*2.5. Activity of Ancestral GOX Proteins Point to Early Evolution of Preferential Glycolate Oxidation*

To analyze the enzymatic activity of the ancestral N3-GOX protein, which represents a proxy for the common ancestor of GOX-like proteins from Archaeplastida, the gene was synthesized, cloned, and expressed in *E. coli*. The subsequent biochemical analysis was done as described for the extant GOX-like proteins, using glycolate and l-lactate concentrations ranging from 0.2 to 100 mM and 0.5 to 200 mM, respectively. The "ancestral" N3-GOX showed a preference for glycolate as a substrate, as reflected in a higher Vmax and kcat/Km value for glycolate (Figure 5 and Table S2). However, compared with the extant proteins At-GOX2 and Sp-GOX from the streptophyte clade, N3-GOX shows a 30 to 35 times lower Vmax (0.87 μmol min−<sup>1</sup> mg−<sup>1</sup> protein) and a slightly higher kcat/Km with glycolate as a substrate (Figure 5, Table 1, and Table S2). Although the Km value for glycolate is lower, the difference is not significant, which is in contrast to most of the other GOX-like proteins [30,38]. While the proteins from *Cyanidioschyzon* and *Spirogyra* show a similar affinity for l-lactate, the affinity for glycolate is at least 10 times lower in the "ancestral" N3-GOX compared with the extant GOX and LOX proteins (Table 1).

**Figure 5.** Biochemical characterization of the synthetic "ancestral" N3-GOX protein. The Km and Vmax values for glycolate and l-lactate were calculated by non-linear regression implemented in Sigma Plot 13.0 using the Michaelis–Menten kinetic model. For each substrate, three biological replicates were used. Asterisk indicates the significant difference (*p* < 0.05) determined with the two-tailed Student's *t*-test.

## **3. Discussion**

GOX-like proteins belong to the (S)-2-hydroxy-acid oxidase protein family. Besides the important role of GOX in the photorespiration of plants, where it oxidizes glycolate to glyoxylate (Figure 1; e.g., [42]), GOX-like enzymes are also present in animals and heterotrophic prokaryotes, where they can also metabolize other hydroxy acids such as l-lactate, and are additionally involved in fatty acid metabolism or in the detoxification of critical intermediates [43,44]. Despite the differences in their substrate preference [30,32,43,45], all of these enzymes are clearly members of a single gene family and share similar homo-tetrameric quaternary structures (see Figure S1) [46,47].

Their wide-spread occurrence in almost all bacterial and eukaryotic groups makes the emergence of the eukaryotic GOX-like proteins from a bacterial ancestor very likely. Indeed, our phylogenetic analysis (Figure 2) indicates that all eukaryotic GOX-like proteins evolved from an ancestral protein of ancient cyanobacteria. This scenario includes the LOX proteins of extant cyanobacteria, which build the sister group to the clade of eukaryotic GOX-like proteins. This sister group topology is preserved when alternative roots are used for tree building (see Figure S5A,C,D). Thus, the phylogenetic reconstruction supports the view that all eukaryotic GOX-like proteins most likely evolved from prokaryotic ancestors, likely acquired from ancient cyanobacteria. This view, also supported by the close biochemical and protein similarities, is more parsimonious than the assumption of an independent evolution of these proteins among eukaryotes, although it is surprising in the sense that it implies gene transfer from a cyanobacterium to a very early eukaryote. As expected, the topology of the tree shown in Figure 2 shares many features with a previously reported reconstruction [32]. However, the inclusion of important additional taxa and different rooting have significant new implications for our understanding of the evolution of GOX-like proteins.

Placing the root of the tree for GOX-like proteins in between eukaryotic and prokaryotic proteins, as was done by Esser et al. [32], unsurprisingly results in the separation of eukaryotic from all prokaryotic GOX-like proteins (see Figure S5B), and an appearance as if the eukaryotic and prokaryotic proteins would have evolved independently. However, given the evidence that Eukarya are closely related to, and may in fact be derived from Archaea [48,49], the interpretation of the eukaryotic GOX-like proteins as being the native eukaryotic form is difficult to reconcile with the phylogenetic reconstructions. If eukaryotic GOX-like proteins indeed originated from the archaeal ancestor, then related GOX-like proteins should be found in at least some Archaea. However, only a few GOX-like sequences have been found among Archaea, and all of them are restricted to Euryarchaeota. In our midpoint-rooted tree (Figure 2 and Figure S5A), the GOX-like proteins from Archaea cluster with the Firmicutes as the sister group to all other sequences. An analysis of the protein signatures has placed the Firmicutes as an outgroup to most other bacteria [36], although other analyses place them closer to cyanobacteria [50]. Furthermore, the rare occurrence of GOX-like proteins among Archaea and their sister-group relation with Firmicutes suggests that related proteins were present in the last universal common ancestor and lost from other Archaea, or, alternatively, have undergone a horizontal gene transfer (HGT) event. In fact, HGT is a prominent force in prokaryotic and archaeal genome evolution [51]. Placing the GOX-like proteins of Archaea as the outgroup, all bacterial and eukaryotic sequences are found in one clade, where cyanobacterial and eukaryotic proteins are found as sister groups (see Figure S5C).

Hence, our phylogenetic tree strongly implies an origin of GOX in all eukaryotes from a GOX-like protein of the 2-hydroxy-acid oxidase family of ancient cyanobacteria. Such a relation is easily understandable for eukaryotic phototrophs, which evolved because of the endosymbiotic uptake of a cyanobacterial cell that eventually formed the plastids [1]. However, the origin of the GOX-like proteins of chromalveolate taxa and Metazoa from cyanobacteria is more difficult to understand. It is widely accepted that animal (and fungal) lineages separated from the phototrophic eukaryotes before the endosymbiotic engulfment of plastids. Hence, the sister group relationship of animal GOX-like proteins with cyanobacterial proteins is not easily explainable. One possibility is that the placement of this clade is incorrect because of a phylogenetic reconstruction artifact; however, the clade is quite distant from Proteobacteria (mitochondria) and Archaea, the other expected placements for eukaryotic genes obtained from prokaryotes. It is interesting to note that a few other animal proteins appear to be of a cyanobacterial origin. For example, the animal alanine–glyoxylate aminotransferase, which is, like GOX, located in the peroxisome, appears to be derived from cyanobacteria [52]. Moreover, the animal aldehyde dehydrogenases and cytochrome P450 enzymes have their closest relatives with cyanobacterial orthologs [53]. Finally, the photorespiratory glycerate 3-kinase (GLYK) is shared by cyanobacteria, fungi, and all eukaryotic phototrophs [20,54,55]. A simple and parsimonious explanation for the unexpected cyanobacterial protein origins is an early HGT between an ancestral cyanobacterium and the common ancestor of eukaryotes. The occurrence of not only one, but several cyanobacterial proteins in the non-phototrophic eukaryotes could further indicate an earlier endosymbiotic event between a eukaryotic cell and a cyanobacterium, which, however, did not result in a stable establishment of a plastid. Thus, these cyanobacterial genes could be seen as relics from earlier transfers, possibly even ones that somehow prepared the eukaryotic host for the final successful plastid incorporation as suggested by the "shopping bag" hypothesis [56], that is, multiple endosymbiotic gene transfer.

Our results support the inference that the photorespiratory GOX in all phototrophic prokaryotes originated from an ancestral cyanobacterial protein [30]. We verified the presence of a biochemically active GOX in glaucophyte algae, red algae [38], and streptophyte green algae (Table 1). Thus, the consistent utilization of a photorespiratory GOX among all groups of Archaeplastida points to an early evolution of the plant-like photorespiratory cycle in the common ancestor of phototrophic eukaryotes. This notion is also supported by the biochemical analysis of the reconstructed ancestral N3-GOX, which displays some preference of glycolate over l-lactate as a substrate. Previously, it was assumed that the photorespiratory glycolate oxidation via GOX only appeared in Streptophyta, because

chlorophytes such as *Chlamydomonas* perform this reaction by a mitochondrion localized glycolate dehydrogenase [57–59]. The genome sequence of *Chlamydomonas* revealed that, in addition to glycolate dehydrogenase, this chlorophyte also possesses a GOX-like protein. However, biochemical analyses showed that, in contrast to the GOX of all other Archaeplastida, the *Chlamydomonas* enzyme must be assigned as LOX, as it clearly prefers the substrate l-lactate over glycolate, similar to the GOX-like proteins from N2-fixing cyanobacteria such as *Nostoc* (Table 1) [30].

The close similarity of cyanobacterial and chlorophyte LOX proteins is also reflected in our phylogenetic tree, where the chlorophytic LOX proteins form an extra clade at the base of all eukaryotic GOX-like sequences, clearly separated from the GOX clade of all remaining Archaeplastida. There are two possible scenarios how LOX evolved among chlorophytes. The most likely explanation is that the LOX in Chlorophyta reflects a functional reversal from an early GOX; this may reflect a clade that diverged before the ancestral cyanobacterial 2-hydroxy-acid oxidase diversified among Archaeplastida, but may also reflect a phylogenetic artifact stemming from differential selection, depending on the substrate. As mentioned before, all GOX-like enzymes can use l-lactate and glycolate to some degree, and it is to be expected that over the course of evolutionary history, they were optimized to fit the specific metabolic requirements of the respective organism as GOX or LOX. A second possible explanation is that an ancient cyanobacterial LOX received by the endosymbiotic uptake of the plastid ancestor was retained as a second gene in the chlorophyte genome, while it was lost from other algal genomes, or has been obtained by a second HGT from cyanobacteria, as suggested by Esser et al. [32]. Regardless of which scenario actually occurred, LOX in chlorophytes also originated from a protein closely related to those of cyanobacteria.

Furthermore, the sequence and biochemical analyses of the GOX-like proteins among Archaeplastida and cyanobacteria revealed that the experimentally verified amino acid residues that determine whether glycolate or l-lactate is the preferred substrate [30] also distinguish the cyanobacterial and chlorophyte LOX from GOX, such as the Cp-GOX from *Cyanophora*, the Cm-GOX from *Cyanidioschyzon*, and the streptophytic Sp-GOX from *Spirogyra* (Table 2). Interestingly, the hypothetical ancestral N3-GOX also possesses the glycolate-preferring amino acid signature, corresponding to its dominant enzymatic activity as GOX. The amino acid residues responsible for the binding of the flavin mononucleotide (FMN) cofactor are also highly conserved in all GOX and LOX proteins, analyzed here or previously [30,38]. Interestingly, we found two clear trends among the biochemically verified GOX enzymes among Archaeplastida (Table 1). The GOX of the early branching Glaucophyta showed the highest affinity for glycolate among phototrophic eukaryotes, whereas streptophyte GOX proteins have a five-times lower affinity for glycolate. The inverse trend is observed regarding the Vmax values, which are the highest among Streptophytes and the lowest for the enzyme from Glaucophyta. These findings are consistent with the hypothesis that Streptophytes, especially C3 plants, are characterized by much higher photorespiratory fluxes compared with algae, in which photorespiration is often inhibited by the presence of a CCM. Thus, a highly active peroxisomal GOX allows for the rapid degradation of glycolate and its recycling to 3-phosphoglycerate [24,60]. The relatively low Vmax of the N3-GOX may also result from incorrectly predicted amino acid residues, which can lead to protein misfolding [61].
