Nuclear PKM2 activates transcription of MEK5 by phosphorylating stat3
We and others have observed that the PKM2 localizes to the cell nucleus (Hoshino et al., 2007
; Schneider et al., 2002
; Stetak et al., 2007
). To understand the functional significance of nuclear localization of PKM2, we examined nuclear PKM2 levels in ten cancer cell lines by immunoblot using an antibody raised against a peptide spanning aa 399 – 412 of PKM2 (PabPKM2). These ten cell lines represent different cancer progression stages and are derived from cancers of different organs/tissue types. Cell proliferation analyses demonstrated that M4C1, SW620, WM266, and H146 are more proliferated than their corresponding lines in the matched pairs (Fig. S1A
). Immunoblot analyses indicated much higher nuclear levels of PKM2 in the more proliferated cancer cells than those in the corresponding less proliferated cells in the matched pair (, and S1B
), The higher nuclear PKM2 levels were not because of differences in PKM2 expression (). The detection of PKM2 in the nuclear extracts was not due to contamination of cytoplasmic proteins as demonstrated by lack of GAPDH in the immunoblot analyses of the nuclear extracts (Fig. S1B
PKM2 regulates MEK5 transcription
To explore a possibility that nuclear PKM2 is involved in regulation of gene expressions, we carried out gene expression array analyses with SW620 cells expressing HA-PKM2. Expressions of 350 genes were upregulated, and expressions of 359 genes were downregulated, by at least two fold (Table S1
). Among those affected genes, expression of MEK5 was affected by over six fold. Expression of MEK5 plays an important role in SW620 cell proliferation. Thus, we verified the role of PKM2 in expression of MEK5 by RT-PCR. Consistent with the gene array analyses, RT-PCR demonstrated a dramatic change in the MEK5 expression in SW620 cells, while only a marginal increase in SW480 cells, a colon cancer cell line that is derived from the same patient with substantially lower proliferation rate, following the expression of PKM2. Consistently, higher nuclear levels of exogenously expressed HA-PKM2 were observed in more proliferated cells (). Similar patterns were also observed with another pair of melanoma cell lines, WM115 and WM266. To further verify the role of PKM2 in transcriptional regulation of MEK5, PKM2 was knocked down in SW620 and SW480 cells (). Quantitative RT-PCR analyses demonstrated a strong reduction in cellular MEK5 mRNA (). We further used chromatin immunoprecipitation (ChIP) to probe whether PKM2 interacted with the MEK5 promoter. The ChIP was carried out with SW620 cells using the PCR primer pair that spanned the region of nt −1,621 – −1,366 of MEK5 promoter. Our ChIP assays clearly demonstrated that PKM2 indeed interacted with the MEK5 promoter ().
We next tested whether up-regulation of MEK5 by PKM2 contributed to cell proliferation. MEK5 was first knocked down in SW620 and WM266 cells. As a result, the cell proliferation rate was dramatically reduced (Fig. S1C
). The increases in proliferation by expression of HA-PKM2 in the MEK5 knockdown SW620 and WM266 cells were also largely reduced (Fig. S1C
), suggesting that up-regulation of MEK5, at least partially, mediated the effects of HA-PKM2 overexpression on cell proliferation. Moreover, we analyzed the MEK5 expression levels in these ten cancer cell lines. MEK5 was expressed in all ten cell lines, but there were substantially higher levels of MEK5 in the more proliferated cancer cells than in the less proliferated cancer cells within the matched pair (Fig. S1D
). Clearly, there is a close correlation between cellular MEK5 levels and nuclear PKM2 levels, and both were correlated closely with the cell proliferation status of the cells (comparing Fig. S1D
with ). To rule out a possibility that high cellular levels of MEK5 promote PKM2 nuclear localization, MEK5 was knocked down, and the nuclear/cytoplasmic PKM2 was examined. It was clear that knockdown of MEK5 did not affect the nuclear levels of PKM2 (Fig. S1E
). We concluded from these studies that up-regulation of MEK5 by nuclear PKM2 contributed to cell proliferation.
How would PKM2 function in regulation of gene transcription? Sequence analyses did not reveal any known DNA binding domain/motifs in PKM2. One possibility is that PKM2 may activate particular transcription factors. Thus, we attempted to probe the interaction of PKM2 with several known transcription factors in nuclear extracts of SW620 and SW480 cells by immunoprecipitation, such as Oct-1, Oct-4, Gadd45, SOX4-1, and stat3. Among these selected targets, stat3 and SOX4 were involved in transcription of MEK5 (Aaboe et al., 2006
; Song et al., 2004
), while Oct-4 was reported to interact with PKM2 in extracts made from ESC (Lee et al., 2008
). Our experiments showed that stat3 was the only transcriptional factor among the selected targets that interacted with PKM2 (Data not shown, , and Fig. S2 A&B
). It might be due to the differences between ESC and cancer cells, we did not detect the PKM2 and Oct-4 interaction in the extracts prepared from cancer cells (data not shown). We noted that stat3 interacts with MEK5 promoter at −1776 nt – −1520 nt region by ChIP (Song et al., 2004
). Interestingly, PKM2 also interacts with MEK5 promoter at the same region (). To address whether the activation of stat3 mediates the effects of PKM2 on upregulation of MEK5 transcription, we first knocked down stat3 in SW620/SW480 and WM266/WM115 cells, and HA-PKM2 was subsequently expressed. It was clear that expression of HA-PKM2 could no longer upregulate MEK5 as measured by cellular levels of both mRNA and protein of MEK5 (). To exclude a possibility that downregulation of MEK5 is solely due to stat3 knockdown, HA-PKM2 was expressed in these cells. Immunoblots indicated that MEK5 was up-regulated upon the HA-PKM2 expression (Fig. S2D
). To further test the role of stat3 in mediating the effects of PKM2 on upregulation of MEK5 transcription, we used a dominant-negative mutant stat3 (Y705F, ref to as stat3-DN) (Xie et al., 2006
). The mutant was co-expressed with HA-PKM2 in SW620/SW480 and WM266/WM115 cells. It was evident that expression of the stat3-DN largely diminished the effects of PKM2 on upregulating MEK5 transcription (Fig. S2 E&F, G&H
). As a control, expression of the stat3-DN in PKM2 knockdown SW480/SW620 cells did not result in up-regulation of MEK5 expression (Fig. S2E
). The activity of stat3 can be inhibited by specific inhibitor (Xu et al., 2008
). We therefore tested the effects of a stat3 inhibitor on the upregulation of MEK5 by expression of HA-PKM2 in SW620 and WM266 cells. Treatment of the HA-PKM2 expressing cells with the inhibitor resulted in a decrease in MEK5 transcription (Fig. S2 I&J
). Thus, we concluded that activation of stat3 mediated the regulatory effects of PKM2 on MEK5 transcription. Consistently, re-analyses of our expression array data revealed changes of a number of stat3 regulatory genes in the PKM2 over-expressing SW620 cells (Table S2
PKM2 regulates MEK5 transcription via activation of stat3
We analyzed whether expression of PKM2 affected the stat3 binding to its target DNA at the MEK5 promoter. To probe whether knockdown or expression of PKM2 affected the interaction of stat3 with the MEK5 promoter, we performed ChIP in SW620 cells in which the endogenous PKM2 was knocked down or HA-PKM2 was expressed. Clearly, the stat3 and MEK5 promoter interaction was strengthened by HA-PKM2 expression and was weakened by PKM2 knockdown (). We then analyzed the effects of PKM2 on the interaction between stat3 and its target DNA by gel mobility shift assays using a 32
P-labled oligonucleotide duplex containing a stat3 binding sequence (Xie et al., 2006
). The experiments were carried out with nuclear extracts of SW620 cells with/without PKM2 knockdown or with/without HA-PKM2 expression. It was clear that a slow migration complex was assembled with the labeled probe and addition of the antibody against stat3 resulted in a supershift complex (). Interestingly, knockdown of PKM2 substantially decreased the assembly of the oligo-protein complex and assembly of stat3 in the complex (the weak supershift) (), while expression of HA-PKM2 increased assembled complex and assembly of stat3 in the complex ().
PKM2 upregulates MEK5 transcription by promoting stat3 DNA interaction and phosphorylation of stat3
Stat3 is activated by phosphorylation at Y705. The phosphorylation increased its DNA binding affinity (Sehgal, 2008
). We observed that PKM2 increased the stat3 binding to its target sequence both in vitro
and in vivo
. We suspected that PKM2 may play a role in stat3 phosphorylation at Y705. To test this conjecture, we analyzed the stat3 phosphorylation in the nuclear extracts prepared from SW620 cells in which the HA-PKM2 was expressed using an antibody against the Y705 phosphorylated stat3 (P-y705/stat3). Clearly, a significant increase in Y705 phosphorylation of stat3 was evident. Examination of the cellular levels of stat3 indicated that the expression levels of stat3 were not affected (). We also observed a significant reduction in the stat3 phosphorylation in SW620 cells upon PKM2 knockdown (). JAK2 and c-Src are the most common protein tyrosine kinases that phosphorylate stat3. We therefore probed whether JAK2 and c-Src became more activated by the HA-PKM2 expression. Using antibodies against JAK2, the phosphorylated JAK2, c-Src, and the phosphorylated c-Src, our experiments showed that there was no increase in JAK2 and c-Src activation (Fig. S3A
). Furthermore, treatment of cells with JAK2 and c-Src inhibitors did not lead to any significant changes of the PKM2-depedent stat3 phosphorylations (Fig. S3 B&C&D
), suggesting that stat3 phosphorylation was not due to activation of JAK2 and Src.
Since PKM2 acts in the glycolysis pathway, we analyzed cellular glucose, lactate, and pyruvate in two pairs of cells, SW620/SW480 and WM266/WM115, in which the HA-PKM2 was expressed. Upon expression of HA-PKM2, there were no significant changes in cellular glucose and production of lactate and pyruvate for either WM266 or SW620 cells (the more proliferated cell lines). Cellular pyruvate and lactate increased slightly in SW480 and WM115 cells (the less proliferated cell lines), indicating a slight increase in glycolytic pyruvate kinase activity in the less aggressive cancer cells, while expression of PKM2 did not lead to any changes in glycolytic pyruvate kinase activity in the more proliferated cancer cells. Overall, no substantial changes in cellular pyruvate and lactate were observed (Fig. S4 A-F
), suggesting that the differences in stimulating cell proliferation by expression of HA-PKM2 were not simply due to the effects on the changes in carbohydrate metabolism.
Dimeric PKM2 is the active protein kinase
We next sought to test whether PKM2 could directly phosphorylates stat3. An in vitro phosphorylation assay using both the E. coli expressed recombinant PKM2 (rPKM2) and the HA-PKM2 immunopurified from nuclear extracts of SW620 cells in the presence of ATP did not yield phosphorylation of a commercially available GST-stat3. Since PKM2 uses PEP as phosphate donor to phosphorylate ADP in the glycolysis, we reasoned that the protein may use the same phosphate donor to phosphorylate a protein substrate. Thus, we replaced ATP by PEP in our in vitro reaction. Immunoblot using the antibody P-y705/stat3 demonstrated that the GST-stat3 was phosphorylated by the HA-PKM2 in the presence of PEP. Consistently, stat3 was not phosphorylated in the presence of ATP (). These results indicated that PKM2 is a protein kinase using PEP as the phosphate donor.
Phosphorylation of GST-stat3 by the rPKM2
The kinase activity of the nuclear HA-PKM2 in phosphorylation of stat3 was substantially higher than that of the rPKM2 expressed in E. coli
. The results led us to compare the protein kinase activity of the nuclear PKM2 and the cytoplasmic PKM2. The same in vitro
phosphorylation reactions were performed with the HA-PKM2 immunopurified from nuclear or cytoplasmic extracts of SW620 cells in the presence of PEP or ATP. The HA-PKM2 from the nuclear extracts had much higher activity than that of protein from the cytoplasmic extracts (). To test whether the Y705 of stat3 is the only phosphorylation site by PKM2 in cells, we expressed a stat3 mutant (Y705A) and GFP-PKM2 in SW620 cells. Phosphorylations of endogenous and exogenously expressed stat3 were examined by immunoprecipitation of HA-tagged stat3 mutant or endogenous stat3 followed by immunoblot using an antibody against phorpho-tyrosine. It was clear that the endogenous stat3 was phosphorylated, while the exogenously expressed mutant was not phosphorylated (Fig. S4G
), indicating that Y705 is the only site.
It was reported that the tetramer and dimer of PKM2 co-exist in proliferation cells (Mazurek et al., 2005
). We therefore questioned whether the differences in the protein kinase activity of nuclear/cytoplasmic HA-PKM2 and the rPKM2 were due to dimer or tetramer of the protein. To investigate whether PKM2 is a dimer or a tetramer in the nucleus and in the cytoplasm, we first fractioned the nuclear and cytoplasmic extracts of SW620 cells by size exclusion chromatography. The levels of PKM2 in each fraction were examined by immunoblot using the antibody PabPKM2. Nuclear PKM2 was only detected in fractions 14 – 16, while cytoplasmic PKM2 was mainly detected in fractions 11 – 16 with the highest concentrations in fractions 11 – 13. According to the MW calibration standard (Fig. S5 A&B
), fraction 11 co-elutes with a MW near 240 kDa, while fraction 14 co-elutes with a MW near 120 kDa (). The gel-filtration chromatography suggested that nuclear PKM2 was completely dimer, while the cytoplasmic PKM2 existed in both dimer and tetramer. The same procedure was also employed to analyze whether the rPKM2 is a dimer or a tetramer. It was evident that the rPKM2 was mostly tetramer with very small amount of dimer (). It is well documented that FBP functions as an allosteric regulatory factor that stabilizes the tetramer PKM2. We therefore asked whether FBP could convert the dimer nuclear PKM2 to a tetramer form. To this end, nuclear extracts of SW620 cells were incubated with 5 mM FBP at room temperature for 2 hours. The dimeric/tetrameric status of PKM2 in the nuclear extracts was analyzed by the same procedure. It was evident that FBP did not convert PKM2 from the dimeric to the tetrameric form ().
Close examination of the crystal structure of the tetramer human PKM2 (Dombrauckas et al., 2005
) reveals that a positive charged residue R399 may plays a critical role in forming the tetramer of PKM2. It is notable that the R399 forms stable charge-charge interactions with residues E418 and E396 of PKM2 located on the other dimer of the tetramer PKM2 (). We therefore created a mutant R399E to disrupt the interactions. Size exclusion chromatography analyses demonstrated that the R399E mutant was mostly dimer (). We reasoned that the dimeric R399E would be more active in phosphorylating stat3. Thus, the in vitro
phosphorylation reactions were carried out with the rPKM2 and the R399E. It was evident that the protein kinase activity of the R399E was substantially higher than that of the wild-type rPKM2 (), while the pyruvate kinase activity of the mutant was dramatically lower than that of the rPKM2 (). The results supported our speculation that dimeric PKM2 is an active protein kinase.
Dimeric PKM2 is active protein kinase and expression of the R399E mutant promotes cell proliferation
To further verify the phosphorylation of stat3 by PKM2, we prepared [32
P]-labeled PEP (Roossien et al., 1983
). The same in vitro
phosphorylation reaction was carried out with rPKM2, rPKM1, and R399E using the [32
P]-PEP. Autoradiography indicated that the GST-stat3 was phosphorylated by the rR399E. The phosphorylation of the GST-stat3 by the rPKM2 was very weak (was only visualized by a substantial overexposure). The GST-stat3 was not phosphorylated by the rPKM1 in the presence of the [32
P]-PEP (). No phosphorylation can be detected even under very high overexposure. To ensure that the in vitro
phosphorylation of stat3 by PKM2 and the R399E was at comparable physiological conditions, we compared the phosphorylation of the GST-stat3 by JAK2 and R399E. Clearly, very similar levels of stat3 phosphorylations were observed by both kinases (Fig. S6A
). The R399E could not phosphorylate other proteins, such as stat5 and BSA under the same conditions (Fig. S6 B&C
), indicating substrate specificity. Extensive analyses of enzyme kinetic of pyruvate kinase in various tissues (Km = 0.07 – 1.2 mM range) indicate that the physiological concentration of PEP is likely to be at μM - mM (Mazurek et al., 2007
; van Veelen et al., 1978
). To test whether PKM2 and the R399E would phosphorylate stat3 at the physiological PEP concentrations, we carried out the in vitro
phosphorylation at various PEP concentrations. The levels of phosphorylated stat3 remained almost constant down to 100 μM of PEP. However, there was a clear decrease in stat3 phosphorylation when PEP concentration fell below 10 μM (Fig. S6D
). These experiments indicated that phosphorylation of stat3 by PKM2 were at physiologically comparable conditions.