*3.2. Phospho-Site Mutations Suggest Di*ff*erences between Photorespiratory GOX Proteins*

Our studies to decipher photorespiratory GOX regulation by protein phosphorylation appeared to indicate that our different GOX enzymes (*At*GOX1 versus *At*GOX2 and *At*GOX1/2 versus *Zm*GO1) did not always respond in a similar manner thus suggesting subtle differences between them even though they have a conserved photorespiratory role in planta. Both mutations at S212/S213 led to a similar decrease in the glycolate-dependent kcat of *At*GOX2 and *Zm*GO1 while the kcat of *At*GOX1 remained unaltered (Table 2). On the other hand, these mutations did not modify the KM glycolate of the different proteins (Table 2). Therefore, while *At*GOX1 and *At*GOX2 have been shown to have a redundant photorespiratory function [7] and similar kinetic properties ([34], Table 2), these observations suggest that they could have subtle regulatory and structural differences which could potentially explain reported differences in planta [11]. Furthermore, since S212/S213 was found to be important for *At*GOX2 and *Zm*GO1 activity but not for *At*GOX1 activity, we propose that the reduced kcat observed with GOXS212/213D does not reflect the actual consequences of a phosphorylated GOX at S212. It is indeed difficult to explain why *At*GOX1 did not behave in a similar way to the other two GOX proteins since S212 is located in a loop between α-helixD and α-helixE that contains no known residues involved in glycolate oxidase activity. Our structural models showed that S212 was not involved in any H-bond formation; however, when replaced by an aspartate it could form a new H-bond with D54

in a loop between β-sheetA and β-sheetB of a neighbouring GOX subunit, but this was the case for all three proteins.

*At*GOX1/2 T4 and *Zm*GO1 T5 mutations to valine were done to suppress any phosphorylation but also to mimic the sequence found in both *At*HAOX1 and *At*HAOX2, two enzymes that use more efficiently long-chain hydroxyl-acids [6], and *Hs*HAOX2. As previously shown [8], *At*GOX1 and *At*GOX2 were less efficient using L-lactate as a substrate with both enzymes showing an 8-fold increase in KM although the kcat values were not significantly different when compared to glycolate (Tables 2 and 3). However, contrary to the literature [8], both *At*GOX1 and *At*GOX2 were able use 2-hydroxy-octanoate as a substrate but again they were less efficient with lower kcat and increased KM values when compared to glycolate (Tables 2 and 3). Interestingly, the C4-plant enzyme *Zm*GO1 had a similar kcat using either L-lactate or glycolate as a substrate while displaying only a 4-fold increase of KM and even though the kcat with 2-hydroxy-octanoate decreased in a similar manner to *At*GOX proteins, its KM 2-hydroxy-octanoate was unaltered when compared to its KM glycolate (Tables 2 and 3). Thus, even if the kinetics parameters of *Zm*GO1, *At*GOX1 and *At*GOX2 were similar for the glycolate oxidase reaction ([34], Table 2), *Zm*GO1 appeared to be able to use more efficiently both L-lactate and 2-hydroxy-octanoate (as seen from a comparison of kcat/KM ratios).

Differences between GOX proteins with respect to 2-hydroxy-octanoate were also observed when comparing the effect of T4/5 and T158/159 mutations to valine. Indeed *At*GOX1T4V, *At*GOX2T4V and ZmGO1T5V each exhibited a different alteration of KM 2-hydroxy-octanoate compared to their GOXWT counterparts (Table 3). This amino acid being important for GOX quaternary structure [30], its replacement by a valine may modify 2-hydroxy-octanoate specificity towards that of *At*HAOX1 and *At*HAOX2 due to possible subtle differences in 3D structure.

T158V and T159V mutations induced the same modifications of KM and kcat when using either of the three substrates tested thus indicating that the mutation affected the global enzymatic mechanism and it did not appear to be involved in determining substrate specificity (Tables 2 and 3). Finally, *At*GOX1/2T158V and *Zm*GO1T159v proteins were poorly active whatever the substrate used however for each protein the KM appeared to be lower than their GOXWT protein counterparts (Tables 2 and 3). This inferred that T158 was involved in substrate binding as well as enzymatic activity and that when mutated to valine (as in Arabidopsis HAOX proteins) this somehow improved substrate binding. In conclusion to this part of the discussion, we can say that V4/5 and V157/158 containing GOX proteins were not transformed into HAOX enzymes as they were unable to use 2-hydroxy-octanoate more efficiently.

## *3.3. Discovering Conditions Inducing GOX Phosphorylation and Identifying GOX Kinases*

To date, the only evidence of GOX protein phosphorylation has come from phosphoproteomics studies using mass spectroscopy analyses. From Table 1 it can be seen that only the peptide containing phosphorylated T158 has been reported more than once and although the accuracy of the methods in this domain have increased over the years, there are still a number of technical constraints and peptide identification is prone to error. It is therefore important to confirm GOX phosphorylation using other methods and/or confirm by mass spectroscopy using less complex samples instead of total protein extracts. Furthermore, it is necessary to identify conditions that bring about GOX phosphorylation so as to better understand its function. The T4 phospho-site of *At*GOX1 and *At*GOX2 was identified in leaves of plants subjected to oxygen depletion (5% of O2 for 3 h) (Table 1). In such low O2 conditions, photorespiration would be less important and GOX activity could be reduced via T4/T5 phosphorylation. Although the T158 phospho-site has been seen in several different phosphoproteome experiments, no significant differences in the quantity of this phosphopeptide were reported whatever the conditions tested (Table 1), even when comparing dark versus light and low versus high CO2 concentrations that are expected to modulate photorespiratory activity [28]. Thus, T158 phosphorylation may have a role that is not linked to photorespiration. Several studies have implicated GOX in response to pathogen attack [9,35,36], and therefore regulation of GOX activity, together with catalase, may be a

way to modulate H2O2 production as part of plant defence signalling and this could involve protein phosphorylation. Arabidopsis phospho-site S212 was identified in a phosphoproteomic study of the triple kinase mutant *snrk2.2*/*2.3*/*2.6* subjected either to ABA or dehydration treatments [26]. SnRK2.2, SnRK2.3 and SnRK2.6 (also known as OST1) are three kinases activated by ABA [37] and we have recently shown that another photorespiratory enzyme (SHMT1) responded to ABA and altered stomatal movements in response to salt stress [38]. Thus, GOX could be a target of leaf ABA signalling either in mesophyll cells or stomata. Indeed, the quantity of the TL(pS)WK phosphopeptide was shown to increase in response to either ABA or dehydration and this increase was absent in *snrk2.2*/*2.3*/*2.6* seedlings [26].

Therefore, phosphoproteomics studies have had very limited success in providing insights into the eventual role of GOX phosphorylation. Since this post translational modification is a rapid and reversible response to environmental stimuli, experiments should be conducted to identify when GOX is differentially phosphorylated. With respect to GOX photorespiratory function, different conditions expected to modulate photorespiratory flux should be examined such as day/night cycle, variable CO2/O2, and stresses like heat, drought, salt and high light as well as pathogen attack. To simplify the detection of GOX phosphopeptides by mass spectroscopy, GOX should be specifically immunoprecipitated from soluble protein extracts at different times during the stress treatments or during the day/night cycle. In this way, it should be possible to identify conditions where GOX phosphorylation is modulated.

Of course, our in vitro data using recombinant phospho-mimetic GOX proteins only give an indication of what GOX phosphorylation might actually be doing with respect to enzyme activity. Indeed, GOX was produced in *Escherichia coli* as a recombinant phospho-mimetic protein and therefore we do not know whether this led to a non-physiological alteration of GOX structure (and activity). Since the modification was constitutive and present as soon as the protein was synthesized, it does not reflect a reversible phosphorylation of a protein. For instance, is the absence of FMN in GOXT4/5D and GOXT158/159D due to structural changes induced by the aspartate that inhibits FMN entry into the apo-protein to form the halo-protein, and therefore would phosphorylation lead to changes that favour FMN removal from the halo-protein? Furthermore, phospho-mimetic mutations do not always lead to the modifications in protein function brought about by an actual phosphorylation. Thus, to assess the real role of GOX phosphorylation, identification of the protein kinases (and also the phosphatases) involved would be crucial. Based on the SUBA database (http://suba.plantenergy. uwa.edu.au/), 33 protein kinases are predicted to be addressed to the peroxisome. Using machine learning methods to identify proteins carrying plant peroxisomal PST1 targeting signals, 11 protein kinases were identified as being addressed to the peroxisome [39]. More recently, a list of about 200 confirmed Arabidopsis peroxisomal proteins was compiled and four kinases and eight phosphatases or phosphatase subunits were identified [40]. A strategy to identify *At*GOX1/2 kinases could be to retrieve knock-out mutant lines for predicted and/or verified peroxisomal protein kinases and compare *At*GOX1/2 phosphorylation status by mass spectroscopy after *At*GOX1/2 immunoprecipitation from soluble proteins extracted from wild-type and mutant plants treated to previously identified conditions that induce *At*GOX1/2 phosphorylation.

In conclusion, photorespiratory GOX has the potential to be regulated by protein phosphorylation at several distinct sites. In general, phospho-mimetic recombinant GOX proteins exhibited reduced activities and this could be explained by predicted changes in structural interactions affecting key residues involved in FMN binding and catalysis. Phospho-mimetic GOX did not show any alteration in substrate specificity although C4-plant *Zm*GO1 did appear to have a relaxed substrate specificity when compared to C3-plant *At*GOX1 and *At*GOX2.
