**1. Introduction**

During plant development and adaptation to environmental changes, the glycolytic network provides an enormous metabolic flexibility. Thereby, flux regulation is achieved by the fine control of key regulatory enzymes, including pyruvate kinase (PK). PK-mediated synthesis of pyruvate represents a bottleneck for acetyl-CoA entering the TCA cycle. On the other hand, reduced pyruvate kinase activity will lead to a backlog of PEP and other glycolytic intermediates, thereby increasing the flux rate of carbon skeletons into branching biosynthetic pathways. The *Arabidopsis thaliana* genome encodes several putative cytosolic and plastidial PKs, and glycolytic metabolites can be exchanged between the cytosol and plastids [1] since both compartments are connected through diverse transporters located in the inner plastid envelope membrane [2–4]. Despite the assumable key regulatory function of pyruvate kinase so far, only plastidial isoforms have been described [5]. Possible reasons for this are numerous. The high number of isoenzymes with potential redundant physiological roles, as well as the compartmentalized system with glycolytic intermediates equilibrating through plastid membrane transporters, may hamper their investigation.

Several ways of regulation for PKs have been verified, including binding of co-substrates and allosteric effectors. In *Saccharomyces cerevisiae* the glycolytic intermediate fructose-1,6-bisphosphate (FBP) increases the affinity to the bivalent cations, Mg2<sup>+</sup> or Mn2<sup>+</sup>, which are essential for PK activity [6]. Studies on cytosolic PK from castor bean identified glutamate as the most effective inhibitor, whereas aspartate functioned as an activator [7]. Furthermore, the TCA cycle intermediates citrate, 2-oxoglutarate, fumarate and malate have the potential to decrease activity of some plant PKs, which indicates a role as feedback regulators [8–10]. A further regulatory aspect may arise from pH-dependent alterations in the PK enzyme's affinity to metabolite inhibitors. This was proposed in a study on PK enzymes that were isolated from cotyledons of *Ricinus communis* [8]. An enhanced PK activity was accompanied by a reduced cytosolic pH, which was caused by H+-symport that affected the uptake of endosperm-derived sugars and amino acids. Protein degradation is another way of controlling PK activity as shown recently for cotton cytosolic pyruvate kinase 6 (GhPK6) [11]. Here phosphorylation-mediated ubiquitination of GhPK6 appears to modulate the cotton fiber elongation process. Subunit association/dissociation was shown to be an additional mechanism to adjust PK activity. Accordingly, human pyruvate kinase muscle isoenzyme 2 (PKM2) assembled to a dimer has remarkably lower affinity to PEP as the respective tetramer [12]. This regulatory mechanism has been proposed for plant PKs as well, as in vitro studies show that specific subgroup combinations are more active than others [5,7]. Plastidial PKs in *Arabidopsis thaliana* have been shown to be essential for seed oil production, whereby enzyme isoforms form higher-order subunit complexes composed of 4α and 4β-subunits [5].

Spatial distribution of glycolytic enzymes within the cell constitutes a further point of regulation, since enzymes may localize at sites of demand for glycolytic intermediates. Proteomic analyses of highly purified mitochondrial fractions revealed the presence of glycolytic key enzymes, including PK isoforms on the outside of the mitochondrion considered to ensure direct import of pyruvate [13]. Furthermore, degradation by the proteasome determines PK-catalyzed pyruvate synthesis, since C-terminal proteolytic processing of cytosolic PK was shown for isoforms derived from soybean [14]. Finally, the multilayered character of PK activity control generates a complex picture on its role as a flux regulator. On the other hand, the complexity of the involved factors underlines the enormous sensitivity by which fine control is attained.

The *Arabidopsis thaliana* genome encodes 14 putative pyruvate kinases, which are likely to be isoforms catalyzing the ADP-dependent conversion of PEP to pyruvate, thereby releasing ATP. These isoenzymes show a broad diversity concerning gene expression rate and tissue specificity [15] and additionally segregate into plastidial and cytosolic subclades according to consensus predictions of their subcellular localization [16].

We identified five cytosolic PK gene candidates that show a significant expression and are likely to be localized to the cytosol. After having confirmed the cytosolic localization of PK2, PK4 and PK5 by heterologous expression of yellow fluorescent protein (YFP) fusion constructs in *Nicotiana benthamiana*, we aimed to identify the different roles of cytosolic PK enzymes in dependence of changing developmental and environmental conditions. By histochemical analysis of plant lines expressing promoter-β-glucuronidase (GUS) fusion constructs, we observed tissue-specific localization of PK expression in diverse developmental stages. Furthermore, biochemical characterization of purified cPK isoenzymes heterologously expressed in *Escherichia coli* showed that PK activity is controlled by the presence of metabolite effectors or binding of enzyme subunits.

In summary, our findings show that regulation of cPK enzyme activity is controlled by distinct gene expression patterns, different sensitivity to allosteric effectors and enzyme subgroup formation.

#### **2. Results**

#### *2.1. Selection of Pyruvate Kinase Candidates to be Involved in Cytosolic Glycolysis*

The *Arabidopsis thaliana* genome encodes for 14 putative PK isoforms. A phylogenetic tree based on PK amino acid sequence alignment (Appendix A Figures A1 and A2) is shown in Figure 1A. Four PK isoforms, namely At1g32440, At5g52920, At3g22960 and At3g49160, are predicted to contain a chloroplast transit peptide according to the Aramemnon database [16], and their localization to the chloroplast was confirmed in vitro by Andre and colleagues (2007) [5]. For the remaining isoforms, the consensus predictions give no clear indication for targeting to a certain organelle (Figure 1B). An alignment of protein sequences of the *Arabidopsis* PK candidates with bona fide PKs from other organisms revealed two PK subclades to exist in *Arabidopsis* [5]. In addition to isoforms localized to the plastids, another subclade consisting of enzymes that target the cytosol was hypothesized.

**Figure 1.** (**A**) Dendrogram of the *Arabidopsis thaliana* pyruvate kinase family based on protein sequences obtained from the Aramemnon database [16]. The alignment was performed using clustalW, and the dendrogram was created using Dendrocsope 3 [17]. The scale bar represents the number of substitutions per site. Significantly expressed genes are highlighted in red (according to the *Arabidopsis* efp browser). (**B**) Consensus prediction of the subcellular location. Consensus scores for PK proteins obtained from the Aramemnon database [16]. (**C**) Expression of putative cytosolic *PK* genes in roots and leaves. Data obtained from published microarray-data, *Arabidopsis* efp browser [15].

According to the *Arabidopsis* efp browser expression database [15], only five out of ten candidate genes coding for putative cytosolic PKs are expressed up to a reasonable level in order to be considered for our analysis (Figure 1C). Relative expression of these putative cytosolic isoforms is generally higher in roots than in leaves as indicated by published microarray data (Figure 1C). Expressed putative cytosolic pyruvate kinase (cPK) candidates were selected for further analysis and are subject of the current study. For clear determination, the genes were named as follows: At5g08570, *cPK1*; At5g56350, *cPK2*; At5g56350, *cPK3*; At2g36580, *cPK4;* and At3g52990, *cPK5*. PK4 and PK5 expression cannot be distinguished from each other, as both genes are identified by the same probe target. Based on an alignment with bona fide PKs from other organisms, the five expressed PK candidates fall into a subclade of cytosol localized isoforms [5]. As for cPK2, cPK4 and cPK5, the localization prediction was unclear, and YFP fusion proteins were constructed for these enzymes and transiently co-expressed with free mCherry fluorescence protein in leaf epidermal cells of *Nicotiana benthamiana*. Confocal laser scanning microscopy analysis of transformed leaf sections revealed that, in contrast to the triose-phosphate/phosphate translocator (TPT) green fluorescent protein (GFP) fusion, all cPK:YFP fusion proteins were co-localized with free mCherry protein in the cytosol and were absent from the chloroplast (Figure 2).

**Figure 2.** Subcellular localization of pyruvate kinase isoforms. cPK2, cPK4, cPK5-YFP fusion proteins (yellow) and TPT-GFP fusion protein (cyan) were expressed under the control of the cauliflower mosaic virus promoter (*Pro35S*), while free mCherry fluorescence protein was expressed under the control of the ubiquitin promoter (*ProUbi*). All PK-YFP fusion proteins were co-localized with free mCherry in cytosolic plasma strands (white arrows), whereby TPT-GFP fusion protein was co-localized with chlorophyll A fluorescence (green) as indicated by the white asterisks. Constructs were transiently expressed in *Nicotiana benthamiana* leaves. Scale bars = 50 μm.
