Cancer Caught in the Act

Glucose Deprivation Contributes to the Development of KRAS Pathway Mutations in Tumor Cells

Yun, et. al. Science 325: 1555-1559 (2009).

Out of approximately 20,000 human genes, why are the usual suspects (p53, Rb, RAS, BRAF, etc.) the ones that are most commonly mutated in cancer? What selective pressure drives their mutation in so many different types of tumors? In a recent Science paper, Yun, et. al. provide evidence that KRAS and BRAF mutations can arise in tumor cell lines grown in a low glucose environment, and that constitutive KRAS- or BRAF-dependent upregulation of glucose transporter 1 (GLUT1) is specifically required for the survival of mutant cells in such inhospitable conditions. Acquisition of such a trait would be especially useful in a blood-starved, oxygen-deprived tumor, as it would allievate the cells' dependence on oxygen for aerobic ATP production. Alternatively, this could also potentially provide a mechanism for a phenomenon observed in many cancer cells known as the Warburg Effect, where cells rely on increased glucose uptake and glycolysis to generate ATP even if oxygen is available ("aerobic glycolysis"). This is an inefficient way to generate ATP, though it has been proposed that this benefits rapidly proliferating cancer cells because glucose can instead go towards building the amino acids, fatty acids, and nucleotides required for new cells.

KRAS and BRAF mutations are never found in the same tumor cell, so Yun, et. al. hypothesized that a common set of genes would be deregulated following mutation of either gene, pointing to the signaling pathways that conspire to give mutant cells a selective growth advantage. Yun, et. al. first compare the expression profiles of several tumor cell lines with mutant KRAS or BRAF to that of otherwise identical cell lines whose mutant KRAS or BRAF allele has been corrected. (Tumor cell lines have frequently acquired many mutations, but the authors try to negate any potential effects of non-KRAS or BRAF mutations by using multiple pairs of cell lines that should be exactly the same except for a single allele of a single gene, or isogenic.) The only gene of interest commonly upregulated in all the mutant cell lines was GLUT1, or glucose transporter 1. The authors show that GLUT1 overexpression is specific to tumor cell lines with mutant KRAS or BRAF, and that KRAS and BRAF mutant cell lines exhibit GLUT1-dependent increased glucose uptake and lactate production consistent with an increased rate of glycolysis. This characteristic of cells with KRAS and BRAF mutations may also have therapeutic implications, as the authors shows a glycolysis inhibitor is selectively toxic to cells harboring mutant KRAS or BRAF alleles.

If GLUT1 overexpression contributes to increased glucose uptake and glycolysis, does its overexpression permit cells to survive in low-glucose environments? The authors show that this is indeed the case, first by demonstrating that cell lines with mutant KRAS or BRAF (and therefore increased GLUT1 expression) could outgrow cell lines with wild-type alleles in low-glucose conditions. Second, the authors grew cell lines with wild-type KRAS and BRAF alleles for multiple generations in low-glucose conditions, and found that over 75% of the surviving cells had obtained stable overexpression of GLUT1 that persisted even after cells were returned to normal conditions. Even more striking was their observation that 4.4% of the wild-type KRAS cells had obtained KRAS mutations, compared to zero when the wild-type KRAS cell lines were grown in normal glucose conditions.

My only complaint with this paper is the title, I think it overstates their findings, as glucose deprivation can cause acquisition of a KRAS mutation but only in 4.4% of surviving tumor cells, and fails to mention their major finding, that KRAS and BRAF mutant tumor cell lines commonly overexpress GLUT1, and have characteristic GLUT1-dependent increases in glucose uptake and rate of glycolysis. GLUT1 was such an intriguing hit, not only because of its obvious links to glucose metabolism, but because its overexpression had previously been shown in several types of cancer and associated with poor prognosis (the KRAS and BRAF status of these tumors isn't mentioned, however). This suggests that a broad spectrum of tumors commonly acquire the trait of increased glycolysis in response to glucose-poor growth conditions by upregulating GLUT1, either via mutation of KRAS, BRAF, or some as yet unidentified gene, or by mutation of GLUT1 itself. Using a clever, unbiased, isogenic tumor cell line system, the authors of this paper caught cancer cells in the act of acquiring the traits needed to survive in harsh environments.

Other references:
Vander Heiden, et. al. Understanding the Warburg Effect: The Metabolic Requirements of Cell Proliferation. Science 324 (5930): 1029-1033 (2009).

Invertebrates: A Few Exons Short of a Full Deck

Regulation of Vertebrate Nervous System Alternative Splicing and Development by an SR-Related Protein
Calarco, et. al. Cell 138: 898-910 (2009).

Most eukaryotic genes are comprised of functional modules, or exons, separated by intervening segments of DNA known as introns. After a gene is transcribed, introns are removed (or "spliced") out of the pre-mRNA by a large RNA-protein complex called the spliceosome, which recognizes characteristic nucleotide sequences at exon-intron boundaries. Other proteins can influence splicing by either promoting or repressing spliceosome formation at potential splice sites. This can lead to the inclusion of different combinations of exons in the mature mRNA, a phenomenon known as alternative splicing. The result of alternative splicing is that a single gene can encode multiple protein isoforms with different amino acid sequences. The fact that at least 95% of human genes are alternatively spliced is cited as one explanation for why humans have fewer genes than some less complex organisms.

An additional layer of complexity is added when expression of a factor that regulates splice site selection is restricted to a certain tissue, ensuring that a particular protein isoform is similarly tissue-specific. Several such splicing factors have already been identified, particularly in the mammalian nervous system, but in the September 4 issue of Cell, Calarco, et. al. identify a neural-specific splicing factor found only in vertebrates, suggesting that it may have played a role in the evolution of the vertebrate nervous system. One major way vertebrate nervous systems differ from their invertebrate counterparts is that they have a central component (i.e., a brain and a spinal cord). Another is that their brains contain several specialized regions dedicated to particular functions. Surprisingly, the brains of all vertebrates have the exact same number of brain divisions (except jawless fish which lack a true cerebellum). Therefore, it is thought that the general structure of what we call a "brain " must have evolved at the same time as the first vertebrate, or shortly thereafter.

Many proteins involved in the regulation of alternative splicing have a characteristic arginine/serine (RS or SR) repeat domain, so in order to find potentially novel splicing factors, the authors used BLAST to identify all mouse protein-encoding genes with RS domains. They then analyzed the expression pattern of the 112 genes they found across a diverse panel of cell and tissue types, identifying a subset of RS domain genes whose expression was restricted to the adult nervous system. The authors chose to focus on one gene whose distinguishing feature is that it is highly conserved only in vertebrates. The authors demonstrate via RT-PCR, immunofluorescence, and Western blot that this gene is highly expressed only in neurons of the brain and sensory organs and encodes a 100 kilodalton protein, and consequently name it nSR100 (or neural-specific SR related protein of 100 kD).

The authors then utilize a custom microarray to determine which alternative splicing events are most affected following nSR100 knockdown. This lab previously developed this strategy using probes for both exon bodies and exon-intron junctions to detect the frequency of common alternative splicing events in different cell and tissue types. Using this technique, they show that loss of nSR100 in a neuronal cell line results in increased exon skipping, suggesting nSR100 normally acts to promote alternative exon inclusion in neurons. The authors verify by exon- specific RT-PCR that nSR100 knockdown does increase exon exclusion in the genes identified on the microarray, as well as show that nSR100 can bind mRNA containing target exons prior to splicing. Excitingly, they also show that nSR100 has a conserved function in neural development, as it is required for differentiation of neural stem cells in mice as well as nervous system development in zebrafish embryos.

The exact mechanism of how nSR100 is promoting alternative exon inclusion remains unclear, though the authors would argue its through antagonism of a known repressor of exon inclusion. Their data also suggests that nSR100 is necessary, but not sufficient, for exon inclusion, indicating a role for other splicing factors in nSR100-mediated regulation of alternative splicing. Furthermore, it remains to be seen what the functional consequences of the nSR100-dependent alternative splicing events are, though the authors point out that protein-protein interaction domains are encoded by some of the exons and that the literature suggests that the affected proteins may be part of an interaction network regulating cytoskeleton remodeling during neuronal differentiation. If that's true, it may be the mind-boggling number of combinations of alternative splicing events that allow a vertebrate's mind to be boggled.

Other References:
Northcutt, R.G. Understanding Vertebrate Brain Evolution. Integrative and Comparative Biology 42(4):743-756 (2002).
Pan, Q. Revealing Global Regulatory Features of Mammalian Alternative Splicing Using a Quantitative Microarray Platform. Molecular Cell 16(6): 929-941 (2004).

Did "Selfish" DNA Change Your Mind?

L1 retrotransposition in human neural progenitor cells.

Coufal, et. al. Nature 460: 1127-1131 (2009).

Retrotransposons (a.k.a. mobile elements or "jumping genes") are short segments of DNA whose only function is to make copies of themselves. Our DNA is absolutely littered with retrotransposons. For example, the active human mobile element L1 (short for LINE-1, or "long interspersed nucleotide element 1") comprises approximately 20% of the human genome! Long considered "junk DNA," increasing evidence suggests that retrotransposons may play important roles in gene function and human development. A recent paper in the August 27 issue of Nature suggests that these mobile elements may be particularly active in the developing human brain, with potential implications for neuronal function and diversity.

These DNA segments have been referred to as "selfish" or parasitic genes because the proteins they encode only function to copy and insert the DNA segment elsewhere in the genome, co-opting cellular resources in the process (known as "retrotransposition"). Since mobilization and integration of these elements could potentially disrupt normal gene function, cells have evolved ways to inhibit retrotransposition. For example, the promoters of retrotransposons are often hypermethylated, which prevents the expression of the two retrotransposon open reading frames encoding the proteins required for replication and integration. Retrotransposons usually take advantage of decreased DNA methylation in germ cells in order to propagate, but until now they were believed to be silent in somatic cells since novel retrotransposition events would not be passed on to progeny.

This paper demonstrates that a human L1 reporter construct can, in fact, retrotranspose in a neuronal progenitor cell (NPC). Furthermore, NPCs in which retrotransposition occurs can differentiate into functional neurons or glia. Their data suggest that the brain may be particularly permissive for L1 retrotransposition due to decreased methylation of L1 promoters compared to the skin (this is a striking -if not surprising- example of two different tissues having completely different epigenetic landscapes). Indeed, the authors demonstrate that L1 retrotransposition may be a more frequent event in the brain than in other tissues, as quantitative PCR indicates that there are significantly more copies of L1 in the brain rather than in the heart or liver of the same individual. It is unclear if these retrotransposition events have any phenotypic consequences, although a previous paper by this group examining human L1 retrotransposition in rat hippocampal stem cells showed that retrotransposition events could occur inside or near neuronally expressed genes, with consequent effects on gene expression and neuronal differentiation.

While this study falls short of actually sequencing the genomic DNA of individual neurons and uncovering the location and frequency of novel retrotransposition events in a human brain, the implications of the findings are exciting. First, this paper challenges the idea that the genetic material you inherit from your parents is a static entity that cannot be changed. It seems the neuronal genome may actually be somewhat dynamic during human development. Second, some studies have shown that a single neuron can affect behavior. What if a retrotransposition event affects a neuron's ability to retain a memory? Or sense a particular smell? Or respond to a neurotransmitter that affects mood? In addition to generating cellular diversity, is this phenomenon a driving force in creating human diversity? Did "selfish DNA" make us who we are?

Other References:

Martin, S.L. Jumping-gene roulette. Nature 460, 1087-1088 (2009).

Muotri, A.R., et. al. The necessary junk: new functions for transposable elements. Human Molecular Genetics 16, R159-R167 (2007).

Muotri, A.R., et. al. Somatic mosaicism in neuronal precursor cells mediated by L1 retrotransposition. Nature 435, 903-910 (2005).