driving lesson for me, to figure out what the equipment can do, how to use it, and only secondarily to examine my cells to see if we can figure out what the fruitbat they are doing. So this picture has no scientific merit, but is very pretty:

I said to the Younger Pawn (aged seven and a half!) this morning "Come and see a koala bear!" and showed her the picture.

"Koalas aren't bears!" she stroppily informed me.

"Oh, what are they?"

"Marsupials" she said, definitely.

Sigh. Trust kids to take the fun out of science.

Why cell cultures are like children

1. You get emotionally attached to them
   
2. When it's all going right, they are rewarding and fulfilling
   
3. When things go wrong it breaks your heart
   
4. Sometimes it's difficult to tell if the weird shit is normal
   
5. It helps to keep them clean
   
6. It's difficult to keep them clean
   
7. It's distressing when they're sick
   
8. You can't just throw them away when they're sick
   
9. It's difficult to trust anyone else to look after them
   
10. Looking after someone else's is more stressful than looking after your own
    
11. It's impossible to get them to clean their own room
    
12. You need to look after them at inconvenient times (weekends, holidays...)
    
13. They're devilishly expensive
    

Here's one for Joolya.

I've been growing some cells that express a splicing factor (probably. But that's a different story) tagged with green fluorescent protein (GFP). For the non-biologists, GFP is a protein that naturally fluoresces green in response to certain wavelengths of light. This is useful for sub-cellular localization of proteins, because it means you can fuse your favourite protein gene to the gene for GFP, stick that into cells, and see where your protein lives. It's incredibly useful, and pretty to boot.

Now, my protein lives in the nucleus, so when it's tagged with GFP the nuclei of my cultured cells glow green. And after the cells have been growing a while, all sorts of weird things happen, but mainly I get multinucleate cells: The nucleus of a cell divides but then something goes wrong and you don't get daughter cells. Most cells have one nucleus each, but these guys (gals?) are forming syncytia, although the term is usually used to describe the fusion of existing cells, rather than what I think is happening here.

Ooh, controversy.

I got my Honours student ('W') to perform the experiment yesterday, and this is just in:

The above is a photograph of an agarose gel, which is used to separate different fragments of DNA. It has been stained with ethidium bromide, which glows under ultraviolet light when bound to DNA.

As good a place as any to record the "freeze-squeeze" method of DNA extraction from gels:

1. Run gel as usual
2. Isolate band using long-wave UV
3. Freeze lump of agarose containing band of interest (upwards of 20 minutes)
4. Place agarose lump in top of SpinX-type spin column, and spin 2 minutes
5. Mush remaining agarose with 200 µl TE and spin 5 minutes
6. Phenol-chloroform extract the flowthrough (use 1:1 volumes)
7. Ethanol precipitate.
8. Resuspend in appropriate volume of water/whatever buffer.

Scene just now in the lab: Your humble correspondent setting up an RT-PCR experiment. The conversation went something like this,

Black Queen: Are you going to use the new block ?

BK: No. I have a programme on the Eppendorf.

BQ: But it's intuitive!

BK: You mean you can give your samples to a grad student and let them get on with it?

BQ: <That Look>

Doing PCR reactions? Then stop using 200 µM dNTPs and use half that. Seriously.

It's to do with the K_m and other nasty kinetic things but trust me, you'll get better data (and fewer errors).

I spent a long time this afternoon thinking about the model upon which the set of experiments I have been doing for too many months is based. And I have come to the conclusion that the model is pants.

Which on the one hand is a shame, because I feel that I've wasted half the past year. On the other hand it does explain why the experiments are not showing what we expected — that is of course why we do experiments, to test models. The thing is that I can not make the model yield the result that was expected of it, even theoretically. That is, even if the model is correct it can not give the answer we were looking for. The best I can do is make half of the pre-messenger RNA transcripts behave and the remainder have ruddy great lengths of intron in them.

That is somewhat annoying, especially seeing as the effect I have been trying to reproduce was published a few years ago and appears to be accepted as true. That published work explains (but does not justify) why I had not previously assessed the model so critically.

I walked into the office the other day to grab a notebook, with a pile of shiny, new tissue culture plastics under my arm.

CK accused me of hoarding the flasks and wotnot like a squirrel. I countered by saying that I intended to be the only person capable of carrying out cell culture in a post-apocalyptic world, cornering the market as it were. Think of me as a pre-alcoholic Mel Gibson.

I'm back!

And my cells are fine, mad props to P for looking after them, my lovely, lovely babies.

Here's a rather weird picture that I captured earlier today, of what appears to be a dividing cell in one of my transfection cloning plates:

The green is the nucleus (expressing my splicing factor) and the purply colour is the surrounding cellular stuff (the colour is just for contrast. Grey is boring).

But look at the shape of that nucleus! And the nobbly bits on the end! Just what is going on here? It's similar to the multinucleation I've been seeing (and saw lots of this morning), but stretched out rather than spread around. Maybe it is a precursor form?

Answers, please, on the back of an email to the usual place.

I'm on leave this week.

This (time off) is a problem for scientists, because as I've intimated previously, experiments do not stop when we do. It is especially bad for cell biologists to go away for more than a couple of days, because our cell cultures need feeding and looking after. This means getting the cells to a stage where they will not be much bother for the absence period and finding someone you trust to check on them and change their media. Think getting someone in to feed the hamster or goldfish and you'll understand.

I was particularly badly burned in this respect early on in my career. As a final year grad student with young Kiwi bride I went to New Zealand for a month. I got my stable, monoclonal transfections growing up and left detailed instructions with the post-doc for their care. When I got back I discovered that he had managed to contaminate at least half of them while trying to freeze them down for long-term storage. That is a lesson I have not forgotten.

The difference between an experienced scientist and a novice in the art is not necessarily that the student makes mistakes where the experienced does not. Rather, the experienced anticipates those mistakes and plans for them; s/he knows when they will occur and how to design an experiment such that the mistake does not matter.

So, to illustrate, I always run asymmetric protein and nucleic acid gels; if I drop them I can tell which way around they are supposed to be. Or, to take a completely random example, when running western blots of multiple samples from two experimental cell types I will make one of the gels subtly different, just in case - oh I don't know - maybe I absent-mindedly ignore cough the labels cough on the soaking trays cough or something like that, then I will, tomorrow, still be able to interpret the experiment.

This entry is mainly for the benefit of my niece.

Nicole, this is the kind of thing we can see through our really cool microscopes. If you like, I can tell you all about it at Christmas, and when you come and visit us I'll take you to see one for real. Click on the small picture for a bigger version.

One of my fellow Rats commented that she liked doing cell culture, because what with the laminar flow hood and everything it felt like she was doing Real Science. I know what she means, so here's another photograph of lab furniture; the cell culture, or laminar flow, hood.

There is an arrangement of fans and filters with the general idea that any nasty stuff inside the hood stays there and does not infect the operator, and everything on the outside stays there and does not infect whatever it is that the operator is working on. Obviously this theory breaks down a little bit because you have to put your hands and flasks of cells, solutions, pipettes and whatnot inside in order to actually do anything. We get around the problems this causes by liberally spraying 70% ethanol over the surface and anything that gets put inside (including hands, which are usually also washed with Hibiscrub first). There is also a reasonably high intensity ultraviolet lamp inside that gets turned on when the hood is not in use, to make life unpleasant for any nasties that do manage to find a way in.

And yes, you do get to feel like a real scientist doing real science, and most of us doing cell culture form an irrational emotional attachment to the cell line(s) we happen to be working with at the moment. We care for, feed and nuture our cells and it can be quite distressing (not just because of the time wasted) when, not if, you go to the incubator one morning and find your precious cultures swarming with bacteria, or fungi, or strange beings from the Planet Claire.

Five months' work:

In the photograph above, 'a' is the control experiment and 'b' shows the effect of Protein de Jour. That 'SF' band (i.e. the white bit) is stronger than the 'LF' band in 'b'. That is telling me that PdJ is changing the levels of splicing (for this particular protein) in the cells I'm working with. The arrowhead in 'c' is pointing out a third band that I can never see in the control, which tells me that something else might be going on, too. That's somewhat exciting.

The Western blot became conclusive. The tantalizing band is the potato acid phosphatase, cross-reacting with the secondary antibody. Frustratingly, the phosphatase is represented by two bands, which threw me. Note to self: Sigma are crap.

I said on Friday that 'Z' does not regulate its own splicing, and the gel I ran this morning supports that conclusion. There is one thing that is stopping me drawing a line under that however, which is to do with the way I did the experiment. I performed what we call a 'transient' transfection. What this means is that I made the cells take up the test DNA, and went ahead and did the analysis on a mixed population: Some of the cells took up a lot of DNA and (hopefully) would have made a lot of protein, other cells took up a middling amount; and some cells just laughed and told me to bugger off.

I should have written this up on Friday evening, but seeing as I was driving to Canberra it would have been inconvenient. I did pack the Queen's laptop, but it wouldn't have worked very well in the sauna. So, this post is a tad late. Apologies.

Friday is meeting day in our department. In the morning I set up a PCR reaction, helped a grad student next door with a cell culture problem, grabbed a coffee and went to our regular two-group lab meeting.

In one of my previous lives I pretended to be a crystallographer.

In other words, I would attempt to persuade concentrated protein solutions to get together and form ordered three-dimensional arrays - crystals - so that I could then shoot X-rays at them. The purpose of this was to determine their structure.

One of the things I was taught when I learned crystallography was to be very, very clean. No dust, human hair, bits of glass or other muck were allowed into the experiment. The most frustrating thing about crystallography is that all proteins are different, and will crystallize (if they crystallize) under different conditions, and there doesn't seem to be any pattern to this at all. The lab's insistence on cleanliness was an attempt to factor out one of the variables in the process. But I soon discovered that this may have been counter-productive. As with a lot of things in research, people disagreed with each other and there was a lot of intuition and opinion without a great deal of solid evidence. I realize that this might come as a surprise to some of you, to those who, perhaps, believe that 'scientists deal with facts'. The truth is that at the frontiers of science we don't know what's going on and we're trying to find out - that's why it's called 'research'. If you want facts, look in a text book (and they all contain mistakes, too).

So as I went on and got more experienced, I began to welcome small amounts of crud in my crystallization experiments. In fact, one recalcitrant protein only ever crystallized once, along what looked like an insect leg. I was never able to repeat that experiment; although I did have gothic fantasies about breeding every different sort of insect I could find and using various bodily insect parts as nucleants. A little too Shelley, perhaps.

I revisited the gel-eating tube question yesterday. I did the experiment in the suspect tube strips and in another brand in parallel.

Guess what? The other brand's reactions were fine, the original strips again ate the gel. How's that for science in action?

The word alone chills the hearts of experienced cell biologists. And when, a couple of months back, someone upstairs was getting strange results from their cell-based assays, the Boss came to me and asked,
'What do you know about mycoplasma?'. The icy black hand of dread gripped me and I spent the rest of the day trying to find a supplier for a mycoplasma detection kit in Australia.

I spent the next two days arguing with various suppliers and AQIS about what I could or could not import and for how much. It seemed that AQIS were upset about the presence of mycoplasma DNA; in effect the small amount of positive control included in the kits I was looking at.

I had 96 PCR samples to run on gels today. It was a big experiment; a transfection time course with four different conditions and three different primer pairs.

I like weekends. Unfortunately science doesn't, and often just half an hour on a Sunday can save a whole day in the week. And that's before you take account of things like cells, which have this habit of growing all day and all of the night.

So it was last weekend, after a pleasant day out with the Queen and Pawns, that I found myself wondering how my transfections were getting on. Now the trains out to Black Castle are pretty good during the week, but a little unpredictable (if not replaced by busses) at other times. The remains of a bottle of wine was also tempting me, and I really didn't think that driving was a good idea. At the same time I was beginning to fret about my cells; were the little blighters growing, was the new batch of FCS good, had I really contaminated a flask. . .?

The Queen was using a 4 litre fermentor to express a certain protein. For some reason, she was using an 'enriched' media recipe from a lab down the corridor. Expression failed, and when I looked at the recipe I saw that the final glucose concentration in the media was in the order of 4%.

This is outrageously high in my opinion: I've expressed various proteins from bugs grown in minimal media (M9, for example) at a maximum of 0.3 % glucose, both for selenomethionine and ¹³C/¹⁵N labelling. When trialling those expressions I did the appropriate titrations and discovered that protein expression falls off at or above 0.5% glucose. It's a case where less is definitely more.