Abstract
Sequential metabolic enzymes in glucose metabolism have long been hypothesized to form multienzyme complexes that regulate glucose flux in living cells. However, it has been challenging to directly observe these complexes and their functional roles in living systems. In this work, we have used wide-field and confocal fluorescence microscopy to investigate the spatial organization of metabolic enzymes participating in glucose metabolism in human cells. We provide compelling evidence that human liver-type phosphofructokinase 1 (PFKL), which catalyzes a bottleneck step of glycolysis, forms various sizes of cytoplasmic clusters in human cancer cells, independent of protein expression levels and of the choice of fluorescent tags. We also report that these PFKL clusters colocalize with other rate-limiting enzymes in both glycolysis and gluconeogenesis, supporting the formation of multienzyme complexes. Subsequent biophysical characterizations with fluorescence recovery after photobleaching and FRET corroborate the formation of multienzyme metabolic complexes in living cells, which appears to be controlled by post-translational acetylation on PFKL. Importantly, quantitative high-content imaging assays indicated that the direction of glucose flux between glycolysis, the pentose phosphate pathway, and serine biosynthesis seems to be spatially regulated by the multienzyme complexes in a cluster-size-dependent manner. Collectively, our results reveal a functionally relevant, multienzyme metabolic complex for glucose metabolism in living human cells.
Keywords: glycolysis, metabolism, microscopic imaging, phosphofructokinase, protein complex, Metabolic Complex
Introduction
Glucose metabolism involves two reciprocal pathways: glycolysis and gluconeogenesis, during which glucose flux partitions between energy metabolism and anabolic biosynthetic pathways (supplemental Fig. S1). To regulate both the flux and allocation of glucose-derived pathway intermediates, the cell needs mechanisms to coordinate the activities of the pathway enzymes. Specifically, cancer cells have evolved to control their metabolic activity by genetic alterations of isozyme expression and/or post-translational modifications of the rate-limiting enzymes of glucose metabolism, thus preferentially directing pathway intermediates into building block biosynthesis (1–3). However, spatiotemporal mechanisms controlling the direction of pathway intermediates at various metabolic nodes in single cells have not been proposed yet. Hence, the lack of such fundamental mechanisms regulating the direction of glucose flux inside a cell prevents us from comprehending glucose metabolism in normal cells and thus its metabolic alterations in human disease cells.
Meanwhile, the sequential metabolic enzymes of glucose metabolism have long been proposed to form multienzyme complexes in a variety of organisms. To date, various in vitro studies (4–15) have suggested that glycolytic enzymes in Escherichia coli, Arabidopsis, Drosophila, yeast, and protists form metabolic complexes in cells. These studies have been mainly supported by measuring individual enzyme activities from chromatographically fractionated pools of cell lysates or biochemically semipurified subcellular fractions. For instance, extensive in vitro biochemical analysis of mitochondrial fractions of plant cells demonstrated that glycolytic enzymes were associated with mitochondria in a cellular respiration-dependent manner (5, 7). In addition to such in vitro investigations, immunofluorescence imaging has demonstrated that various glycolytic enzymes in mammalian erythrocytes form a glycolytic complex on the inner surface of the erythrocyte membrane in the presence of the anion transporter band 3 protein (16–18). The assembly and disassembly of this complex was dependent on both the phosphorylation state of the band 3 protein and the oxygenation state of hemoglobin (16). The interactions between glycolytic enzymes and the band 3 protein were further supported by FRET and chemical cross-linking techniques (18, 19). Furthermore, colocalization and direct interaction between fructose-1,6-bisphosphatase (FBPase)3 and aldolase have been studied both in vitro and in myocytes (8, 9, 20, 21), proposing the formation of metabolic complexes with α-actinin on the Z-line of vertebrate myocytes. Therefore, these studies have supported the formation of multienzyme metabolic complexes in nature.
However, there are still many challenges ahead when exploring new dimensions of glycolytic enzymes and their complexes, particularly in living human cells. Given the tissue specificity of the band 3 protein in erythrocytes or the unique Z-line structure of myocytes, the observed metabolic complexes in these cells do not fully provide mechanistic insights of how such enzyme complexes are organized in other human cell types absent their reported scaffolds. Importantly, the metabolic influence of these complexes on cells remains to be further elucidated. Therefore, we sought to identify such complexes in living human cancer cells and their functional contributions to cellular metabolism.
In this work, we provide several lines of compelling evidence that every cytoplasmic, rate-limiting enzyme involved in glycolysis, as well as gluconeogenesis, is spatially compartmentalized into three different sizes of cytoplasmic clusters in human cervical adenocarcinoma HeLa and human breast carcinoma Hs578T cells. As controls, we validate that the varying sizes of the enzyme cluster observed in HeLa and Hs578T cells are independent of the expression levels of tagged enzymes, as well as the tagging method. Subsequent biophysical analyses using FRET and fluorescence recovery after photobleaching (FRAP) techniques corroborate the formation of multienzyme metabolic complexes in live cells. We further demonstrate that the multienzyme complex for glucose metabolism is a spatially distinct cellular entity from other cytoplasmic cellular bodies, including stress granules (22), aggresomes (23, 24), and purinosomes (25, 26). Importantly, we provide evidence to support the cluster-size-dependent functional roles of the multienzyme metabolic assemblies at single-cell levels. Collectively, we demonstrate the existence of a multienzyme metabolic complex for glucose metabolism in living human cells, providing new mechanistic insights regarding how a cell regulates the direction of glucose flux between energy metabolism and anabolic biosynthetic pathways at single-cell levels.
Results
Formation of cytoplasmic PFKL clusters in human cancer cells
We first investigated subcellular locations of the metabolic enzymes of glucose metabolism using fluorescent protein tags under fluorescence live-cell microscopy. We found that human liver-type phosphofructokinase 1, tagged with a monomeric form of enhanced green fluorescent protein (PFKL-mEGFP), forms discrete cytoplasmic clusters of varying sizes in transfected HeLa cells (Fig. 1, A–C). This clustering pattern of PFKL-mEGFP was also identified in human breast carcinoma Hs578T cells (Fig. 1, D–F) and human pancreatic adenocarcinoma Pa04C and Pa18C cells (supplemental Fig. S2, A and B). Moreover, the PFKL clusters were formed when mEGFP was alternatively tagged at the N terminus of PFKL (mEGFP-PFKL) (supplemental Fig. S2C).
Figure 1.
Subcellular localization of PFKL-mEGFP in human cancer cells. A–F, PFKL-mEGFP displays three different sizes of cytoplasmic fluorescent clusters in transfected HeLa and Hs578T cells: small clusters showing 0.1 μm2 (A and D), medium-sized clusters having an area between 0.1 and 3 μm2 (B and E), and large-sized clusters under 8 μm2 (C and F). G, the percentage of Hs578T cells displaying each size was also quantified by inspecting ∼3800 transfected cells; 1.6 ± 1.4, 58.3 ± 4.7, 13.4 ± 3.3, and 26.7 ± 3.6% of cells showed no cluster, small-, medium-, and large-sized clusters, respectively. The error bars indicate the standard deviations of 13 independent experiments. H–K, in addition, the total and mean fluorescent intensity per cell, the average size of clusters per cell and the number of clusters per cell were also quantified from Hs578T cells to investigate correlations between the parameters. A total of 109 transfected cells were selected for analysis from triplicate imaging sessions. However, we found no correlation between the expression level of PFKL-mEGFP and the number of clusters per cell (H and I) or the average size of clusters per cell (J and K). Scale bars, 10 μm.
To better characterize the diverse clustering phenomena, we analyzed the size distribution of PFKL-mEGFP clusters at single-cell levels when Hs578T cells were grown in RPMI 1640 supplemented with 10% dialyzed FBS (dFBS) (described under “Experimental procedures”). Based on the entirety of the data presented in this work, we have categorized the varying sizes of PFKL-mEGFP clusters into three distinguishable subgroups for clarification (Fig. 1, A–G). In the first subgroup, which represents 58.3 ± 4.7% of transfected Hs578T cells (Fig. 1G), PFKL-mEGFP proteins assembled throughout the cytoplasm into a number of small clusters. The small-sized clusters are defined as having less than 0.1 μm2 (Fig. 1, A and D) based on the calculated area of the point spread function for the mEGFP emission (i.e. 0.1 μm2) (27). Line scan fluorescent intensity analysis across the cell also supports that this phenomena was clearly distinguishable from the diffusive pattern exhibited by other mEGFP-tagged metabolic enzymes including, but not limited to, hypoxanthine-guanine phosphoribosyltransferase and C1-tetrahydrofolate synthase (25, 28) (supplemental Fig. S3). In the second subgroup, ∼97% of PFKL-mEGFP clusters in 13.4 ± 3.3% transfected cells displayed medium-sized clusters, ranging from 0.1 to 3 μm2 in size (Fig. 1G and supplemental Fig. S4A). However, relative to the second subgroup, the third subgroup of transfected cells exhibited an increased fraction of large-sized clusters (i.e. ∼16% versus 3%) (supplemental Fig. S4B), ranging between 3 and 8 μm2, which were randomly distributed in the cytoplasm in the presence of smaller clusters (Fig. 1, C and F). 26.7 ± 3.6% of transfected Hs578T cells are assigned to the third subgroup (Fig. 1G). In summary, we have defined three differently sized PFKL-mEGFP clusters at single-cell levels in HeLa and Hs578T cells, namely small-, medium-, and large-sized clusters.
Large-sized clusters are found in various cancer cells, but not in non-cancerous human breast tissue cells (Hs578Bst)
We then investigated whether the clustering formation of PFKL-mEGFP would be relevant for human cancer cells. Because three different sizes of PFKL-mEGFP clusters were observed in human breast cancer Hs578T cells, we selected to test the formation of PFKL-mEGFP clusters in a non-cancerous human normal breast cell line, Hs578Bst, because this cancer/non-cancerous pair of breast tissue cell lines were derived from the same patient (29). From our cluster size analysis of non-cancerous breast Hs578Bst cells, we found that PFKL-mEGFP formed a number of small clusters (i.e. ∼0.1 μm2) throughout the cytoplasm (Fig. 2A), as well as medium-sized clusters (< 3 μm2) (Fig. 2B). However, non-cancerous Hs578Bst cells did not induce large-sized fluorescent clusters. Although additional non-cancerous normal cells may need to be evaluated, we hypothesize that large-sized clusters of PFKL-mEGFP in Hs578T cells are a cancer-relevant cellular phenomenon.
Figure 2.
No large-sized clusters in non-cancerous human breast tissue cells (Hs578Bst). Non-cancerous human breast tissue cells, Hs578Bst, exhibited only two clustering patterns when transfected with PFKL-mEGFP (A and B), which are an absolute scale equivalent to small- (Fig. 1, A and D) and medium-sized (Fig. 1, B and E) clusters observed in Hs578T cells. Scale bars, 10 μm, unless otherwise indicated.
We further evaluated the formation of large-sized PFKL clusters in the cancerous Hs578T cells when they were maintained in the non-cancerous Hs578Bst-supporting medium. Please note that although cancerous Hs578T cells were cultured in the RPMI 1640 and 10% dFBS, the non-cancerous Hs578Bst cells were cultured in Hybri-Care medium (ATCC) supplemented with 1.5 g/liter NaHCO3, 10% FBS, and 30 ng/ml mouse EGF. However, in the Hs578Bst-supporting medium, 23.4 ± 4.5% of Hs578T cells showed the formation of large-sized clusters, which is similar to the percentage of cells (26.7 ± 3.6%) showing large-sized clusters in Fig. 1G. Therefore, we conclude that large-sized clusters in cancerous Hs578T cells are induced regardless of the differences of the two cell culture conditions.
Expression-independent formation of PFKL-mEGFP clusters in Hs578T cells
To evaluate whether the varying size of PFKL-mEGFP clusters is dependent on the protein expression level, we further quantitated the mean or total fluorescent intensity per cell as an indication of the protein expression level in single cells. However, we did not identify any positive or negative correlation between the mean or total fluorescent intensity per cell and the average size of clusters per cell or the number of clusters per cell (Fig. 1, H–K, and supplemental Fig. S4C). As an added note, the relationship between the size of individual cells and the average size of clusters per cell or the number of clusters per cell also appears to be random (supplemental Fig. S4, D and E). Collectively, we conclude that neither the formation of the fluorescent clusters by PFKL-mEGFP, nor their size, is governed by the expression levels of PFKL-mEGFP at single-cell levels.
Cluster-like distribution of endogenous PFKL in fixed HeLa and Hs578T cells
To examine whether endogenous PFKL exhibits similar cytoplasmic clustering patterns in non-transfected cells, we performed immunocytochemistry against endogenous PFKL in HeLa and Hs578T cells. Although chemical manipulation sometimes alters subcellular distribution of endogenous proteins during fixation and permeabilization (30), we observed in >95% of fixed cells that the subcellular distribution of endogenous PFKL appears to be discrete clusters in the cytoplasm (supplemental Fig. S5), much as we observed in transfected cells expressing PFKL-mEGFP. We did not detect immunofluorescent signal absent primary, secondary, or both antibodies in our controls. In addition, our results are consistent with the subcellular distribution of immunostained PFKLs in other fixed human cells, as can be seen at the Human Protein Atlas. Taken together, these data indicate that the subcellular distribution of PFKL-mEGFP in transfected cells mirrors the distribution pattern of endogenous PFKL, setting us to investigate real-time spatial dynamics of the PFKL-mEGFP clusters in live cells.
The mEGFP tag does not interfere with the subcellular localization of metabolic enzymes, nor does it induce cytoplasmic clusters
We also demonstrated as controls that other enzymes in glucose metabolism whose subcellular locations are genetically determined were properly localized in their designated organelles when tagged with mEGFP in the presence and absence of monomeric orange fluorescent protein-tagged PFKL (PFKL-mOFP). mEGFP-tagged hexokinase 1 and 2 and pyruvate carboxylase were explicitly associated with mitochondria in HeLa and Hs578T cells (supplemental Fig. S6A) (28), whereas mEGFP-tagged glucose-6-phosphatase 3 stained the endoplasmic reticulum in HeLa and Hs578T cells (supplemental Fig. S6B). The ectopic expression of mEGFP alone did not induce any manner of fluorescent foci with or without PFKL-mOFP in our experimental conditions in either HeLa or Hs578T cells. We conclude that along with other metabolic enzymes showing diffusive patterns in the cytoplasm with mEGFP tags (supplemental Fig. S3C) (25, 28), the mEGFP tag does not alter subcellular locations of metabolic enzymes involved in glucose metabolism in living human cells.
Additionally, we have constructed tetracysteine (TC) motif-conjugated variants of PFKL. Briefly, the TC motif, which is composed of only six amino acids (e.g. N-CCPGCC-C), can be fluorescently labeled with small biarsenical compounds (31). To determine the specificity of TC-mediated fluorescent staining in our conditions, we first transfected Hs578T cells with dually tagged PFKL with TC and mEGFP (i.e. PFKL-TC-mEGFP). We then incubated the cells with a biarsenical resorufin-derivative agent, namely ReAsH-EDT2 (ThermoFisher Scientific), to visualize the PKFL clusters. Clearly, colocalization of ReAsH (in the red channel) and mEGFP (in the green channel) signals supports the specific labeling of ReAsH to PFKL-TC-mEGFP in our conditions (supplemental Fig. S7, A–C). Finally, our live-cell imaging with PFKL-TC demonstrated three differently sized clusters in the presence of ReAsH (supplemental Fig. S7D) as PFKL-mEGFP in the cytoplasm of Hs578T cells. Collectively, we conclude that the mEGFP tag did not arbitrarily promote the formation of fluorescent clusters in cells.
Mobility of PFKL-mEGFP in live Hs578T cells
We have also investigated whether the fluorescent clusters formed by PFKL-mEGFP are composed of mobile proteins or insoluble protein aggregates. Using FRAP, we photobleached the various sizes of PFKL-mEGFP clusters ranging from ∼0.3 to ∼8 μm2 in live Hs578T cells and then monitored the recovery of fluorescent intensities in the photobleached areas (supplemental Fig. S8). We first performed FRAP on the minor population of Hs578T cells, in which transfected PFKL-mEGFP does not form any kind of cluster (i.e. <2% of transfected cells). The diffusion coefficient of the non-clustering PFKL-mEGFP was 0.114 ± 0.045 μm2/s (NFRAP = 10). However, the diffusion coefficients of PFKL-mEGFP within the medium- and large-sized clusters are similar to each other; i.e. Dapp = 0.018 ± 0.009 μm2/s (NFRAP = 80). The ∼7-fold difference in the diffusion coefficients of PFKL-mEGFP indicates that the diffusion of PFKL-mEGFP is noticeably hindered in cluster-positive Hs578T cells, suggesting potential interactions with other biomolecules. Nonetheless, this FRAP analysis strongly supports the possibility that the PFKL-mEGFP clusters are composed of “mobile” enzymes rather than insoluble protein aggregates, regardless of their sizes in Hs578T cells.
Colocalization of PFKL with the other cytoplasmic, rate-limiting enzymes involved in glucose metabolism
From our FRAP analysis, we anticipated that PFKL-mEGFP would be involved in specific or non-specific protein-protein interactions in cluster-positive Hs578T cells. To identify potential protein partners of PFKL in the spatially distinct compartments, we first carried out dual-color colocalization microscopy with other enzymes in glucose metabolism. Fourteen enzymes participate in glycolysis and/or gluconeogenesis. Half of these, including PFKL, catalyze irreversible reactions and thus control the rate-limiting steps in either glycolysis or gluconeogenesis. Our live-cell imaging revealed that three cytoplasmic, rate-limiting enzymes in either glycolysis or gluconeogenesis (i.e. mEGFP-tagged liver-type fructose-1,6-bisphosphatase (FBPase-mEGFP), pyruvate kinase M2 (mEGFP-PKM2), and phosphoenolpyruvate carboxykinase 1 (PEPCK1-mEGFP)) formed coclusters with PFKL-mOFP in both HeLa and Hs578T cell lines (Fig. 3, A–I). The coclustering efficiency between FBPase-mEGFP and PFKL-mOFP among cotransfected Hs578T cells was ∼85%. Importantly, we observed coclusters having various sizes between ∼0.5 and ∼8 μm2 in Hs578T cells, indicating that medium- and large-sized clusters are multienyzme compartments. Colocalization of PFKL-mEGFP with FBPase-mOFP was also monitored after we reversed the fusion of the fluorescent tags in HeLa cells (supplemental Fig. S9). Collectively, this indicates that all four cytoplasmic, rate-limiting enzymes in glucose metabolism (i.e. PFKL, FBPase, PKM2, and PEPCK1) are spatially organized into multienzyme complexes in live human cells.
Figure 3.
The formation of multienzyme metabolic complex by the cytoplasmic rate-limiting enzymes in human glucose metabolism. FBPase-mEGFP (A), mEGFP-PKM2 (D), and PEPCK1-mEGFP (G) were cotransfected with PFKL-mOFP (B, E, and H, respectively) into HeLa cells. All the enzymes were found to colocalize with PFKL; green channels correspond to the mEGFP-fusion constructs, whereas red channels correspond to PFKL-mOFP in the merged images (C, F, and I). Similar colocalization was observed in dually transfected Hs578T cells as well. The direct interaction between FBPase-mEGFP and PFKL-mOFP was measured by FRET in the presence (black line with closed diamonds) and the absence (gray line with open squares) of spatial colocalization of the two enzymes in live Hs578T cells (J). The red arrow indicates the time of acceptor photobleaching. The fluorescent intensity in the donor channel (i.e. green) was collected every 0.5 s for at least 50 s. Scale bar, 10 μm.
Direct interaction between PFKL and FBPase upon colocalization
We further investigated whether the colocalization of these enzymes indicates the occurrence of direct protein-protein associations within the cluster site. Particularly, PFKL and FBPase catalyze the same step of glucose metabolism, though in opposite directions. The activities of both enzymes are reciprocally regulated by a set of allosteric metabolites, and thus it has been proposed that they might directly interact as a means to further this reciprocal control mechanism. However, evidence of this proposed protein-protein interaction had been circumstantial in the 1970s (32–36), only to be overlooked afterward (37). Consequently, we took advantage of our colocalization event of these two enzymes in live Hs578T cells to reexamine this hypothesis. To study their direct interaction in live cells, we measured FRET signals by employing an acceptor photobleaching method under confocal fluorescence microscopy. In this technique, we detect the increased emission of the donor signals upon the photobleaching of the acceptors because of the loss of FRET. Indeed, a direct protein-protein interaction between FBPase-mEGFP and PFKL-mOFP was detected upon their colocalization within the various sizes of the clusters (i.e. ∼0.3 to ∼8 μm2; NFRET = 30) in live Hs578T cells (Fig. 3J, black line with closed diamonds). As a negative control, FRET signals were not detected in the absence of colocalization (Fig. 3J, gray line with open squares). Therefore, our intracellular FRET signals strongly support the idea that the spatial compartmentalization of metabolic enzymes in glucose metabolism would be the means of their intracellular contacts.
Post-translational acetylation of PFKL for the formation of the multienzyme complex
Since a proteomic investigation catalogued the existence of a multitude of acetylated metabolic enzymes in human cells (38), lysine acetylation on pyruvate kinase, phosphoglycerate mutase, and phosphoenolpyruvate carboxykinase has been reported to influence their enzymatic activities or expression levels in human cells (39–43). However, the role of acetylation on PFK remains to be elucidated. We have focused on protein acetylation as a potential regulator of the PFKL clustering events observed in HeLa, Hs578T, Pa04C, and Pa18C cells (Fig. 1 and supplemental Fig. S2). Site-directed mutagenesis was carried out to generate mEGFP-tagged mutants, PFKL-K689R and PFKL-K689A, in which the post-translational acetylation is abolished (38). Interestingly, both PFKL-K689R and PFKL-K689A mutants eliminated clustering, redistributing the enzyme diffusely in HeLa cells (Fig. 4A). When we introduced a PFKL-K689Q-mEGFP mutant whose mutation would mimic the acetylation on Lys-689, we reproduced the formation of the varying sized clusters (Fig. 4B) as wild-type PFKL-mEGFP formed in cells (Fig. 1). We also generated another point mutant of PFKL, PFKL-S529A-mEGFP, which abolishes a post-translational glycosylation event in HeLa cells. Although Ser-529 is identified to play an important role under hypoxic conditions in HeLa cells (3), the S529A mutation did not influence the subcellular distribution of PFKL in HeLa cells whether it is singly or dually transfected with other pathway enzymes (Fig. 4C). Collectively, these data imply that wild-type PFKL-mEGFP proteins appear to be acetylated in our conditions, and the acetylation of Lys-689 in PFKL appears to be important for its clustering in the cytoplasm.
Figure 4.
Site-directed mutagenesis on Lys-689 of PFKL-mEGFP. K689R of PFKL-mEGFP (A) abolished the cytoplasmic clustering in cells. However, acetylation-mimic K689Q mutation (B) produced the formation of various-sized clusters in cells analogous to wild-type. A S529A mutation on PFKL-mEGFP (C) did not alter the cytoplasmic clustering in cells. Scale bars, 10 μm.
In addition, we performed dual-color colocalization microscopy to explore the effect of PFKL acetylation on multienzyme complex formation. First, as we anticipated, the PFKL-K689R-mEGFP mutant did not colocalize with FPBase-mOFP (supplemental Fig. S10). FBPase-mOFP remained diffuse, indicating the importance of Lys-689 acetylation on their colocalization in live cells. Second, we expanded our site-directed mutagenesis studies with FBPase and PKM2 to investigate the potential role of their acetylation events on the colocalization with wild-type PFKL. However, single point mutants of FBPase-mEGFP containing K329A, K329R, K329Q, or K329E did not influence its colocalization efficiency with PFKL-mOFP. Neither did single mutants of mEGFP-PKM2 containing K305A, K305R, K305Q, K62A, or K62Q. We conclude that their acetylation status is not the determinant for their complexation with PFKL. Collectively, we deduce that the acetylation of PFKL is required for the formation of multienzyme metabolic complexes.
Functional characterization of the multienzyme complexes in Hs578T cells
Considering the protein contents and their interaction within medium- and large-sized clusters of multienzyme complexes (Fig. 3), we hypothesized that the different sizes of protein clusters might play functionally different roles in cells. To characterize subcellular functions of spatially resolved multienzyme assemblies in live cells, we have treated PFKL-mEGFP-expressing Hs578T cells with glucose flux regulators and subsequently quantified the number of cells displaying the three differently sized clusters.
First, we have promoted the pentose phosphate pathway in Hs578T cells with two small molecules. Methylene blue has been known to effectively deplete the pool of NADH/NADPH in cells because of its reduction potential, resulting in the promotion of the pentose phosphate pathway (44–46). In the presence of methylene blue (5 nm), we have detected the promotion of medium-sized clusters from small-sized clusters at single-cell levels. Indeed, our quantitative high-content imaging analysis has revealed that 12.3% more cells displayed medium-sized clusters in the presence of methylene blue, whereas cells showing small-sized clusters decreased 15.4% (Fig. 5, A and C). In addition, we have treated Hs578T cells with fructose 1,6-bisphosphate (15 mm), which allosterically inhibits the metabolic activity of PFK in cells while activating FBPase. Such metabolic perturbation is known to shunt glucose flux into the pentose phosphate pathway (47, 48). Excitingly, we have detected that 15.9% more cells formed medium-sized clusters in the presence of fructose 1,6-bisphosphate, whereas cells showing small-sized clusters decreased 12.9% (Fig. 5, A and C). We also confirmed that as a control, the percentage of cells displaying various sized clusters was not changed by treatment of vehicles. These data imply that the ensemble-level promotion of the pentose phosphate shunts by methylene blue and fructose 1,6-bisphosphate in the literature (44–48) appears to be the consequence of the increased population of cells showing medium-sized clusters. Collectively, we conclude that medium-sized clusters are metabolically responsible for shunting glucose flux into the pentose phosphate pathway in cells.
Figure 5.
Functional contributions of medium- and large-sized clusters to cell metabolism. The population (%) of Hs578T cells displaying each size of PFKL-mEGFP cluster was analyzed in the presence of glucose flux regulators. The pentose phosphate shunt was promoted by methylene blue (MB, 5 nm) and fructose-1,6-bisphosphate (F16P, 15 mm), respectively (A and C). In addition, EGF (30 ng/ml) was incubated with Hs578T cells to divert glucose flux into serine biosynthesis (B and C). The error bars indicate the standard deviations of at least three independent experiments. C lists the average percentages (%) of cells displaying the given sized clusters along with their standard deviations (±). Statistical analyses were performed using two-sample two-tailed t test. *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001. N.S., not statistically significant.
Second, we have diverted glucose flux into serine biosynthesis by treatment of EGF. According to the literature (49–52), the EGF treatment up-regulates the activities of PFK and phosphoglycerate dehydrogenase, which catalyzes the first step of serine biosynthesis while simultaneously down-regulating PKM2 activity in cancer cells. The mechanism of EGF action was indeed demonstrated to divert glucose flux into serine biosynthesis (53). When we supplemented 30 ng/ml EGF to Hs578T cells that had been cultured in the RPMI 1640 medium with 10% dFBS, we were able to promote the cell population displaying large-sized clusters from 26.7 to 38.9% while decreasing the number of cells showing medium-sized clusters from 13.4 to 7.7% (Fig. 5, B and C). However, we note here that when Hs578T cells were cultured in the Hybri-Care medium containing 30 ng/ml EGF and 10% non-dialyzed FBS, the population of Hs578T cells showing large-sized clusters was barely changed (i.e. 26.7% in the RPMI1640 versus 23.4% in the Hybri-Care). In addition, non-cancerous Hs578Bst cells did not form large-sized clusters in the Hybri-Care medium in the presence of EGF (Fig. 2). These seemingly contrasting observations can be explained by the fact that the chemical formulations of RPMI 1640 and Hybri-Care are significantly different from each other with respect to the appended compounds of amino acids, vitamins, inorganic salts, metabolites, and cofactors and also that the different levels of nutrients between dialyzed and non-dialyzed FBS are also substantial. Collectively, although we need more in-depth studies with EGF along with these components, we conclude that to increase the net flux of serine biosynthesis at ensemble levels, the cells with large-sized clusters preferentially divert glucose flux into serine biosynthesis.
PFKL-mediated multienzyme assemblies are different from stress granules, aggresomes, and purinosomes
Lastly, we compared the PFKL-mediated multienzyme assemblies with other cellular bodies identified in cultured human cells. Specifically, we included aggresomes formed by chimeric GFP170* and GFP250 proteins (23, 24), stress granules by their scaffold protein, EGFP-tagged the RasGAP-associated endoribonuclease G3BP protein (EGFP-G3BP) (22, 54), and purinosomes (25, 28). As we previously demonstrated with FRAP in Hs578T cells (28), the protein markers for aggresomes and stress granules did not show gradual fluorescent recoveries as a function of time in our conditions. Based on the recovery curve of PFKL-mEGFP in this work (supplemental Fig. S8), it is clear that PFKL-mEGFP clusters are different from aggresomes and stress granules in live cells. However, the fluorescent recovery curve analysis does not differentiate the PFKL-mediated multienzyme clusters from another metabolic complex, the purinosome (25, 28).
To further validate the authenticity of the PFKL clusters, we performed dual-color colocalization microscopy. To begin with, we confirmed that aggresomes and stress granules were spatially different cellular bodies in dually transfected cells with PFKL-mOFP (Fig. 6, A–I). It is important to note that their observed sizes in mammalian cells are comparable with the medium- and large-sized clusters of PFKL-mediated multienzyme assemblies. Also, stress granules were detected in less than 10% of transfected cells, which indicates that our culture conditions do not promote cellular stress. These data imply that PFKL clusters are not a by-product of either protein aggregation or cellular stress in our imaging conditions. Importantly, we demonstrate that PFKL-mOFP did not colocalize with a purinosome marker in Hs578T and HeLa cells (Fig. 6, J and K). Collectively, confocal FRAP analysis and dual-color colocalization microscopy support that PFKL clusters are spatially distinguishable from aggresomes, stress granules, and purinosomes.
Figure 6.
No colocalization of PFKL with other cellular bodies in cells. A–I, EGFP-G3BP (A), GFP170* (D), and GFP250 (G) were cotransfected with PFKL-mOFP (B, E, and H, respectively) in Hs578T cells. Green channels correspond to the EGFP-fusion constructs, whereas red channels correspond to PFKL-mOFP in the merged images (C, F, and I). PFKL-mEGFP formed spatially distinct cellular bodies apart from stress granules and aggresomes. J and K, PFKL-mEGFP and mOFP-tagged formylglycinamidine ribonucleotide synthase were cotransfected into HeLa cells. These two proteins, representing the metabolic complexes in glucose metabolism and de novo purine biosynthesis (i.e. the purinosome), respectively, do not colocalize in cells (J). A random representative region in J was zoomed in for clarification (K). Scale bars, 10 μm, unless otherwise indicated.
Discussion
In this work, we provide compelling evidence that the cytoplasmic, rate-limiting enzymes of glucose metabolism (total of four enzymes; PFKL, FBPase, PKM2, and PEPCK1) are spatially organized into multienzyme complexes in living cells. In addition, the large size of the compartments (i.e. displaying >3 μm2) was detected in various human cancer cells but not in the non-cancerous human breast cell line Hs578Bst. Importantly, the formation of medium- and large-sized clusters at single-cell levels corresponds to the metabolic shunts of glucose flux into the pentose phosphate pathway and serine biosynthesis, respectively, supporting potential functional contributions of the metabolic complexes to cellular metabolism. Collectively, we report the identification of a multienzyme metabolic complex for human glucose metabolism in living cells, which we propose function to regulate the direction of glucose flux in a cluster-size-dependent manner.
At the same time, all the presented experimental evidence strongly indicates that the PFKL-mEGFP clusters are neither insoluble protein aggregates nor fluorescent protein-mediated artifacts. First, all the mEGFP-tagged metabolic enzymes were properly localized either in the cytoplasm or into the organelles based on current understanding of their functional roles in a cell (Fig. 1 and supplemental Figs. S2, S3, and S6) (25, 28). Second, our cluster size analysis with the mean and total fluorescence intensities of transfected PFKL-mEGFP confirmed that PFKL clustering was not governed by the expression level of the tagged protein (Fig. 1 and supplemental Fig. S4, C–E). Third, endogenous PFKL displayed a similar distribution pattern (supplemental Fig. S5) as that observed in transfected cells expressing PFKL-mEGFP (Fig. 1), as well as in other fixed human cells from the Human Protein Atlas database. Fourth, the mEGFP tag itself did not promote the formation of fluorescent clusters in Hs578T cells (supplemental Fig. S7). Fifth, FRAP measurements demonstrated that PFKL-mEGFP, whether located within clusters or not, is mobile in transfected live cells (supplemental Fig. S8). Sixth, FRET signals in live cells revealed that PFKL and FBPase directly interacted upon their colocalization (Fig. 3J), indicating the dynamic nature of fluorescent protein-tagged enzymes inside the clusters. Seventh, dissociation of the assembly was detected by site-directed mutagenesis, which abolished post-translational acetylation on PFKL (Fig. 4). Lastly, the multienzyme complexes were biophysically differentiated from aggresomes, stress granules, and purinosomes (Fig. 6) (28). Collectively, all the evidence provided here strongly support that these PFKL-mediated multienzyme assemblies are not technique-mediated artifacts but rather bona fide compartments utilized in human cells.
The results described above also represent a significant advance upon what has been gleaned to date. Although there are many articles reporting glycolytic enzyme complexes in various organisms over several decades (5–7, 9–12, 14), we have had a very limited understanding of their dynamic properties inside living cells with respect to their metabolic functions. In addition, most of the previous studies have centered on the identification of glycolytic complexes, excluding gluconeogenic enzymes, thus providing limited insights of the interplay between glycolysis and gluconeogenesis upon the protein complexation. Hence, our functional characterization of the multienzyme complex in living cells will advance our understanding of the reversible nature of glucose metabolism and metabolic shunts not only for normal healthy cell metabolism but also for dysregulated cancer cell metabolism.
This multienzyme assembly for glucose metabolism is also analogous to the purinosome assembly, which regulates de novo purine biosynthesis in human cells (25). In terms of their sizes, ∼97% of PFKL-mEGFP clusters were ranged from 0.1 to 3 μm2 in size (supplemental Fig. S4A), which is in good agreement with the ∼97% of purinosome clusters displaying less than 3 μm2 in HeLa and Hs578T cells. Also, the average size of PFKL-mEGFP clusters in HeLa cells, ranging from 0.2 to 1 μm2 (supplemental Fig. S4A, inset), is comparable with the average size of purinosomes in HeLa cells, ranging from ∼0.1 to ∼0.8 μm2 (55). In addition to their sizes observed under wide-field and confocal fluorescence microscopy, we notice that the diffusion coefficients of purinosome-participating enzymes (i.e. 0.007–0.075 μm2/s in Ref. 28) are also in good agreement with the diffusion coefficients of PFKL-mEGFP (i.e. Dapp = 0.018 ± 0.009 μm2/s; supplemental Fig. S8). However, these two complexes are spatially independent metabolic granules in cells (Fig. 6, J and K). Although more advanced analysis may be necessary, it seems that their spatial relationship may reflect their biochemical network (supplemental Fig. S1), providing us with novel insights into how biochemically defined metabolic networks are spatially regulated in a coordinated fashion at the single-cell level.
We may identify the cancer cell-relevant size of clusters formed by the rate-limiting enzymes in human glucose metabolism. Our data suggest that cancer cells seem to promote the clustering events in a larger volume to induce large-sized clusters for their altered metabolic needs (Fig. 1, A-F, versus Fig. 2). Indeed, our experimental data support that large-sized clusters primarily shunt metabolic intermediates into serine biosynthesis at subcellular levels, which is considered one of the hallmarks of altered glucose metabolism in various cancer cells (1, 3, 56, 57). Therefore, although extensive functional studies are required, we propose that the varying sizes of this assembly represent various metabolic roles, like traffic signals, to guide the direction of glucose-mediated carbon flux at various metabolic nodes in the cell.
Collectively, we demonstrate the formation of a multienzyme metabolic complex for glucose metabolism in living human cells, namely the “glucosome.” We envision that comprehensive understanding of such multienzyme metabolic assemblies, the “metabolon,” will open new avenues to address their functional and/or regulatory contributions to human metabolic diseases, like cancer and beyond.
Experimental procedures
Materials
The cDNAs of the human enzymes involved in glucose metabolism were acquired from the PlasmID Repository, the DNA Resource Core at Harvard Medical School. Most cDNAs, except for the plasmids expressing EGFP-tagged hexokinases (Addgene), were amplified by PCR using Pfu DNA polymerase (Stratagene) with pairs of restriction sites on primers. Subsequently, the genes were cloned into either a pmEGFP-N1 plasmid, which possesses an A206K mutation in the EGFP sequence of pEGFP-N1 (Clontech) to produce mEGFP (58) or the pmOFP-N1 (25) plasmid expressing monomeric orange fluorescent protein (mOFP). The resulting cloned plasmids were confirmed by restriction enzyme digestions and DNA sequencing (GeneWiz).
Consequently, we have used the following mEGFP/mOFP-fusion constructs to visualize proteins under fluorescence live-cell microscopy: hexokinase (HK1-EGFP and HK2-EGFP) (59), glucose-6-phosphatase 1 and 3 (G6Pase1/3-mEGFP), liver-type phosphofructokinase 1 (PFKL-mEGFP, mEGFP-PFKL, and PFKL-mOFP), liver-type fructose 1,6-bisphosphatase (FBPase-mEGFP/mOFP), pyruvate kinase M2 (mEGFP/mOFP-PKM2), phosphoenolpyruvate carboxykinase 1 (PEPCK1-mEGFP), and pyruvate carboxylase (PC-mEGFP). As controls, the mEGFP-tagged formylglycinamidine ribonucleotide synthase (FGAMS-mEGFP), EGFP-tagged the RasGAP-associated endoribonuclease G3BP protein (EGFP-G3BP), and the internal segment of the Golgi complex protein 170 fused to GFP (GFP170*) and the C-terminal fragment of p115 fused to GFP (GFP250) were used as protein markers representing purinosome (25), stress granules (22), and aggresomes (23, 24), respectively. EGFP-G3BP and GFP170*/GFP250 were acquired from Drs. J. Tazi (Institut de Genetique Moleculaire de Montpellier, Montpellier, France) and E. S. Sztul (University of Alabama, Birmingham, AL), respectively.
In addition, a TC motif of six amino acids (N-CCPGCC-C) was introduced to PFKL-mEGFP, resulting in PFKL-TC-mEGFP and PFKL-TC. Site-directed mutagenesis was also used to either abolish or mimic reported lysine acetylation sites on human PFKL, FBPase, and PKM2. Single mutants reported here were generated using QuikChange site-directed mutagenesis kits (Agilent).
Cell culture and transfection
Human cervix adenocarcinoma HeLa, human breast carcinoma Hs578T (HTB-126), and human breast normal Hs578Bst (HTB-125) cell lines were obtained from the ATCC. HeLa and Hs578T cells were maintained in the RPMI 1640 (Mediatech, catalog no. 10-040-CV) supplemented with 10% dFBS (Atlanta Biological, catalog no. S12850) and 50 μg/ml gentamycin sulfate. Hs578Bst cells were maintained as recommended in Hybri-Care medium (ATCC, catalog no. 46-X) supplemented with 1.5 g/liter NaHCO3, 10% FBS, 30 ng/ml mouse EGF (Sigma), and 50 μg/ml gentamycin sulfate. In addition, human pancreatic adenocarcinoma Pa04C and Pa18C cell lines were a gift of Dr. Anirban Maitra (Johns Hopkins School of Medicine) and were cultured in minimum essential medium (Mediatech, catalog no. 10-010-CV) supplemented with 20% FBS (Atlanta Biological, catalog no. S11550), 1% l-glutamine (200 mm; Gibco, catalog no. 25030), and 1% penicillin-streptomycin (Gibco, catalog no. 15070-63). The cells were maintained in a HeraCell CO2 incubator (37 °C, 5% CO2, and 95% humidity).
To prepare cells for transfection and subsequent imaging, HeLa, Hs578T, Hs578Bst, Pa04C, and Pa18C cells were gently removed from the culture flask by replacing the culture medium with trypsin-EDTA (Corning, catalog no. 25-053-Cl). Fresh, antibiotic-free growth medium was subsequently used to harvest and resuspend cells that were used to plate either glass-bottomed 35-mm Petri dishes (MatTek) or 8-well chambers (LabTek) such that next-day confluency was ∼70–90%. The following day, the cells were transfected with either Lipofectamine 2000 (Invitrogen) or Xfect (Clontech). For dual transfection, the two plasmids were used in the same transfection mixture whether using Lipofectamine 2000 or Xfect. When using Lipofectamine 2000, the Opti-MEM-I reduced serum medium (Opti-MEM-I; Gibco, catalog no. 11058) was used for transfection, but the medium was exchanged with fresh antibiotic-free growth medium after a 5-h incubation (37 °C, 5% CO2, and 95% humidity), followed by ∼18–24 h of incubation in the incubator. Conversely, Xfect-treated cells in antibiotic-free growth medium did not require a medium exchange and were left in the incubator for ∼18–24 h following the initial transfection.
Fluorescence live-cell imaging
On the day of imaging (∼18–24 h post-transfection), the cells were washed with buffered-saline solution (20 mm HEPES, pH 7.4, 135 mm NaCl, 5 mm KCl, 1 mm MgCl2, 1.8 mm CaCl2, and 5.6 mm glucose) for three 10-min incubations, followed by a ∼1–2-h incubation at ambient temperature. All samples were then imaged at ambient temperature (∼25 °C) with a 60× 1.45 NA objective (Nikon CFI Plan Apo TIRF) using a Photometrics CoolSnap EZ monochrome CCD camera on a Nikon Eclipse Ti inverted C2 confocal microscope. Wide-field imaging was carried out using the following filter sets from Chroma Technology: mEGFP detection by a set of Z488/10-HC clean-up, HC TIRF dichroic, and 525/50-HC emission filter; and mOFP detection by a set of Z561/10-HC clean-up, HC TIRF dichroic, and 600/50-HC emission filter.
For cell-based metabolic flux assays, small molecules, including fructose-1,6-bisphosphate (15 mm) and methylene blue (5 nm), were added to cells after washing three times with buffered-saline solution. Images showing PFKL-mEGFP clusters were acquired before and after cells had been incubated with the small molecules at various time points. Control experiments were also carried out with 1–30 μl of vehicle. In addition, to study the effect of EGF, 30 ng/ml mouse EGF (Sigma) was supplemented to the RPMI 1640 medium along with 10% dialyzed FBS. The EGF effect on PFKL-mEGFP clustering was quantified after cells had been cultured in the EGF-added RPMI 1640 plus dialyzed FBS medium for at least 3 weeks.
Immunocytochemistry
We have also performed immunocytochemistry against endogenous PFKL in HeLa and Hs578T cells. The cells were fixed with freshly prepared 3% formaldehyde, permeabilized with 0.2% Triton X-100, and blocked with 10% normal donkey serum (Jackson ImmunoResearch Laboratories). The cells were then incubated with a rabbit polyclonal anti-PFKL antibody (Thermo Scientific; PA5-21685) and a Cy3-conjugated goat anti-rabbit IgG (H+L) (Jackson ImmunoResearch Lab). Controls for non-specificity and autofluorescence included the fixed cells incubated with primary only, secondary only, and neither antibody.
Fluorescence recovery after photobleaching
FRAP was performed as described previously (28). Briefly, confocal imaging was performed using a JDSU argon ion 488-nm laser line (50 milliwatt) for mEGFP detection via a 488/561 dichroic mirror with 525/50 emission filter and photomultipliers. To photobleach specific areas of interest in live cells, the argon ion 488-nm laser line was applied at 50–75% power for 0.5 s. Because the equipment has better precision when the bleach diameter is larger than 1 μm, the target bleaching area was maintained at ∼3 μm in diameter, regardless of fluorescent cluster size. At least 10 images were obtained before bleaching, and subsequent images were acquired every 0.5 s for at least 50 s. Fluorescence recovery was individually fitted after the degree of background photobleaching was normalized. Apparent diffusion coefficients (Dapp) were then calculated as we have described before (28, 60).
FRET
To measure FRET in live cells, FBPase-mEGFP and PFKL-mOFP were dually transfected into Hs578T cells using Xfect (Clontech). The next day, coclustering Hs578T cells were subjected to measurement of their FRET signals using a confocal microscopy-based acceptor photobleaching method. In this technique, we detected the increased emission of the donors' signals upon the acceptors' photobleaching because of the loss of FRET. Briefly, to photobleach the mOFP-tagged acceptor molecules in coclustering areas, a Coherent sapphire 561/20-nm laser was applied at ∼40% power for 0.5 s. At least 10 images were obtained before the acceptor bleaching, and subsequent images were acquired every 0.5 s for at least 50 s. After the degree of photobleaching of the mEGFP donor molecules was corrected in each data point, the temporal increase of the emission of the mEGFP-tagged donor molecules from the same areas was graphed to reveal their direct interaction in live cells. Dual-color confocal imaging were achieved via a 488/561/640 dichroic mirror with 525/50 and 600/50 emission filters and photomultipliers. Note that there is sufficient spectral overlap between the donor mEGFP emission and the acceptor mOFP excitation for FRET measurement (61).
Cluster size analysis
Cluster size analysis was accomplished using ImageJ processing software (National Institutes of Health). Prior to analysis, fluorescent wide-field images were edited to isolate in-focus, single, whole cells from an image. This was accomplished by cropping the original image and, in some instances, manually outlining the cell to entirely remove surrounding pixel intensity information. The latter step was necessary when cropping alone could not isolate a single cell in a group. Neither affected the original pixel information of the image. Edited images were then processed through ImageJ using a custom script and macro that automates the counting of fluorescent clusters using its built-in module, the so-called robust automatic threshold selection (RATS). Briefly, the images were scaled according to the pixel size of the microscope (i.e. 0.12 μm/pixel) before the RATS segmentation tool was used to automatically identify fluorescent clusters within a cell by outlining. Default parameters for RATS were used in this analysis (i.e. noise threshold = 25, λ factor = 3). Once fluorescent clusters were isolated, the inverse look-up table function was used to generate a mask of the original image that only displayed fluorescent clusters. The module for particle analysis was then applied to this mask to attain both the number and area of fluorescent clusters within an image. This process was repeated for all subsequent cell images. The operator then evaluated the original cell images against the particle mask to eliminate data in which more than one cluster was counted as a single particle. The data were then analyzed and graphed using Microsoft Excel.
Single-cell fluorescence intensity analysis
Nikon imaging software and ImageJ processing software were used to compare the fluorescence intensity between cells in fluorescent images, which were captured using a mercury arc lamp at 75% power with 50-ms exposure time. Briefly, the freehand selection tool was used to manually outline single cells, which were subsequently analyzed for their cluster sizes as described above. In parallel, the cropped raw images were subjected to Nikon imaging software, with which we defined the boundary of cells to quantify the total fluorescent intensities and the size of cells. The mean fluorescent intensities were calculated by dividing the total fluorescent intensities of whole cells with the area of the cells. Of note, the mean or total fluorescence intensities were graphed with the number of clusters per cell or the average size of clusters per cell that we obtained from the cluster size analysis.
Line-scan fluorescence intensity analysis
Nikon imaging software was used to quantify the fluorescence intensity of cells. Using the 12-bit raw images, we drew a user-defined line across a cell, which included at least 10 pixels of background on each end of the line. The fluorescent intensity was then normalized to arbitrary units by setting the background to 0 and the maximum intensity to 100 before analysis.
Author contributions
M. K. and S. A. designed the research; C. L. K., M. K., M. J., D. L. S, E. L. K., S. M. B., and S. A. performed the research; C. L. K., M. J., D. L. S, S. M. B., B. T. L., S. J. R., and S. A. contributed new reagents; C. L. K., M. K., M. J., J. R., and S. A. analyzed the data; and C. L. K., M. K., and S. A. wrote the paper.
Supplementary Material
Acknowledgments
We thank Drs. Stephen J. Benkovic (Pennsylvania State University) and James C. Fishbein (UMBC) for critically reading the manuscript. We also appreciate help from Dr. Bruce Johnson (CUNY Advanced Science Research Center) with computer-aid fluorescent image analysis and thank Anand Sundaram, John Arthur, and Angela Koomson for contributions in site-directed mutagenesis during their training.
This work was supported by UMBC start-up funds (to S. A.). This work was also supported in part by 2016 AACR-Bayer Innovation and Discovery Grant 16-80-44-ANSO (to S. A.); NIGMS, National Institutes of Health Grant T32GM066706 (to C. L. K., M. J., and D. L. S.); NIGMS, National Institutes of Health Grant MARC U*STAR 2T34GM008663 for enhancing minority access to research careers (to S. M. B.); a Howard Hughes Medical Institute undergraduate education grant (to S. M. B.); and NIGMS, National Institutes of Health Initiative for Maximizing Student Development Grant 2R25GM55036 (to E. L. K.). The authors declare that they have no conflicts of interest with the contents of this article. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
This article contains supplemental Figs. S1–S10.
- FBPase
- fructose-1,6-bisphosphatase
- FRAP
- fluorescence recovery after photobleaching
- PFKL
- liver-type phosphofructokinase 1
- mEGFP
- monomeric form of enhanced green fluorescent protein
- dFBS
- dialyzed FBS
- TC
- tetracysteine
- PKM2
- pyruvate kinase M2
- PEPCK1
- phosphoenolpyruvate carboxykinase 1
- mOFP
- monomeric orange fluorescent protein
- RATS
- robust automatic threshold selection.
References
- 1. Christofk H. R., Vander Heiden M. G., Harris M. H., Ramanathan A., Gerszten R. E., Wei R., Fleming M. D., Schreiber S. L., and Cantley L. C. (2008) The M2 splice isoform of pyruvate kinase is important for cancer metabolism and tumour growth. Nature 452, 230–233 [DOI] [PubMed] [Google Scholar]
- 2. Vander Heiden M. G., Lunt S. Y., Dayton T. L., Fiske B. P., Israelsen W. J., Mattaini K. R., Vokes N. I., Stephanopoulos G., Cantley L. C., Metallo C. M., and Locasale J. W. (2011) Metabolic pathway alterations that support cell proliferation. Cold Spring Harbor Symp. Quant. Biol. 76, 325–334 [DOI] [PubMed] [Google Scholar]
- 3. Yi W., Clark P. M., Mason D. E., Keenan M. C., Hill C., Goddard W. A. 3rd, Peters E. C., Driggers E. M., and Hsieh-Wilson L. C. (2012) Phosphofructokinase 1 glycosylation regulates cell growth and metabolism. Science 337, 975–980 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Araiza-Olivera D., Chiquete-Felix N., Rosas-Lemus M., Sampedro J. G., Peña A., Mujica A., and Uribe-Carvajal S. (2013) A glycolytic metabolon in Saccharomyces cerevisiae is stabilized by F-actin. FEBS J. 280, 3887–3905 [DOI] [PubMed] [Google Scholar]
- 5. Giegé P., Heazlewood J. L., Roessner-Tunali U., Millar A. H., Fernie A. R., Leaver C. J., and Sweetlove L. J. (2003) Enzymes of glycolysis are functionally associated with the mitochondrion in Arabidopsis cells. Plant Cell 15, 2140–2151 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Ginger M. L., McFadden G. I., and Michels P. A. (2010) Rewiring and regulation of cross-compartmentalized metabolism in protists. Philos. Trans. R. Soc. Lond. B Biol. Sci. 365, 831–845 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Graham J. W., Williams T. C., Morgan M., Fernie A. R., Ratcliffe R. G., and Sweetlove L. J. (2007) Glycolytic enzymes associate dynamically with mitochondria in response to respiratory demand and support substrate channeling. Plant Cell 19, 3723–3738 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Mamczur P., Dus D., and Dzugaj A. (2007) Colocalization of aldolase and FBPase in cytoplasm and nucleus of cardiomyocytes. Cell Biol. Int. 31, 1122–1130 [DOI] [PubMed] [Google Scholar]
- 9. Mamczur P., Rakus D., Gizak A., Dus D., and Dzugaj A. (2005) The effect of calcium ions on subcellular localization of aldolase-FBPase complex in skeletal muscle. FEBS Lett. 579, 1607–1612 [DOI] [PubMed] [Google Scholar]
- 10. Mowbray J., and Moses V. (1976) The tentative identification in Escherichia coli of a multienzyme complex with glycolytic activity. Eur. J. Biochem. 66, 25–36 [DOI] [PubMed] [Google Scholar]
- 11. Srere P. A. (1987) Complexes of sequential metabolic enzymes. Annu. Rev. Biochem. 56, 89–124 [DOI] [PubMed] [Google Scholar]
- 12. Sullivan D. T., MacIntyre R., Fuda N., Fiori J., Barrilla J., and Ramizel L. (2003) Analysis of glycolytic enzyme co-localization in Drosophila flight muscle. J. Exp. Biol. 206, 2031–2038 [DOI] [PubMed] [Google Scholar]
- 13. Wan C., Borgeson B., Phanse S., Tu F., Drew K., Clark G., Xiong X., Kagan O., Kwan J., Bezginov A., Chessman K., Pal S., Cromar G., Papoulas O., Ni Z., et al. (2015) Panorama of ancient metazoan macromolecular complexes. Nature 525, 339–344 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Wang J., Tolan D. R., and Pagliaro L. (1997) Metabolic compartmentation in living cells: structural association of aldolase. Exp. Cell Res. 237, 445–451 [DOI] [PubMed] [Google Scholar]
- 15. Pette D., Luh W., and Buecher T. (1962) A constant-proportion group in the enzyme activity pattern of the Embden-Meyerhof chain. Biochem. Biophys. Res. Commun. 7, 419–424 [DOI] [PubMed] [Google Scholar]
- 16. Campanella M. E., Chu H., and Low P. S. (2005) Assembly and regulation of a glycolytic enzyme complex on the human erythrocyte membrane. Proc. Natl. Acad. Sci. U.S.A. 102, 2402–2407 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Campanella M. E., Chu H., Wandersee N. J., Peters L. L., Mohandas N., Gilligan D. M., and Low P. S. (2008) Characterization of glycolytic enzyme interactions with murine erythrocyte membranes in wild-type and membrane protein knockout mice. Blood 112, 3900–3906 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Puchulu-Campanella E., Chu H., Anstee D. J., Galan J. A., Tao W. A., and Low P. S. (2013) Identification of the components of a glycolytic enzyme metabolon on the human red blood cell membrane. J. Biol. Chem. 288, 848–858 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Real-Hohn A., Zancan P., Da Silva D., Martins E. R., Salgado L. T., Mermelstein C. S., Gomes A. M., and Sola-Penna M. (2010) Filamentous actin and its associated binding proteins are the stimulatory site for 6-phosphofructo-1-kinase association within the membrane of human erythrocytes. Biochimie 92, 538–544 [DOI] [PubMed] [Google Scholar]
- 20. Rakus D., and Dzugaj A. (2000) Muscle aldolase decreases muscle FBPase sensitivity toward AMP inhibition. Biochem. Biophys. Res. Commun. 275, 611–616 [DOI] [PubMed] [Google Scholar]
- 21. Rakus D., Mamczur P., Gizak A., Dus D., and Dzugaj A. (2003) Colocalization of muscle FBPase and muscle aldolase on both sides of the Z-line. Biochem. Biophys. Res. Commun. 311, 294–299 [DOI] [PubMed] [Google Scholar]
- 22. Tourrière H., Chebli K., Zekri L., Courselaud B., Blanchard J. M., Bertrand E., and Tazi J. (2003) The RasGAP-associated endoribonuclease G3BP assembles stress granules. J. Cell Biol. 160, 823–831 [DOI] [PMC free article] [PubMed] [Google Scholar] [Retracted]
- 23. Fu L., Gao Y. S., and Sztul E. (2005) Transcriptional repression and cell death induced by nuclear aggregates of non-polyglutamine protein. Neurobiol. Dis. 20, 656–665 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. García-Mata R., Bebök Z., Sorscher E. J., and Sztul E. S. (1999) Characterization and dynamics of aggresome formation by a cytosolic GFP-chimera. J. Cell Biol. 146, 1239–1254 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. An S., Kumar R., Sheets E. D., and Benkovic S. J. (2008) Reversible compartmentalization of de novo purine biosynthetic complexes in living cells. Science 320, 103–106 [DOI] [PubMed] [Google Scholar]
- 26. Pedley A. M., and Benkovic S. J. (2017) A new view into the regulation of purine metabolims: the purinosome. Trends Biochem. Sci. 42, 141–154 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Cole R. W., Jinadasa T., and Brown C. M. (2011) Measuring and interpreting point spread functions to determine confocal microscope resolution and ensure quality control. Nat. Protoc. 6, 1929–1941 [DOI] [PubMed] [Google Scholar]
- 28. Kyoung M., Russell S. J., Kohnhorst C. L., Esemoto N. N., and An S. (2015) Dynamic architecture of the purinosome involved in human de novo purine biosynthesis. Biochemistry 54, 870–880 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Hackett A. J., Smith H. S., Springer E. L., Owens R. B., Nelson-Rees W. A., Riggs J. L., and Gardner M. B. (1977) Two syngeneic cell lines from human breast tissue: the aneuploid mammary epithelial (Hs578T) and the diploid myoepithelial (Hs578Bst) cell lines. J. Natl. Cancer Inst. 58, 1795–1806 [DOI] [PubMed] [Google Scholar]
- 30. Schnell U., Dijk F., Sjollema K. A., and Giepmans B. N. (2012) Immunolabeling artifacts and the need for live-cell imaging. Nat. Methods 9, 152–158 [DOI] [PubMed] [Google Scholar]
- 31. Hoffmann C., Gaietta G., Zürn A., Adams S. R., Terrillon S., Ellisman M. H., Tsien R. Y., and Lohse M. J. (2010) Fluorescent labeling of tetracysteine-tagged proteins in intact cells. Nat. Protoc. 5, 1666–1677 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Emerk K., and Frieden C. (1975) Rabbit muscle phosphofructokinase: the effect of the state of the enzyme and assay procedure on the kinetic properties. Arch. Biochem. Biophys. 168, 210–218 [DOI] [PubMed] [Google Scholar]
- 33. Koerner T. A. Jr., Voll R. J., and Younathan E. S. (1977) A proposed model for the regulation of phosphofructokinase and fructose 1,6-bisphosphatase based on their reciprocal anomeric specificities. FEBS Lett. 84, 207–213 [DOI] [PubMed] [Google Scholar]
- 34. Proffitt R. T., and Sankaran L. (1976) Specific, reversible inactivation of phosphofructokinase by fructose-1,6-bisphosphatase. Involvement of adenosine 5′-triphosphate, oleate, and 3-phosphoglycerate. Biochemistry 15, 2918–2925 [DOI] [PubMed] [Google Scholar]
- 35. Söling H. D., Bernhard G., Kuhn A., and Lück H. J. (1977) Inhibition of phosphofructokinase by fructose 1,6-diphosphatase in mammalian systems: protein-protein interaction or fructose 1,6-diphosphate trapping? Arch. Biochem. Biophys. 182, 563–572 [DOI] [PubMed] [Google Scholar]
- 36. Uyeda K., and Luby L. J. (1974) Studies on the effect of fructose diphosphatase on phosphofructokinase. J. Biol. Chem. 249, 4562–4570 [PubMed] [Google Scholar]
- 37. Benkovic S. J., and deMaine M. M. (1982) Mechanism of action of fructose 1,6-bisphosphatase. Adv. Enzymol. Relat. Areas Mol. Biol. 53, 45–82 [DOI] [PubMed] [Google Scholar]
- 38. Zhao S., Xu W., Jiang W., Yu W., Lin Y., Zhang T., Yao J., Zhou L., Zeng Y., Li H., Li Y., Shi J., An W., Hancock S. M., He F., et al. (2010) Regulation of cellular metabolism by protein lysine acetylation. Science 327, 1000–1004 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Hallows W. C., Yu W., and Denu J. M. (2012) Regulation of glycolytic enzyme phosphoglycerate mutase-1 by Sirt1 protein-mediated deacetylation. J. Biol. Chem. 287, 3850–3858 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Lv L., Li D., Zhao D., Lin R., Chu Y., Zhang H., Zha Z., Liu Y., Li Z., Xu Y., Wang G., Huang Y., Xiong Y., Guan K. L., and Lei Q. Y. (2011) Acetylation targets the M2 isoform of pyruvate kinase for degradation through chaperone-mediated autophagy and promotes tumor growth. Mol. Cell 42, 719–730 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Park S. H., Ozden O., Liu G., Song H. Y., Zhu Y., Yan Y., Zou X., Kang H. J., Jiang H., Principe D. R., Cha Y. I., Roh M., Vassilopoulos A., and Gius D. (2016) SIRT2-mediated deacetylation and tetramerization of pyruvate kinase directs glycolysis and tumor growth. Cancer Res. 76, 3802–3812 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Xiong Y., Lei Q. Y., Zhao S., and Guan K. L. (2011) Regulation of glycolysis and gluconeogenesis by acetylation of PKM and PEPCK. Cold Spring Harbor Symp. Quant. Biol. 76, 285–289 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Xu Y., Li F., Lv L., Li T., Zhou X., Deng C. X., Guan K. L., Lei Q. Y., and Xiong Y. (2014) Oxidative stress activates SIRT2 to deacetylate and stimulate phosphoglycerate mutase. Cancer Res. 74, 3630–3642 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44. Delgado T. C., Castro M. M., Geraldes C. F., and Jones J. G. (2004) Quantitation of erythrocyte pentose pathway flux with [2–13C]glucose and 1H NMR analysis of the lactate methyl signal. Magn. Reson. Med. 51, 1283–1286 [DOI] [PubMed] [Google Scholar]
- 45. Lewis I. A., Campanella M. E., Markley J. L., and Low P. S. (2009) Role of band 3 in regulating metabolic flux of red blood cells. Proc. Natl. Acad. Sci. U.S.A. 106, 18515–18520 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46. Martino G., Anastasi J., Feng J., Mc Shan C., DeGroot L., Quintans J., and Grimaldi L. M. (1993) The fate of human peripheral blood lymphocytes after transplantation into SCID mice. Eur. J. Immunol. 23, 1023–1028 [DOI] [PubMed] [Google Scholar]
- 47. Bolaños J. P., Delgado-Esteban M., Herrero-Mendez A., Fernandez-Fernandez S., and Almeida A. (2008) Regulation of glycolysis and pentose-phosphate pathway by nitric oxide: impact on neuronal survival. Biochim. Biophys. Acta 1777, 789–793 [DOI] [PubMed] [Google Scholar]
- 48. Kelleher J. A., Chan P. H., Chan T. Y., and Gregory G. A. (1995) Energy metabolism in hypoxic astrocytes: protective mechanism of fructose-1,6-bisphosphate. Neurochem. Res. 20, 785–792 [DOI] [PubMed] [Google Scholar]
- 49. Borlak J., Singh P., and Gazzana G. (2015) Proteome mapping of epidermal growth factor induced hepatocellular carcinomas identifies novel cell metabolism targets and mitogen activated protein kinase signalling events. BMC Genomics 16, 124. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50. Criscitiello C., Azim H. A. Jr., Schouten P. C., Linn S. C., and Sotiriou C. (2012) Understanding the biology of triple-negative breast cancer. Ann. Oncol. 23, vi13–vi18 [DOI] [PubMed] [Google Scholar]
- 51. Locasale J. W., Grassian A. R., Melman T., Lyssiotis C. A., Mattaini K. R., Bass A. J., Heffron G., Metallo C. M., Muranen T., Sharfi H., Sasaki A. T., Anastasiou D., Mullarky E., Vokes N. I., Sasaki M., et al. (2011) Phosphoglycerate dehydrogenase diverts glycolytic flux and contributes to oncogenesis. Nat. Genet. 43, 869–874 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52. Yang W., and Lu Z. (2013) Regulation and function of pyruvate kinase M2 in cancer. Cancer Lett. 339, 153–158 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53. Possemato R., Marks K. M., Shaul Y. D., Pacold M. E., Kim D., Birsoy K., Sethumadhavan S., Woo H. K., Jang H. G., Jha A. K., Chen W. W., Barrett F. G., Stransky N., Tsun Z. Y., Cowley G. S., et al. (2011) Functional genomics reveal that the serine synthesis pathway is essential in breast cancer. Nature 476, 346–350 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 54. Nadezhdina E. S., Lomakin A. J., Shpilman A. A., Chudinova E. M., and Ivanov P. A. (2010) Microtubules govern stress granule mobility and dynamics. Biochim. Biophys. Acta 1803, 361–371 [DOI] [PubMed] [Google Scholar]
- 55. Chan C. Y., Zhao H., Pugh R. J., Pedley A. M., French J., Jones S. A., Zhuang X., Jinnah H., Huang T. J., and Benkovic S. J. (2015) Purinosome formation as a function of the cell cycle. Proc. Natl. Acad. Sci. U.S.A. 112, 1368–1373 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 56. Anastasiou D., Yu Y., Israelsen W. J., Jiang J. K., Boxer M. B., Hong B. S., Tempel W., Dimov S., Shen M., Jha A., Yang H., Mattaini K. R., Metallo C. M., Fiske B. P., Courtney K. D., et al. (2012) Pyruvate kinase M2 activators promote tetramer formation and suppress tumorigenesis. Nat. Chem. Biol. 8, 839–847 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 57. Chaneton B., Hillmann P., Zheng L., Martin A. C., Maddocks O. D., Chokkathukalam A., Coyle J. E., Jankevics A., Holding F. P., Vousden K. H., Frezza C., O'Reilly M., and Gottlieb E. (2012) Serine is a natural ligand and allosteric activator of pyruvate kinase M2. Nature 491, 458–462 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 58. Zacharias D. A., Violin J. D., Newton A. C., and Tsien R. Y. (2002) Partitioning of lipid-modified monomeric GFPs into membrane microdomains of live cells. Science 296, 913–916 [DOI] [PubMed] [Google Scholar]
- 59. Sun L., Shukair S., Naik T. J., Moazed F., and Ardehali H. (2008) Glucose phosphorylation and mitochondrial binding are required for the protective effects of hexokinases I and II. Mol. Cell. Biol. 28, 1007–1017 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 60. Kang M., Day C. A., Kenworthy A. K., and DiBenedetto E. (2012) Simplified equation to extract diffusion coefficients from confocal FRAP data. Traffic 13, 1589–1600 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 61. Shaner N. C., Campbell R. E., Steinbach P. A., Giepmans B. N., Palmer A. E., and Tsien R. Y. (2004) Improved monomeric red, orange and yellow fluorescent proteins derived from Discosoma sp. red fluorescent protein. Nat. Biotechnol. 22, 1567–1572 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.