In the previous two parts I've described how cell biologists (and scientists in related fields) began to uncover the causes of cancer. Today I'll wrap things up with a recent discovery that goes full circle. But first lets have a recap and an expansion on some key points.
I started this series of posts by describing the Warburg effect. This was first described by Otto Warburg about 100 years ago and led to the golden age of research into metabolism. Here's a summary of the principle as described in one of the papers that I'll be covering today:
Otto Warburg noted that tumour cells, unlike their normal counterparts, use aerobic glycolysis with reduced mitochondrial oxidative phosphorylation for glucose metabolism, and proposed that this was an early and essentially irreversible step that ultimately led to tumorigenesis
Stimulated by this finding scientists intensely studied the bodies metabolic pathways. We learned quite a bit about how our cells made and consumed large macromolecules, but we gained little insight into the nature of cancer.
As the years went by we did learn something but from other approaches.
1) From microscopy, we learned how cells divided. Microtubules emanating from two points formed a spindle that served to organize the duplicated chromosomes and to pull these apart during cell division. Most of the first chemotherapeutic agents targeted microtubules. These drugs are still used today but have the side effect that they kill any cell that divides quickly. So not only are rapidly dividing cancer cells nuked but also hair follicle cells, the lining of the gut and the progenitors of immune cells.
2) From studies in yeast, we learned that the steps of cell division are tightly regulated by what is known as "the cell cycle". At every stage during replication the cell must ascertain whether the previous stage was successfully completed. If so, a group of signals are activated and another group of signals are destroyed and as a result the cell advances to the next stage. Groups of proteins would be made and destroyed at every step - these proteins were known as the cyclins for obvious reasons. A second byproduct of this research is that we learned how proteins are degraded. Proteins are targeted for destruction by being linked to a small, highly conserved protein called ubiquitin. These ubiquitinated proteins were recognized by the trash compactor of the cell called the proteosome.
3) From viruses we learned that cells are full of signalling molecules that formed vast and complicated protein interaction networks in cells. These signalling cascades allowed cells to communicate with eachother and to evaluate the nature of their immediate surroundings. Based on these signals cells would decide whether to live, die, multiply, chill-out, differentiate, remain stationary or migrate. If genes that encode proliferation signals sustain mutations that made the proteins overactive, the cells become more cancer like (or transformed). These genes are called proto-oncogenes. Other genes encode factors that trigger cell death when cells go haywire. If these tumor suppressor genes are inactivated by mutation that also makes your cells more cancerous. A current model o=is the two hit hypothesis where cancer is caused by a combination of mutations each coming in succession. This topic was covered in part I.
4) Recently we learned from a whole interesting line of experiments that signals that sense cell size and metabolic state also contribute to cancer. This ancient signalling pathway, called the TOR pathway is conserved in most nucleated cells. It's components include some of the most important proto-oncogenes and tumor suppressors. One of the major outputs of this pathway is the regulation of how the ribosome functions. It turns out that by regulating how the ribosome translates different mRNAs, TOR signalling can upregulate the expression of a whole slew of proto-oncogenes. In some way this makes some logical sense. If the metabolic state of the organism is good then it will want to grow (make more pro-proliferative proteins) if the metabolic state of the organism is bad then it must prepare for bad times (make more stress resistance proteins). This idea was explored in part II. Recently, Susan Lindquist has explored the stress idea further - it turns out tat since many tumor cells downregulate anti-stress signals and make lots of mutated overactive enzymes that don't fold well. It looks like stressing animals through either heat or through chemical treatment selectively kills some tumors and not normal tissue. (We have been informed that many drugs are being tested in various clinical trials.)
5) We also learned about a whole stew of other cellular/organismal processes that are important for the generation of cancer cells and tumor growth. I won't go into them but just list some:
- Dr. Judah Folkman's work on vascularization (getting a blood supply to the growing tumor)
- Metastasis or the ability of cells to migrate, degrade the surrounding matrix and invade foreign tissue
- Immortalization of cells by renewing their telomeres
- The connection between tumors and stem cells
- DNA damage, chromosome instability ...
Well there was an inkling form TOR signaling that it was somehow important. But for many years most researchers thought that the Warburg effect itself was just a reflection of the fact that tumors were poorly vascularized and did not get a high supply of oxygen. To survive cancer cells simply got their energy by converting sugars to lactate (i.e. non-oxidative glycolysis). This view has now changed drastically.
I'll start off this story with Erich Eigenbrodt. You see despite all the fuss with the cell cycle and all these other topics described above, Dr Eigenbrodt decided to follow up the now ignored Warburg hypothesis. He first observed that many glycolytic enzymes were upregulated in most cancer cell lines. You may think, no big deal, the cancer cells just adapted to low oxygen levels, but in 1981 he made a strange discovery. He found that cancer cells did not express the normal form of pyruvate kinase, but instead had the embryonic version of this enzyme. Pyruvate kinase is a key enzyme responsible for shunting the products of glycolysis into the citrate acid cycle.
Dr Eigenbrodt's idea was that this alteration deeply affected the metabolic state of the cell that had deep impacts on how the cell used glucose. Pyruvate afterall is a key metabolite (click here to see how). This resetting of the metabolic state was key in allowing the cells to fulfill their jobs or to help meet their needs. Thus adipose cells make lots of fat and thus need a lipogenic metabolic state and express a particular pyruvate kinase. Liver cells make sugar and need their metabolism in a gluconeogenic state and thus express a different pyruvate kinase. Embryonic and cancer cells need to make nucleic acids (DNA) and thus have a metabolism in the nucleogenic state. The theory was that each of these states required particular forms of this key enzyme that acted as a valve that either shunted sugar into the energy production pathway or towards other metabolic pathways such as lipid, glycogen or nucleic acid production.
Eventually Dr. Eigenbrodt pattented the testing of urine and stool for embryonic pyruvate kinase as a cancer diagnostic. Sadly Dr. Eigenbrodt died in 2004 at the age of 55. His work remains largely unknown outside the field.
Last month the whole topic was revived with a pair of papers in Nature.
Remember how cancer is a disease of intracellular signals? And that signals take the form of phosphorylation events? For example protein A adds a phosphate to protein B and thus changing the later's activity. Well often protein C will come along and now bind to the phosphorylated form of protein B.
Well the Cantley lab were looking for new factors that could bind to proteins whose tyrosines were phosphorylated. One of these factors that they discovered was ... the embryonic form of pyruvate kinase (PKM2). Strangely, they found that the embryonic form of this protein, but not the adult form (PKM1) had this property. Moreover the pocket that binds to phospho-proteins also binds an activator of the pyruvate kinase, a protein called FBP. It turns out that the binding of phopho-proteins helps to dislodge the FBP protein from PKM2 and thus inactivate this form of pyruvate kinase. Since protein phosphorylation, especially on tyrosine side chains, is massively upregulated in cancer cells, this result was a strongly implied that the whole energy production pathway is inhibited in highly proliferative cells. This was inhibition of energy production was also stimulated by the fact that the PKM2 form of pyruvate kinase is less active than the adult enzyme.
So it looks like cancer cells and embryonic cells have overall lower pyruvate kinase activity. This activity now becomes inhibited by pro-proliferative signals. Sure enough the authors demonstrate that PKM1 expression, but not PKM2, reduces the tumoroigenicity of cancer cells in mice.
So what is PKM2 doing? And what is the significance of it's inhibition by proteins containing phospho-tyrosines?
No changes in adenine nucleotide levels or the ATP/ADP ratios were observed in M2- versus M2KE-expressing cells, suggesting that this cannot account for the defect in cell proliferation observed in the M2KE-expressing cells. However, acute inhibition of PKM2 activity in proliferating cells by tyrosine kinase signalling may result in a temporary build-up of upstream glycolytic intermediates that can be used by the cell as precursors for fatty acid and nucleic acid synthesis, which could provide an advantage to PKM2-expressing cells for proliferation. Consistent with this model, we observed a 25% increase in the incorporation of metabolites from 14C-glucose into lipids on pervanadate treatment of PKM2-expressing cells.
We have realy come full circle. I started off this series of posts with Otto Warburg. This researcher thought that metabolism would be the key to understanding cancer. Then we found out that cancer was cause by mutations in genes involved in signal transduction. The products of these genes are used b cells to interpret cell-to-cell communication. Cells told eachother when to divide, when to move, etc. and these signalling molecules carried out those instructions. Some of these mutations overactivated genes whose normal job was to interpret pro-proliferative signals, and other mutations inhibited the products of genes that acted to kill cells and turn off cell division. The first group of genes are called proton-oncogenes, the second group are named tumor suppressor genes. Next we discovered that some of these signaling molecules interpreted the metabolic state of the organism and the size of cells. These signalling molecules, which are part of the TOR signaling pathway, regulate ribosome production and function. The more ribosomes you made, the faster a cell could duplicate all the proteins needed for its daughter cells. TOR signaling also regulated whether a subset of mRNAs were translated by the ribosome, mRNAs encoding pro-proliferative signals. Finally we rediscovered that the metabolism of cancer cells are very different. Cancer cells are set up in such a way as they convert most of the glucose not into energy but into fatty acids and nucleotides, the components of new cells. This is accomplished by tuning down the energy production pathway an making energy production sensitive to pro-proliferative signals (i.e. phospho-tyrosines.)
OK I'll be cleaning up these entries. If you spot any errors or typos let me know.
Heather R. Christofk, Matthew G. Vander Heiden, Ning Wu, John M. Asara & Lewis C. Cantley
Pyruvate kinase M2 is a phosphotyrosine-binding protein p181
Nature (08) 452:181-187 doi:10.1038/nature06667
Heather R. Christofk, Matthew G. Vander Heiden, Marian H. Harris, Arvind Ramanathan, Robert E. Gerszten, Ru Wei, Mark D. Fleming, Stuart L. Schreiber & Lewis C. Cantley
The M2 splice isoform of pyruvate kinase is important for cancer metabolism and tumour growth
Nature (08) 452:230-234 doi:10.1038/nature06734
I loved the M2 isoform paper and it is so sad that I don't know the short hairpin RNA methodology behind, want to learn it. Anyway, Alex, good stuff the Warburg hypothesis constantly keeps me excited. Did you talk about the role of DCA (dichloroacetate) in inhibiting pyruvate dehydrogenase kinase? Also there is the recent nice review: Mitochondria in cancer cells: what is so special about them? Trends Cell Biol. 2008 Feb 22
Fabulous job with this series Alex, thanks for doing it, lots of interesting stuff to consider here for us neurobiologists!
Just so you know, your link to part II is heading to part I rather than the TOR part of the series.
Thanks for that info. To be honest I'm not that up to date on PK enzymology (was DCA mentioned in the papers?)
Corrections done. If I can finish my protein prep on time, I'll spend some time tonight cleaning up any sloppy sentences.
Very nice. You should submit to the cancer research blog carnival...
Great work Alex, I think that having the space to include proper historical perspective is a great advantage blogs have over papers. It provides a nice record of some of the real background of these big papers. I read this for our journal club and I have a couple of comments of stuff that was unclear, so I thought would pass along to see if you had any ideas.
In the cancer metabolism paper, they show more pyruvate, lactate and TCA intermediates in the M2 lines than in the M1 lines, which sort of makes no sense, since they are all downstream of the (less active) PK.
They also point out a pretty large reduction in FBP (the activator) in the M2 lines. Its possible that the reduced FBP alone could result in blocked pyruvate kinase activity in these lines (independent of the phosphotyrosines).
The last thing is that Eigenbrodt showed that upregulation of tyrosine phosphorylation results in reduced pyruvate kinase activity, the question was just whether it was tyrosine phosphorylation of PK or of something else (ref 5 in the pTyr paper).
That's it and I agree, you should submit this to the Cancer research carnival thingy
Well according to Eigenbrodt all these metabolic genes are differentially expressed in cancer cells - it could be that these transformed cells have reach a different expression profile then the typical cell and thus explaining the differences in metabolite levels. Any systems biologist out there want to take up the challenge?
As for FBP - yes it looks like several factors may contribute to the lowered PK activity. Although their experiments with the siRNAs and the re-expressions indicate that the PK isoform itself may have alot to do with it. But who knows - there may be potential fedback loops (such as PKM2 activity lowers FBP expression).
As for pTyr activity - thanks for that. I'll try to read the Eigenbrodt paper. I must add that pTyr activity goes sky-high in any rapidly dividing cell line. If you immuno-stain for pTyr you see that the highest concentration of pTyr proteins is in focal adhesions. Translation: there's a lot of changes in signalling going on. I think that the Cantley lab has shown that pTyr affects M2 directly as oppose to indirectly. I'll have to read how far Eigenbrodt got. Hopefully his contributions will now be appreciated.
you made a tremendous work in summarizing the metabolic hypothesis of cancer.I would like one small bricks to your edifice.It is the fact that the toxicity of tobacco smoke is because of a well known but little studied poison of cell respiration: carbon dioxide.(Carbon dioxide inhalation causes pulmonary inflammation.Abolhassani M, Guais A, Chaumet-Riffaud P, Sasco AJ, Schwartz L.Am J Physiol Lung Cell Mol Physiol. 2009 Apr;296(4):L657-65.)