**1. Introduction**

Oxygenic photosynthesis is among the most important biological processes on Earth, because it produces the vast majority of organic carbon and nearly all of the atmospheric oxygen. This process is thought to have evolved approximately 2.5 billion years ago in cyanobacteria, and was later conveyed into a eukaryotic host cell via endosymbiosis, giving rise to plastids [1–3]. The phototrophic eukaryotes Glaucophyta, Rhodophyta (red algae), and Viridiplantae (green algae and land plants) form the monophyletic group of the Archaeplastida, which all harbor primary plastids that evolved from a common cyanobacterial ancestor [4] (for an alternative hypothesis of multiple origins and convergent evolution of primary plastids, see [5]). Other eukaryotic algal groups contain plastids, which evolved via secondary- or higher-order endosymbiosis of a green or red algal ancestor [4]. The green lineage (Viridiplantae) comprises two major clades, the "classical" green algae (Chlorophyta), and their sister lineage, the charophyte green algae, from which land plants evolved (Streptophyta) [6,7].

All organisms performing oxygenic photosynthesis fix inorganic carbon as CO2 by Rubisco (Ribulose 1,5-bisphosphate carboxylase). In addition to the carboxylation of ribulose 1,5-bisphosphate (RuBP) with CO2, Rubisco also catalyzes the oxygenation of RuBP, being the first step in a process termed "photorespiration", leading to the appearance of the potent enzyme inhibitor 2-phosphoglycolate (2PG) [8,9]. Despite the evolution of different types of Rubisco with varying substrate affinities and specificity factors, the oxygenase reaction cannot be avoided under the present atmospheric conditions, containing 21% O2 and 0.04% CO2 [10]. As 2PG is toxic for the plant's metabolism [11–13], it is rapidly metabolized to glycolate, which is efficiently metabolized in the photorespiratory pathway [14]. However, this salvage process is energetically costly and recovers only 75% of the organic carbon, while 25% is lost as CO2. It has been estimated that in a crop, C3 plant photorespiration might decrease the yield by approximately 30% under present day atmospheric conditions [15]. Thus, photorespiration is one key target in molecular breeding attempts to improve crop productivity [16,17].

Initially, it was assumed that photorespiration evolved rather late when plants started to colonize the continents and became exposed to the high O2-containing atmosphere with low CO2 (e.g., [18]). In contrast, we hypothesized that photorespiration and oxygenic photosynthesis coevolved in ancient cyanobacteria [14,19]. This hypothesis is based on the discovery of the essential function of the photorespiratory 2PG metabolism under ambient air conditions in the model cyanobacterium *Synechocystis* sp. PCC 6803 [19]. However, the cyanobacterial photorespiratory 2PG metabolism differs from that of plants. First, it is not compartmentalized in the prokaryotic cell. Second, glycolate oxidation is performed by glycolate dehydrogenases among cyanobacteria, and not by glycolate oxidase (GOX), as in plants (Figure 1). Third, in addition to a plant-like 2PG metabolism, *Synechocystis* can also metabolize glyoxylate via the bacterial glycerate pathway and by its complete decarboxylation [19]. The subsequent phylogenetic analysis also revealed that not all enzymes of the plant 2PG cycle originated from the cyanobacterial ancestor. In particular, the photorespiratory enzymes that operate in the mitochondrion of plants are more closely related to enzymes of proteobacteria, the ancestors of this organelle [20,21]. We hence hypothesized that these enzymes, which originally also existed in the cyanobacterial endosymbiont, were replaced by the already established mitochondrial enzymes of a proteobacterial origin [22].

**Figure 1.** Glycolate oxidation can be catalyzed either by a glycolate dehydrogenase (top) or a glycolate oxidase (bottom).

However, in contrast to plants and similar to cyanobacteria, chlorophyte algae, such as *Chlamydomonas reinhardtii*, also use glycolate dehydrogenase for glycolate oxidation, not GOX [23]. The presence of glycolate dehydrogenases in green algae and the later discovery of similar enzymes in cyanobacteria correlated with the existence of an efficient inorganic carbon concentrating mechanism (CCM), which suppresses photorespiration to a great extent in these organisms. Therefore, it was initially assumed that cyanobacteria and chlorophytes with a lower photorespiratory flux due to the CCM prefer glycolate dehydrogenases, which have a higher affinity to glycolate than GOX [24,25]. Accordingly, it was proposed that plant-like photorespiration, including the peroxisomal glycolate oxidation via GOX, only evolved late among streptophytic green algae, and represented an adaption to

a low CO2 and high O2 concentration, which was indeed necessary for the later terrestrialization of the plant kingdom [18].

Recently, numerous new genome sequences became publicly available, which permitted a broader analysis of the distribution of GOXes and glycolate dehydrogenases among different algal lineages. Those searches also revealed that some cyanobacterial genomes possess a GOX-like protein [20]. The subsequent biochemical analysis of the GOX-like protein from the diazotrophic cyanobacterium *Nostoc* sp. PCC 7120 showed that this protein is rather a l-lactate oxidase (LOX). LOX and GOX (here summarized as GOX-like proteins) belong to the group of 2-hydroxy-acid oxidases, display highly similar primary and tertiary structures (see Figure S1), and share the same catalytic mechanism [26–29]. Phylogenetic analysis using GOX-like proteins from heterotrophic bacteria, cyanobacteria, eukaryotic algae, and plants implied that the cyanobacterial GOX-like protein is the common ancestor of all plant GOX proteins, which was consistent with the hypothetical transfer of genes for photorespiratory enzymes into the plant genome via primary endosymbiosis from the previously assumed ancestral N2-fixing cyanobacterium [30]. However, more recent phylogenetic analyses point at the likely uptake of a non-N2-fixing, early-branching unicellular cyanobacterium as a primary endosymbiont [31]. Another study included GOX-like proteins from non-photosynthetic eukaryotes in the phylogenetic analysis, and postulated that plant and animal GOX proteins share a common non-cyanobacterial ancestry [32].

Hence, to solve this controversy and better understand the evolutionary origin of the photorespiratory GOX, we reanalyzed GOX-like proteins using biochemical and phylogenetic approaches. Our results support the view that GOX became part of the photorespiratory metabolism early in the evolution of Archaeplastida, and that a cyanobacterial GOX-like protein most likely served as the ancestor for GOX in all eukaryotic lineages.
