PhD – 21 months in 

Do you remember my optimistic blog post about finding my bearings in the lab after a month of the PhD? I also included pictures of a failed western blot and slightly crushed centrifuge tubes.

Well, twenty months later and I’m still making mistakes. Often they’re new and different mistakes, which could almost be exciting. But today I made the same mistake and lost a lot of plasmid-growing bacteria (bacteria I am using as work horses to produce specific DNA for me) in a centrifuge (which I subsequently cleaned!)…

Photographic evidence attached.

Behrens lab retreat 2016

Imagine spending a weekend in these idyllic surroundings in the Peak District with nothing to do but talk about and discuss science.


The Peak Mermaid Inn – taken at sunrise on November 13th 2016

Well, that’s exactly what we, the Behrens lab, did last weekend. We invited a keynote speaker, Roland Rad, and Dieter Saur’s group from the Technical University of Munich to join us. Each of us gave a talk about the most interesting or exciting aspects of our projects and in between we drank copious amounts of coffee. In the evenings we cooked enough food to feed a small regiment, drank beer, played pool, darts or table football, all punctuated by heated debates about science. Although this wasn’t a relaxing weekend by normal standards, it was motivating and inspiring and a good reminder of why I enjoy being a scientist: a combination of rational and logical thinking, curiosity and the drive to learn new things for their own sake, all shared with people who, by and large, know more than I do and think differently.

Of the talks I just want to highlight one in particular, because my project also uses one of the techniques mentioned. Dieter Saur is a medical doctor and has his own lab group, which studies mainly gastrointestinal diseases, including pancreatic cancer. In a recently published paper (Schönhuber et al, 2014) they describe an experimental system in mice called the “dual recombinase system“. This is a genetic system that allows the study of complex diseases such as cancer. Until recently it was only possible to simultaneously switch on a gene that drives tumour progression and switch off a gene that prevents tumour formation in a cell type or organ of interest (e.g. in the pancreas). Using the dual recombinase system it is possible to make genetic alterations sequentially. For example, in the beginning of a mouse’s development one can activate a potent tumour driver called Ras and delete an important tumour suppressor called p53. And then, once a tumour has formed, one can additionally delete genes that may be important to maintain the established tumour. Alternatively, the dual system also makes it possible to make genetic changes to the normal cells surrounding the (pancreatic) tumour. If all goes well then I will be able to use these tools to conduct experiments like this in the next year or so.

15044797_10154563370871405_2126581266_o zip-line

Oh and admittedly we did have an activity scheduled that was slightly less scientific: we got all geared up and went on a GoApe outing. Secured by a harness and after some rigorous safety instructions we got to fly down zip lines, balance over gaping abysses and jump over the void below.

Lastly, the following week saw Queen Mary University London and Barts host the 11th UK cancer stem cell symposium. There were several interesting talks, including by group leaders at the Crick Institute, but the most unusual talk was given by a philosopher called Lucie Laplane. She did her PhD in philosophy and combined this with a research master’s in stem cell biology. Putting the two fields together she came up with a classification of (cancer) stem cells using definitions and guidelines borrowed from philosophy, applied to biology. [In general, researchers agree that stem cells are cells that can self-renew (i.e. generate new copies of themselves) and can produce differentiated/specialised daughter cells.] The most important point was how to pin down what kind of characteristic “stemness” is or what makes a stem cell a stem cell:


Framework for defining (cancer) stem cells – copied from Lucie Laplane’s talk at the symposium

For instance, in some cases a stem cell might always be a stem cell no matter what the environment is like (i.e. categorical); other stem cells may be dispositional in nature, meaning that they always have the potential to act as a stem cell but only do so in a permissive environment. Alternatively, being a stem cell might not be property of a single cell at all but rather an attribute of an entire organ (i.e. systemic). Laplane argued that the way we define (cancer) stem cells has a huge impact on how we try to treat diseases such as cancer. For example, if cancer stem cells are “systemic” then even the best therapies targeted against these cells will fail because the system/the tumour will make new cancer stem cells from other tumour cells. Hans Clevers, one of the Gods in the stem cell field, wrote a glowing review of the book here.


Laplane, Lucie. Cancer Stem Cells: Philosophy and Therapies. Harvard University Press, 2016.

Schonhuber N, Seidler B, Schuck K, Veltkamp C, Schachtler C, Zukowska M, Eser S, Feyerabend TB, Paul MC, Eser P, Klein S, Lowy AM, Banerjee R, Yang F, Lee C-L, Moding EJ, Kirsch DG, Scheideler A, Alessi DR, Varela I, Bradley A, Kind A, Schnieke AE, Rodewald H-R, Rad R, Schmid RM, Schneider G, Saur D (2014) A next-generation dual-recombinase system for time- and host-specific targeting of pancreatic cancer. Nat Med 20: 1340-1347

A Second Word on Evolution

Life as a PhD student is busy and doesn’t leave much time for other activities, including this blog. So last time, about a month ago, I left you with the question of how the genetic code may have evolved over time.

For decades some scientists have hypothesised that the genetic code evolved by a so-called direct templating mechanism (also known as the stereochemical hypothesis). That is, the strings of ribonucleotides that make up an RNA molecule could physically interact with amino acids, the building blocks of proteins. This interaction would promote the reaction of adjacent amino acids to start forming a longer polypeptide chain. For a review on the different hypotheses see Koonin & Novozhilov (2009).

One of the proponents of the stereochemical hypothesis is Bojan Zagrovic and his research group at the Max F. Perutz Laboratory in Vienna. They have published several papers on this topic and almost a year and a half ago I went to a symposium where Bojan Zagrovic gave a talk on exactly this topic. I wrote about the various presentations I heard there and then several months later a friend I had met during the Cold Spring Harbor Laboratory (CSHL) undergraduate research programme sent me a message saying he had been inspired, by the blog post, to do some research of his own.

In particular, John wanted to investigate whether there was a pattern behind the observed interactions between the amino acids in proteins and the ribonucleotides in RNA. To do this he and Rachel (another student from CSHL) used computational biology approaches to study a large published dataset of protein-RNA complexes. They found that there is a correlation between these physical interactions and the way the genetic code is laid out.

Once these findings had been made they wrote up a draft manuscript, including some figures, which were produced by Grace, a colleague of John’s at Carleton College in the USA. John asked whether I would mind reading the manuscript to give feedback and of course I was happy to do that. We started e-mailing back and forth and decided to extend the computational experiments, and I edited and expanded the text.

The most interesting result was that we could use the knowledge derived solely from the interaction data (blue and red bars) to predict, significantly more accurately than expected by chance (yellow bars), the amino acid sequence of a protein from its mRNA precursor:

Screen Shot 2015-12-07 at 23.00.28

Combining amino acid-nucleobase affinities with mRNA nucleobase content to predict amino acid sequences without universal genetic code. Copied directly from our paper.

In particular, the proteins that form the ribosome – the molecular machine that translates mRNA into protein in modern-day cells – were more accurately predicted than a random protein from our dataset, possibly suggesting that direct interactions between RNA and amino acids led to the formation of the first primitive ribosomes. However, as you can see, the prediction accuracies do not exceed 15% so all results from this paper need to be taken with a pinch of salt; I think the best we can do is say that our results strengthen the stereochemical hypothesis but by no means prove it. [In any case, the scientific method is only good at disproving theories.] Since the journal, Scientific Reports, is an open access journal anyone can read the paper here.

Overall, I am just proud that we managed to publish our work after a long and iterative process, including one revision. All of this was done long-distance via Skype and e-mail. We were all working or studying full-time at the same time and moreover, we did this without the help of a professor/group leader. In fact, none of us even has a PhD (yet).

Lastly, I have noticed a mini-surge in views of my blog posts pertaining to PhD interviews. Clearly the invitations for the next year have been sent out and I hope whoever is reading this is finding it helpful and: good luck!


Cannon JGD, Sherman RM, Wang VMY, Newman GA (2015) Cross-species conservation of complementary amino acid-ribonucleobase interactions and their potential for ribosome-free encoding. Scientific Reports 5: 18054

Hlevnjak M, Zagrovic B (2015) Malleable nature of mRNA-protein compositional complementarity and its functional significance. Nucleic Acids Research 43: 3012-3021

Koonin EV, Novozhilov AS (2009) Origin and evolution of the genetic code: the universal enigma. IUBMB Life 61: 99-111

Polyansky AA, Zagrovic B (2013) Evidence of direct complementary interactions between messenger RNAs and their cognate proteins. Nucleic Acids Research 41: 8434-8443

de Ruiter A, Zagrovic B (2015) Absolute binding-free energies between standard RNA/DNA nucleobases and amino-acid sidechain analogs in different environments. Nucleic Acids Res 43: 708-718

The Joys (?) of Revision #3

Subtitle: boredom. The only good thing is that finishing my project, writing up the dissertation and doing the thesis “defence” (also known as a viva voce exam) are worth 50% of the final grade this year. The dissertation writing itself was actually enjoyable because it’s good to make sure you know what you did and what you didn’t achieve with the experiments. For instance, the in silico melanoma model I constructed from available experimental data could accurately predict that melanoma cells that had become resistant to the widely used drug, vemurafenib, would be more motile and possibly more invasive. However, the model wrongly predicted how (i.e. by which mechanism) the cells had become more motile. All in all not a bad outcome for a 16-week project, I think.

project title page

During the writing process I regularly made use of notes I’d taken during a lecture on scientific writing. Among other things I had to remind myself a) which tenses to use for each section of the thesis (e.g. past tense to describe results and methods, present tense to describe data previously presented in the literature) and b) when to use “compare to” versus “compare with”: the former when comparing things that are different with respect to each other and the latter when comparing things that are similar. Also not to confuse “while” and “whereas”. Never to begin a sentence with a numeral. To use simple sentences, although that is sometimes quite hard when my mind becomes corrupted by German influences.

Only five more days to go until freedom.

In silico versus in vitro #2

There is no adequate excuse for not keeping up with regular blog entries. Arguably, working in the lab from 9 a.m. to 8.30 p.m. – in a mad rush to collect some data before the lab meeting next week – in addition to preparing for seminars and the imminent badminton match against Oxford help to explain the recent scarcity of cell/science-related news. Especially when a statistically significant number of experiments in the lab end up looking like this Western blot on the right (image copied from here):

notpub-three kinds of westernsHowever, on the more positive side of things I have learnt a couple more cell/molecular biology techniques since I started working in the lab several weeks ago. Beforehand I only knew about the theory of these methods and how to interpret results, but not how to actually carry out the experiments:

  • Immunohistochemistry (IHC): this technique allows specific staining of (mouse) tissue sections. One can, for example, compare specific markers of proliferation or cell death in healthy skin versus tumours/melanoma lesions.
  • Quantitative reverse transcription polymerase chain reaction (qRT-PCR): this allows analysis of gene expression (specifically transcription) in tumour samples or cultured cells under different conditions. For example, I have been comparing the expression of certain transcription factors in cells that are sensitive to a melanoma drug versus cells that have acquired resistance to that drug.
  • Cell cycle analysis by fluorescence-activated cell sorting (FACS): lastly, this method is used to quantify the proportion of cells within a population that are actively dividing, non-dividing or dead. To do this, the amount of DNA in each cell is stained using a dye such as propidium iodide, or newly synthesised DNA can be labelled using bromodeoxyuridine.

Now on to the simple task of analysing the data…