In our examination of PKM isoform expression and PK activity in a series of over 100 astrocytomas, we found that PKM1 expression and PK activity was consistently low (relative to normal brain tissue) across as wide-range of gliomas, but that PKM2 expression increased in a GBM-specific manner. Both low levels of PKM1/PK activity and high levels of PKM2 were, however, critical for continued growth of glioma cells.
Numerous groups have demonstrated that GBM express more PKM2 than normal brain, although the extent to which PKM expression and PK activity vary with respect to glioma grade has only partially and indirectly been examined. David et al examined the ratio of PKM2 to PKM1 RNA expression in 4 grade I gliomas, 4 low-grade astrocytomas, and 4 GBM 
. The results of these studies agree with the present data in showing that all glioma express more PKM2 mRNA than PKM1 mRNA. The methodology in the David study, however, did not allow a direct comparison of PKM2 expression or PK activity between the tumor groups, and as such could not evaluate how these changes with regard to glioma grade. Similarly, Chinnaiyan et al compared PK activity and PKM2 protein expression in groups of grade 3 and 4 glioma sub-typed by expression profiling into proneural or mesenchymal sub-groups 
. Although the proneural glioma were suggested to have higher levels of PK activity and lower levels of PKM2 protein expression than the mesenchymal tumors, grade 3 and grade 4 tumors were represented in both groups, precluding both comparison to the present data and any conclusions about PKM expression and activity based on tumor grade. The present studies therefore represent the first complete analysis of PKM expression and PK activity with regard to glioma grade, and are the first to our knowledge to show grade-specific differences in PKM2 expression, but not PK activity, in glioma.
One of the most striking observations of the present study is the significant up-regulation of PKM2 expression in GBM relative to all other forms of glioma. Previous studies have suggested that even slow-growing benign oncocytoma exhibit high levels of PKM2 expression, and pre-malignant Barrett's intestinal metaplasia also exhibited significant PKM2 over-expression 
. It was therefore surprising that benign grade I glioma exhibited levels of PKM2 expression no different from those of frankly malignant grade III glioma, but less than that of GBM. A trivial explanation of this data is that the grade I-III tumor samples, as a result of the diffuse nature of the tumors, contain a larger percentage of normal cells than the grade IV tumors, and that the admixture of normal cells expressing high levels of PKM1 and tumor cells expressing high levels of PKM2 results in a grade I-III glioma sample that appears as a whole to have a lower level of PKM2 expression than its truly tumorigenic components. All samples used, however, were verified by pathologic examination to contain >90% tumor, and even if this explanation were true, mixed samples would be expected to have higher levels of both PKM1 and PKM2, rather than just the increased levels of PKM2 (relative to normal brain) noted. The related question of whether all glioma over-express PKM2 is an open to interpretation and based on the point of reference. Relative to normal brain, all glioma express higher levels of PKM2 and lower levels of PKM1. These tumors, however, most likely did not arise from normal differentiated brain cells and more likely arose from a stem or progenitor population 
. Further complicating the matter, each grade of glioma may arise from a distinct progenitor population, and even with a given grade there may be different cells of origin 
. Nonetheless, the pattern of PKM1/PKM2 expression in grade I-III glioma more closely resembles that of NSC population than normal brain. Whether these tumors represent a frozen stage of differentiation or a dramatic shift in metabolic profile remains an unanswerable question. It is clear from the data, however, that GBM represent a unique type of glioma, quantitatively different from the other grades of glioma with respect to PKM2 expression.
If GBM exhibit significantly higher levels of PKM2 expression than other glioma, one important question is how this grade-specific up-regulation occurs. A number of genetic events including c-myc over-expression, growth factor over-expression, and Ras pathway activation have been linked to the control of PKM transcript splicing 
, and the increased incidence of these alterations in GBM may tilt the balance of splicing toward production of the PKM2 transcript and help explain the grade-specific increases in PKM2 expression noted. It's worth noting, however, that the present study examined both de novo
GBM, as well as secondary GBM that arose from lower grade tumors based on their mutant IDH status. The fact that both groups of GBM exhibit increased PKM2 expression despite their very different genetic compositions 
suggests that the up-regulation of PKM2 in GBM appears to be driven by fundamental processes shared by all GBM rather than by known genetic alterations that play a direct role in the PKM splicing process but differ between the two GBM groups analyzed.
The work presented also suggests that modulation of both PKM expression and PK activity are important for continued glioma cell growth. The GBM cells used for these studies were representative of primary GBM in terms of patterns of PKM isoform expression and PK activity, and in these cells the up-regulation of PKM1 in the face of high endogenous levels of PKM2 increased PKM activity and suppressed growth, suggesting that optimal glioma growth is limited by high PK activity. Although the basis for this effect is unclear, the data are consistent with idea that the use of glucose for pyruvate and ATP production limits the ability of cells to create the macromolecules needed for increased proliferation 
, resulting in a generalized accumulation of cells G0/G1. The modulation of the end products of PK activity may therefore be a reasonable strategy to inhibit growth of these GBM cells. In contrast, suppression of PKM2 levels in the face of low PKM1 levels also suppressed cell growth. Similar findings in other systems were attributed to levels of PK activity insufficient to sustain growth 
, although in the present study the PK activity levels in the PKM2 knock-down cell were comparable to those in primary GBM. An alternative explanation may be provided by recent studies showing that PKM2 not only has metabolic effects but can also translocate to the nucleus and facilitate cyclin D1 expression and cell cycle progression 
. While the effects of PKM2 knock-down noted in the present study are consistent with the non-metabolic effects of PKM2, the accumulation of cells was not associated with a generalized slowing in cell cycle progression but rather a specific G2 arrest. It may therefore be possible that PKM2 has non-metabolic effects on regulators of the G2 checkpoint that cooperate with, or supersede, those related to cyclin D1.
The present findings provide the first detailed picture of PKM expression and activity across a range of gliomas of different grades. These studies provide a more complex picture of PKM isoform regulation than previously described, and suggest that while PK activity is uniform across all grades of glioma, PKM2 expression is up-regulated not at the benign/malignant transition and not in a gradual manner related to tumor grade, but rather most dramatically in GBM. Therapeutic approaches targeting metabolic changes in glioma may therefore benefit from considering glioma subtypes separately, and in particular in focusing on the potentially unique role of PKM2 over-expression in GBM.