When talking about organelles, it is believed that benefits drive symbiotic relationships. However, it can be pretty challenging to trace the original benefit for which the endosymbiont is retained and integrated into the host cell. It is most likely that photosynthesis was the reason for adopting a cyanobacterial symbiont by a heterotrophic eukaryotic host (Figure 1) because there is no other way to get photosynthesis into the eukaryotic cell lacking it. Photosynthesis is managed by such complex molecular machinery that convergent evolution of the machinery appears highly improbable [6]. While the lateral gene transfer (LGT) likely stands behind the evolution of photosynthesis in prokaryotes [15,16,17], it played, except the endosymbiotic LGT, no role in the evolution of photosynthesis in eukaryotes. Instead, the evolutionary history of eukaryotic phototrophs is full of endosymbiotic events involving prokaryotic (at least twice) or eukaryotic phototrophs (many times) as donors of photosynthetic ability. Consequently, compartmentalization always physically separates photosynthesis from the host cell in eukaryotes (Figure 1). Photosynthesis has been transmitted throughout the tree of life for the apparent reason of the acquisition of a photoautotrophic lifestyle. Although it was supposed for a long time that plastid endosymbiotic events are rare in evolution [18,19], it recently appeared that at least two and six independent events were likely responsible for the appearance of primary and complex plastids, respectively, not counting complex plastids replacements. In addition to the primary endosymbiotic Archaeplastida, a relatively recent event involving heterotrophic amoebae and a cyanobacterium was proposed for the rhizarian Paulinella chromatophora, again with the apparent benefit of photoautotrophy [20,21,22,23,24]. Complex plastids have likely been independently acquired in Euglenophyta, Chorarachniophyta, Alveolata, Stramenopila, Haptophyta, and Cryptophyta [25,26,27,28,29,30,31] (Table 1).

Photosynthesis as a beneficial function is, however, not essential for the host cell‘s survival. Many photoautotrophic lineages became secondarily heterotrophic (Figure 1 and Table 1). Various eukaryotes have lost photosynthesis, being either forced by the lack of access to light or by chemical (e.g., antibiotic) disruption of the photosynthetic molecular machine [11,31,44,45,46,47,48,49,50]. For example, apicomplexan parasites (Sporozoa, Apicomplexa) likely became secondarily heterotrophic because of the easy availability of the organic carbon from the host or by the switch of the ancestral photoparasite from translucent to the opaque host [11]. Additionally, we have to consider that photosynthesis is not only beneficial for the primary producer, as it is quite costly and stands behind the outstanding production of reactive oxygen species (ROS), which can heavily damage the cell [45]. Additionally, a strictly phototrophic lifestyle forces the organisms to live in the access to light and thus limits their environmental distribution.

Therefore, many phototrophs are, in fact, mixotrophs, which can still live heterotrophically. Such organisms may be prone to losing photosynthesis when getting to the nutrient-rich environment or finding a successful heterotrophic strategy, such as predation or parasitism. Moreover, many protists combine the phototrophic and heterotrophic lives to overcome a reduced availability of various compounds present in host or prey but rare in their environment, such as, for example, nitrogen, phosphorus, iron, and sulfur [11]. Therefore, photosynthesis was frequently lost from green parasitic algae [31,44], euglenophytes [46], apicomplexan parasites [11,31,47,48,49,50], and dinoflagellates [48]. Various secondarily heterotrophic strategies, including parasitism, have evolved repeatedly during the evolution of life among former phototrophs [11,38,39]. It is worth noting that such trophic switches are often found in the same taxonomic groups, such as, for example, in myozozans, group of alveolate protists, consisting of dinoflagellates, apicomplexan parasites, and apicomonads (chromerids and colpodellids). Apicomplexans and dinoflagellates are also the only known algae shown to lose their plastids completely. The plastid is absent from the apicomplexan parasitic genus Cryptosporidium [42] and gregarines [51], and the parasitic dinoflagellates of the genus Hematodinium [40]. The plastid losses have happened exclusively in parasites thanks to their ability to scavenge the essential compounds originally produced by plastids from the host (Table 2).

In contrast with plastids, the hypothetical benefit responsible for acquiring mitochondria is the subject of speculation. Frankly speaking, it is not even clear if the mitochondrion was indispensable for forming an early eukaryotic cell [12] or when it was acquired in the course of evolution (early versus late acquisition). The discovery of the eukaryote lacking a mitochondrion proved that the organelle is not essential for the eukaryotic cell as it exists now when the organism is a secondary anaerobic (Figure 2) [61]. Generally, mitochondria are great examples of reductive evolution. The diversity of this organelle involves mitochondria with large circular genomes (e.g., in jakobids [52,53,62]), mitochondria with highly reduced genomes (e.g., those in apicomplexans; [63]), mitochondria-related organelles (MRO), e.g., mitosomes and hydrogenosomes without genomes [60], and, in the end, completely lost mitochondrion (in oxymonads; Table 2 [61]). Consequently, various such organelles possess diverse molecular machinery: the complete respiratory chain of aerobic mitochondria or variously reduced respiratory chains lacking particular complexes: complex I (myzozans, e.g., apicomplexan parasites, and some fungi), complexes III and IV, while complexes I, II, and V retained [64], complexes I and III in chromerids [58,63] and the dinoflagellates of the genus Amoebophrya [59]), or complexes III, IV, and ATP synthase in (some) hydrogenosomes [60]. The missing complexes are substituted by alternative sources of electrons (e.g., alternative NADH dehydrogenases, L- and D- lactate cytochrome c oxidoreductases), or the electron transport chain was even completely lost (MRO).

Mitochondria host a wide range of metabolic functions, with oxidative phosphorylation being the most prominent (Figure 2), although it is found only in classical mitochondrial organelles, and lost from highly reduced MROs. In addition to that, mitochondria can be responsible for the metabolism of amino acids and nucleotides, steroid biosynthesis, heme synthesis, fatty acid catabolism, iron-cluster biogenesis, and many others [60,65,66,67,68]. Such metabolic complexity makes the search for the original beneficial function of the mitochondrion difficult. Various hypotheses have been formulated to explain the primary reason for acquiring a mitochondrion. It is further complicated by extensive genetic rearrangements of the organelle during organellogenesis because a substantial part of the mitochondrial proteome does not originate from the supposed mitochondrial ancestor [60].

The hydrogen hypothesis [69] proposed that the primary benefit of pre-mitochondrial symbiont was hydrogen production for the host, methanogenic Archaea. Some eukaryotes, such as Acanthamoeba castellanii (Amoebozoa), Brevimastigomonas motovehiculus (Rhizaria), Blastocystis spp. (Stramenopila), Nyctotherus ovalis (Alveolata) still contain hydrogen-producing mitochondria with complete (Acanthamoeba) or reduced respiratory chains [60,64,70]. Others host hydrogenosomes, organelles believed to represent modified hydrogen-producing mitochondria [13,60], which have lost the ATP generating part of the respiratory chain and retain just complex I (NADH hydrogenase), or complex II (succinate dehydrogenase) or both [56]. Martin et al. [13] also claim that the mitochondrial ancestor was a facultatively anaerobic bacterium.

Another possible beneficial function of a pre-mitochondrion was proposed by Thomas Cavalier-Smith [71], who speculated that the ancestor of mitochondrion was a photosynthetic purple bacterium, and the primary benefit was photoautotrophy, similar to plastids. This hypothesis assumes that both symbiont and host were facultative aerobes and that the host already has oxidative phosphorylation. A phototrophic symbiont would have an immediate intracellular synergy between a photosynthetic symbiont fixing CO₂ and respiring and a phagotrophic host using oxygen and excreting CO₂ [71].

Other hypotheses suppose that the primary benefit of the mitochondrion is related to dealing with free oxygen in the environment, preferring aerobic heterotrophic respiring bacterium as a mitochondrial ancestor [60]. Oxygen appeared in higher levels during the Great Oxidation Event between 2.4 and 2.1 Bya (billion years ago) [72]. If we take into account the possibility of early acquisition of mitochondrion, the oldest estimates of its appearance touch 2.1 Bya (1655–2094 Mya), while the youngest move around 1 Bya (943–1102 Mya) [73]. One of the earliest views on mitochondrial evolution supposes that the mitochondrion-free anaerobic eukaryotic ancestor acquired aerobic mitochondrion to detoxify oxygen accumulated in the environment after the Great Oxidation Event [74,75]. This scenario would even fit the timing of appearance of eukaryotes between 1 and 2 Bya. However, geochemical data indicate relatively low oxygen levels during the diversification of eukaryotes [72,76]. Moreover, Zimorski et al. [76] argued that those are metabolic processes in the mitochondrion, particularly electron transport chain generating reactive oxygen species, which may harm the cell and must be detoxicated. Therefore, oxidative phosphorylation can hardly be the primary benefit, at least in light of the oxygen detoxification hypothesis. However, we cannot ignore the fact that there is no eukaryote without classical mitochondrion with oxidative phosphorylation known that can permanently live in the presence of oxygen. The use of enzymes that can react with free oxygen does not allow obligate anaerobes to inhabit an oxygen environment. When looking at strictly anaerobic bacteria, they rely on low-potential flavoproteins used for anaerobic respiration. Exposure to oxygen likely causes superoxide and hydrogen peroxide production. They inactivate enzymes with these functional groups through the oxidation of dehydratase iron-sulfur clusters and sulphydryls. However, anaerobes utilize several classes of dioxygen-sensitive enzymes absent from aerobes, which maintain the redox balance during anaerobic fermentation. Their reaction mechanisms require exposure of the solvent of radicals or low-potential metal clusters to the oxygen that can react with it. Additionally, hydroxyl radicals damaging DNA and other biomolecules are generated because hydrogen peroxide oxidizes free iron [77]. Analogously, we can expect that oxygen could not be tolerated by anaerobes involved as the host cell in the early eukaryotic endosymbiotic events. A mitochondrial ancestor could eventually invade the host cell as an energy parasite. The proposed intracellularly parasitic ancestor of the mitochondrion was predicted to bear the ATP/ADP translocase importing ATP from the host. In such a case, there was no beneficial function behind a selection of an endosymbiont (parasite) in the event because the benefit was provided by the host cell [78].