Strains of both species were selected as representative isolates from raw meat. Two strains per species were chosen to cover their previously reported high intra-species variability (Fuertes-Perez et al., 2019, 2021). P. carnosum TMW 2.2021^T (DSM 105454^T) is the described type of strain of the species (Hilgarth et al., 2018b) that was isolated from MAP raw chicken meat. Strain TMW 2.2149 was previously isolated from MAP pork (Fuertes-Perez et al., 2019). P. phosphoreum strains TMW 2.2103 and TMW 2.2134 were isolated from MAP beef and poultry meat, respectively (Fuertes-Perez et al., 2019).

Strains were inoculated in a pre-culture of meat extract media according to Fuertes-Perez et al. (2019), prepared with 20 g/L meat extract with 20 g/L NaCl and pH 5.8 from the same glycerol stock every time. Pre-cultures were incubated at 15°C overnight in Erlenmeyer flasks for aerobic growth conditions or in gas tight Schott bottles for anaerobic conditions. Cells from the pre-culture were harvested, washed once with 0.85% NaCl (w/v) solution, and re-suspended again in the same solution for further inoculation of the cultures.

Growth of the selected strains was tested on gas tight locked glass bottles filled with 0.4 L of Meat-Simulation-Media (MSM) prepared according to Kolbeck et al. (2019). MSM contains 6% meat extract (w/v) (Merck, Darmstadt, Germany) as the minimum amount at which growth was observed,0.5% glycerol (w/v) (Gerbu Biotechnik GmbH, Heidelberg, Germany), and 0.05 mM Tween80 (Gerbu Biotechnik GmbH, Heidelberg, Germany). Additionally, the media contains 2 μg/ml heminchloride (Roth, Karlsruhe, Germany) dissolved in dimethylsulfoxide (99.8%) (Roth, Karlsruhe, Germany) which was added after autoclaving the media. The pH of the media was adjusted to 5.8 with 100% lactic acid.

Bacteria were inoculated at a start optical density of 0.1 at 600 nm. The growth was monitored by optical density measurement for 48 h or until the stationary phase was reached with constant gas flow, stirring at 120 rpm and at 15°C. Gas mixtures utilized and pumped into the bottles during growth were as follows: (a) air, (b) N₂ (100%), (c) O₂/N₂, (70/30%), (d) N₂/CO₂ (70/30%), (e) O₂/CO₂/N₂ (21/30/49%), and (f) O₂/CO₂ (70/30%). Samples for proteomic analysis were collected by centrifugation (4,000 xg, 5 min, 4°C) of 100 ml of culture during exponential growth, when the calculated amount of cells was above log 7 CFU ml⁻¹. Cells were washed twice with 0.85% NaCl solution, snap-frozen with liquid nitrogen, and stored at −80°C. During the whole sampling process, samples were kept on ice. All experiments were performed in triplicate for each gas atmosphere and each strain.

Optical density measurements obtained for the triplicates of each experiment were used as input for the open source software RStudio (v. 3.3.0) together with the CRAN package grofit (v. 1.1.1-1) to obtain lag-phase (λ), maximum optical density (OD_max), and maximum growth rate (μ_max) of the bacteria. Parameters for the analysis were kept as default. Differences between the mean values of each parameter were analyzed with IBM SPSS Statistics v. 28.0 software (IBM Corp., Armonk, NY) by performing one-way analysis of variance (ANOVA) between the gas atmospheres used for each strain, followed by a post-hoc Tukey test with a confidence interval set at 95% (p < 0.05).

Preparation of samples for proteomics analysis was performed with an in-solution sample processing protocol. Shortly, cells were resuspended in urea lysis buffer [8 M urea, 5 mM EDTA, 100 mM ammonium hydrogen carbonate, 1 mM dithiothreitol (DTT), pH 8.0] and mechanically disrupted with acid washed glass beads on a vortex at maximum speed for 10 min at 4°C. The protein concentration was determined by bicinchoninic acid assay (BCA Protein Assay Kit, Thermo Fisher Scientific, US) according to manufacturer's instructions. A total of 20 μg of protein per sample were reduced (10 mM DTT for 30 mins at 30°C) and carbamidomethylated [55 mM chloroacetamide (CAA) for 30 mins at room temperature in darkness]. Digestion of the proteins was carried out by adding trypsin at a 1/100 enzyme/protein ratio (w/w) for 1 h and afterwards by adding another 1/100 enzyme/protein ratio overnight at 37°C.

After digestion, stage tip purification was performed. Therefore, the pH of the samples was measured (pH < 3) with pH strips (MColorpHast, Merck, GER). The in-house built C18 tips using 3 disks (3M) that were equilibrated consecutively with 250 μl 100% acetonitrile (ACN), 250 μl elution solution [40% ACN,0.1% formic acid (FA)], and 250 μl washing solution (2% ACN,0.1% FA) at 1,500 g. Every sample was loaded on the column (5 min at 500 g), and the sample was three times desalted with washing buffer (2% ACN,0.1% FA) for 2 min at 1,500 g. Finally, peptides were eluted with two times 50 μl elution solution (40% ACN,0.1% FA) for 2 min at 500 g. The solvent of all samples was completely subtracted in a centrifugal evaporator (Centrivap Cold Trap −50, Labconco, US) and freshly suspended before MS measurement in washing solution (2% ACN,0.1% FA) before ~0.1 μg of digest was injected into the mass spectrometer per measurement.

Liquide chromatography tandem mass spectrometry (LC-MS/MS) measurements were carried out on an Ultimate 3,000 RSLCnano system coupled to a Q-Exactive HF-X mass spectrometer (Thermo Fisher Scientific, US). Full proteome analyses were performed by delivering 0.1 μg of peptides to a trap column (self-packed, ReproSil-pur C18-AQ, 5 mm, Dr. Maisch, 20 mm x 75 mm) at a flow rate of 5 μl/min [High Performance Liquid Chromatography (HPLC) grade water with 0.1% formic acid]. Peptides were transferred to an analytical column (ReproSil Gold C18- AQ, 3 mm, Dr. Maisch, 450 mm 75 mm, self-packed) after 10 min of loading, and separated with a 50 min linear gradient that ranged from 4 to 32% of solvent B [0.1% formic acid in acetonitrile and 5% (v/v) dimethyl sulfoxide (DMSO)] at 300 nl/min flow rate. Both nanoLC solvents contained 5% DMSO to boost MS intensity [solvent A = 0.1% formic acid in HPLC grade water and 5% (v/v) DMSO] (Hahne et al., 2013). The Q-Exactive HF-X mass spectrometer was set in data dependent acquisition (DDA) and positive ionization mode during operation. MS1 spectra (360–1,300 m/z) were recorded at a resolution of 60,000 using a maximum injection time (maxIT) of 45 ms and an automatic gain control (AGC) target value of 3 x 10⁶. In case of the full proteome analyses, up to 18 peptide precursors were selected for fragmentation. Precursors with charge state 2 to 6 were the only ones selected and dynamic exclusion of 25 s was enabled. Higher energy collision-induced dissociation (HCD) and normalized collision energy (NCE) of 26% were used for peptide fragmentation. The precursor isolation window width was set to 1.3 m/z. MS2 Resolution was 15,000 with an AGC target value of 1 x 10⁵ and maximum injection time (maxIT) of 25 ms (full proteome).

The software MaxQuant (version 1.6.3.4), with its built-in search engine Andromeda (Cox et al., 2011; Tyanova et al., 2016a), was used to perform peptide identification and quantification. MS2 spectra were searched against the NCBI proteome database of P. carnosum TMW 2.2021^T (NPIB01), TMW 2.2149 (WMDL01), and P. phosphoreum TMW 2.2103 (WMCZ01), TMW 2.2134 (WMCU01), supplemented with common contaminants (built-in option in MaxQuant). Trypsin/P was specified as proteolytic enzyme. Precursor tolerance was set to 4.5 ppm and fragment ion tolerance to 20 ppm. Results were adjusted to 1% false discovery rate (FDR) on peptide spectrum match (PSM) level and protein level by a target-decoy approach that uses reversed protein sequences. A minimal peptide length of seven amino acids was established, and the “match-between-run” function was disabled. Carbamidomethylated cysteine was set as a fixed modification, while oxidation of methionine and N-terminal protein acetylation were set as variable modifications. The proteins, differentially regulated between two growth conditions, were evaluated using the label-free quantification algorithm provided by MaxQuant (LFQ)(Cox et al., 2014). Intensity based absolute quantification (iBAQ) (Schwanhäusser et al., 2011) was carried out to evaluate the expression of proteins within the same sample.

Data processing was performed using the Perseus software (Tyanova et al., 2016b). The workflow included filtering out proteins only identified by site, reverse or from potential contaminants, and performing a log₂ transformation of the values. We only considered proteins that were detected in at least two out of three replicates in each gas condition. For differential protein analysis, we performed Welch t-tests between each pair of gas conditions. Proteins that met the requirements of p < 0.05 and log2 fold change > 2 were considered differentially accumulated. Functional annotation of the proteins was obtained from the databases NCBI, Rapid Annotation Subsystem Technology (RAST) server (Aziz et al., 2008), TIGR annotation (Ouyang et al., 2007), and the Kyoto Encyclopedia of Genes and Genomes (KEGG), and manually curated using BLAST.

We performed six different comparisons between conditions to identify the effects of the following: O₂ (21%) (I. air_vs._N₂); high O₂ (70%) (II. O₂/N₂_vs._N₂, III. Air_vs._O₂/N₂); CO₂ (30%) under anoxic conditions (IV. N₂_vs._N₂/CO₂); and CO₂ under oxic conditions (V. N₂/CO₂_vs._O₂/CO₂/N₂, VI. O₂/CO₂/N₂_vs._O₂/CO₂).

According to a comparative genomics study on photobacteria reported by Fuertes-Perez et al. (2021), all strains that were analyzed encode a complete respiratory chain in their genomes. Figure 2 contains an entire summary of genes present in the genomes of each strain and a summary of protein detection. Some of the respiratory enzymes were not detected in the proteomic data of this study. Cytochrome c oxidase (coxABC, cyoA-E, ccoNOP), cytochrome bc complex (qcrABC), and cytochrome b were absent in all strains under all conditions. Specific subunits of come other complexes were also missing, including nqrD, nuoHJKLMN, cydBX, F0F1 ATP synthase subunit AC, and succinate dehydrogenase subunit CD. While most detected subunits were peripheral, such as ATP synthase subunits α-ε (Jonckheere et al., 2012) and NADH-dehydrogenase sub-units nuoEFG (Falk-Krzesinski and Wolfe, 1998), many of the enzymes that were not detected by proteomics in this study were integral membrane proteins (IMPs). IMPs are notoriously challenging proteins for proteomics analyses due to their low solubility when they contain amphipathic structures and their low accumulation levels (Whitelegge, 2013; Jeffery, 2016; Vit and Petrak, 2017).

We detected under all conditions the non-electrogenic NADH dehydrogenase (ndh), proton-translocating NADH-dehydrogenase complex subunits nuoEFG, Na⁺ translocating NADH-dehydrogenase complex subunits nqrACF, cytochrome bd oxidase subunit cydA, succinate dehydrogenase complex subunits AB, and ATP synthase subunits B and α-ε. The synthesis of an additional proton-translocating NADH dehydrogenase complex by P. phosphoreum might influence the efficiency of the respiratory chain and explain, to some extent, the aerobically faster growth of the species in comparison to P. carnosum. As predicted before by Fuertes-Perez et al. (2021), they use both the non-electrogenic and sodium-translocating version of Complex I, which is in agreement with the sodium requirement of these bacteria (Hilgarth et al., 2018a,b). Additionally, P. phosphoreum also synthesizes the proton-translocating version. We only have evidence of the expression of cytochrome bd oxidase complex. However, it is able to catalyze by itself the complete reduction of O₂ to water and bypass both complex II and III of the respiratory chain (not detected but present in the genome), coupling the generated proton motive force to the ATP synthesis by the ATP synthase complex (Giuffre et al., 2014), and therefore still functioning as a complete respiratory chain.

The effect of the presence of O₂ (air-like conditions) was determined by growth experiments and the comparison of air_vs._N₂ conditions (Table 1). The growth of three of the four strains of photobacteria was positively influenced by the presence of O₂ (21%), with statistically significant (p < 0.05) increase of μ_max and OD_max up to 3 (TMW 2.2021^T) and 4 times (TMW 2.2103), respectively. Strain TMW 2.2149 showed low growth values in all conditions and displayed improvement of only the OD_max. Regarding the proteome, P. carnosum shows a stronger regulatory response to presence/absence of O₂ than P. phosphoreum, indicating a higher amount of differentially accumulated proteins and adaptive mechanisms to the change in environmental conditions (Figure 3, Table 2).

Direct adaptation to aerobic conditions are observed in P. carnosum by upregulation of respiratory chain enzymes succinate dehydrogenase (TMW 2.2021^T), cytochrome bd oxidase (TMW 2.2149), and one copy of the ATP-synthase proton pump. Additionally, we detected a slight increase of abundance of enzymes with oxidoreductase activity in the presence of O₂ for both species to maintain the redox homeostasis of the metabolic machinery. P. carnosum accumulated enzymes of the pyruvate oxidation (TMW 2.2021^T), TCA cycle, and production of lipoic acid under oxic conditions (Figure 3). The lipoic acid is essential as a cofactor for the energy metabolism (including pyruvate dehydrogenase reaction) (Spalding and Prigge, 2010; Solmonson and DeBerardinis, 2018), and serves as an antioxidant against reactive O₂ species (Packer et al., 1995). The accumulation of proteins affecting mentioned pathways is likely aimed at increasing the energetic yield under aerobic atmosphere in order to enhance growth.

The accumulation of O₂ consuming acetolactate synthase enzyme in P. carnosum that strains under oxic conditions is also interpreted as an adaptive mechanism of the bacteria to the environmental gas atmosphere. In addition, the biosynthesis of valine, leucine, and isoleucine is upregulated under oxic conditions for P. carnosum (Figure 3), among the most common amino acids in the proteome of the species (Fuertes-Perez et al., 2021).

Despite detection under all conditions of enzymes superoxide dismutase and catalase/peroxidase in all strains, P. carnosum TMW 2.2021^T already expressed anti-oxidative stress enzymes with 21% O₂. Concomitantly, alkyl hydroperoxide reductase, a primary scavenger of hydrogen peroxide in Escherichia coli (Seaver and Imlay, 2001), was detected in higher amounts than in anaerobic conditions for the same strain. Unlike P. phosphoreum, showing no differential detection, results suggest that P. carnosum has a higher sensitivity and, therefore, has an earlier response to stress. This was also supported by previously predicted higher sensitivity to oxidative stress (Fuertes-Perez et al., 2021) and the demonstrated sensitivity to other types of stress such as high pressure, temperature, and salt concentration (Hilgarth et al., 2018b; Hauschild et al., 2020).

Fatty acid oxidation complex subunits had an enhanced accumulation under aerobic conditions: fadJ, fadB for P. carnosum TMW 2.2149, and fadI for P. phosphoreum TMW 2.2103 and TMW 2.2134). This suggests enhanced utilization of lipids under oxic conditions that provides a higher ATP yield (Leverve et al., 2007). As a consequence, photobacteria will contribute to the rancidity during meat spoilage (Mozuraityte et al., 2016) and also provide free fatty acids by lipase activity for other bacteria, leading to accelerated spoilage.

The effects of high O₂ concentration (70%) could be observed by comparison of growth experiments and the differentially accumulated proteins between the following conditions: I. O₂/N₂_vs._N₂ and air_vs._O₂/N₂ (Table 1). P. carnosum strains show significantly lower μ_max and OD_max values with a higher O₂ availability compared to low O₂ or anoxic conditions. P. phosphoreum displays preference for low O₂ concentrations in all three parameters, but the parameters μ_max and OD_max show significantly higher values under high O₂ conditions compared to anaerobic growth of more than 2 times the value. This might be a result of the already suggested higher sensitivity of P. carnosum to oxidative stress (Fuertes-Perez et al., 2021). Meanwhile P. phosphoreum is able to withstand the stress with minimum growth reduction while benefiting from higher energetic yield of the aerobic metabolism.

Proteins affected by high levels of O₂ were similar to those observed in air-like conditions and, in many cases, even more enhanced by the increase in O₂ concentration. The effect of the presence of O₂ is comparable regardless of concentration of O₂ on the respiratory chain and pyruvate oxidation for P. carnosum, and on the oxidoreductase activity and fatty acid oxidation for both species. Additionally, P. carnosum strain TMW 2.2149 showed enhanced glycerol utilization (glycerol kinase glpK) with increased O₂ concentration (Figure 3).

Iron uptake was upregulated for P. carnosum strains under high O₂ conditions for its utilization in heme- and iron-sulfur biosynthesis that is required for aerobic respiration (Paul et al., 2017). On the other hand, heme utilization protein hutZ and heme carrier protein hutX had a higher accumulation for P. phosphoreum strains which are both part of an operon that binds heme and was suggested to act either as storage for said molecule or to facilitate its traffic from the membrane to proteins (Wyckoff et al., 2004). Finally, we found an increase of accumulation in iron-sulfur cluster assembly proteins for P. phosphoreum strains, namely, cofactors that are required for several essential pathways such as respiration, carbon metabolism, and protection from oxidizing agents (Mendel et al., 2020).

Additionally, we detected an increase in the response to oxidative stress in both species as an upregulation of several preventive enzymes such as alkyl hydroperoxide reductase, DNA starvation/stationary phase protection protein [linked to protection against multiple types of stress including oxidative (Karas et al., 2015)], thiol peroxidase (prevents membrane lipid oxidation (Cha et al., 2004), and superoxide dismutase, catalase, peroxidase, and thioredoxin [antioxidant activity (Koharyova and Kolarova, 2008)]. P. phosphoreum strain TMW 2.2103 also showed upregulation of the histidine biosynthesis pathway (Figure 3), with reported antioxidant and reactive O₂ species scavenger activities (Wade and Tucker, 1998).

Despite the higher availability of O₂, growth appears hindered in all cases when comparing optimum growth at air-like conditions and growth under high O₂ concentration. Results therefore prove that the increase in O₂ concentration does have an inhibitory effect to some extent in photobacteria, which is most likely derived from the increase in oxidative stress. However, growth was still observed. We conclude that high O₂ alone is not able to inhibit photobacteria or prevent their growth to spoilage relevant levels.

Comparisons previously analyzed in order to reveal effect of oxic conditions (air_vs._N₂ and O₂/N₂_vs._N₂) were also the base to determine the effects of growth in absence of O₂. The lack of O₂ appears to have an expected detrimental impact on the growth of photobacteria compared to air-like conditions, particularly on the maximum OD₆₀₀ reached, with the aforementioned exception of P. carnosum strain TMW 2.2149 and its μ_max. We found that P. phosphoreum does accumulate proteins of the respiratory chain under anoxic conditions, which could suggest a compensatory adaptation of the species to the absence of O₂ and, therefore, deviation from the higher energetic yield of aerobic respiration. However, it is important to note that the media (as the meat system) used does not contain alternative electron acceptors, such as trimethylamine N-oxide (TMAO), nitrate, or sulfate, predicted to be used by photobacteria (Fuertes-Perez et al., 2021). Therefore, their absence is likely to contribute to the observed growth reduction due to lacking respiratory activity. This idea is supported by Hilgarth et al. (2018b) who reported similar growth of photobacteria under anaerobic and air conditions on marine agar containing nitrate (Hilgarth et al., 2018b). The removal of O₂ alone is not able to inhibit the growth of photobacteria on meat, but merely limit it (Table 3).

There is an enhanced accumulation of one gene copy of the ATP-synthase proton pump for P. phosphoreum strains (also observed for P. carnosum TMW 2.2021^T), while the other copy was accumulated only for P. carnosum under oxic conditions. Gene duplication in this case might respond to an environmental adaptive strategy, with one copy serving as the main proton pump in optimal oxic conditions, and the other as a compensatory copy under anoxic atmospheres (Kondrashov, 2012).

There is an accumulation of enzymes involved in the use of alternative electron acceptors/donors in both species, which is particularly stronger in P. carnosum strains, many of which were detected under all conditions. We detected upregulation in some of the strains of trimethylamine-N oxide reductase, fumarate reductase, and nitrite reductase in addition to formate dehydrogenase (Figure 4). In particular, nitrite reductase, formate dehydrogenase, and hydroxylamine reductase had a higher accumulation than the rest on the four strains analyzed. Formate is used by bacteria as alternative electron donor and is coupled to the reduction of electron acceptors such as fumarate or nitrate (Ferry, 1990). In addition, cytochrome c (napC/nirT) family protein was also only detected under anoxic conditions for strain TMW 2.2021^T and was previously reported as mediator during anaerobic respiration with nitrate or nitrite using formate as electron donor (Simon et al., 2000). Results suggest that photobacteria use more than one type of electron acceptor during anaerobic respiration. Results also suggest that nitrate/nitrite and formate might be the preferred redox couple. Nitrite and nitrate are both compounds commonly used in meat preservation, even in the European Union, mostly commonly on cured meats. In raw unprocessed meat, the natural availability of nitrate/nitrite is very low, impairing the anaerobic respiration by means of said compounds (Ferysiuk and Wójciak, 2020). However, both nitrate and nitrite are common in water, and their use might be a remaining conserved feature from the common lifestyle of photobacteria as marine bacteria.

Both species show an accumulation of enzymes involved in the use of carbohydrates under anoxic conditions resulting from the lack of O₂ and alternative electron acceptors (no respiration) and the switch to fermentative/sugars utilization pathways. Still, both species also appear to have different preferences for the carbohydrate itself. P. carnosum strains heavily increase the expression of glycogen and maltose degradation/transport pathways (Figure 4). P. phosphoreum strains, on the other hand, due to lack of glycogen and maltose utilization enzymes, regulated mainly the ribose metabolism. Glycogen, ribose, and maltose can be commonly found on raw meat with average values of 1.87 g/kg, 0.5-1 mmol/kg, and 0.02-0.2 mmol/kg, respectively (Koutsidis et al., 2008a,b). The ability of P. carnosum to metabolize the three sugars, in contrast to P. phosphoreum which is only able to utilize ribose, had already been reported (Fuertes-Perez et al., 2021) together with the production of acid from their utilization (Fuertes-Perez et al., 2019).

In addition to the utilization of carbohydrates, we also observed an accumulation of unspecific peptidases/proteases on both species. The lack of alternative electron acceptors present in the media hinders anaerobic respiration, reducing the energetic yield and, in turn, might enhance the diversification of carbon sources in order to increase the total energy output.

We observed an enhanced accumulation of a battery of enzymes involved in pH balance and alkalization that might represent a response to acidification of the media during carbohydrate fermentation, as previously reported by Fuertes-Perez et al. (2019). The carbonic anhydrase (only P. carnosum strains) helps maintain pH homeostasis by interconverting CO₂ and acid (Occhipinti and Boron, 2019). Both the nitrite reductase and hydroxylamine reductase from nitrogen metabolism are accumulated on both species and are able to produce ammonia from nitrite or hydroxylamine to increase the pH (Figure 4). To a lesser extent, we observed an increase in accumulation of choline trimethylamine-lyase (cutC) in one strain (TMW 2.2021^T) of P. carnosum which is able to deaminate choline into trimethylamine (TMA). Choline can be found on raw meat at average levels of 0.7 mg/g (Lewis et al., 2015), while TMA is one of the main spoilage products generated by photobacteria on fish (Dalgaard, 1995). We also observed enhanced detection of anaerobic glycerol-3-phosphate dehydrogenase on P. phosphoreum strains and P. carnosum strain TMW 2.2021^T that catalyzes the production and accumulation of glycerol and helps maintain osmoregulation during osmotic stress conditions (Albertyn et al., 1994).

The comparison of anoxic conditions with and without addition of CO₂ (N₂_vs._N₂/CO₂) allows the determination of the direct effect of CO₂ alone on analyzed photobacteria. In terms of growth, it appears to negatively impact the growth of photobacteria by significantly decreasing the μ_max and the OD_max of all strains in comparison to the rest of conditions (except O₂/CO₂).

Most pathways appear unaffected under anoxic conditions when comparing the presence and absence of CO₂. No common strategy to the strains of each species that was specific to high environmental levels of CO₂ rather than a response to anaerobic metabolism was identified to counteract the presence of CO₂. We only observed strain-specific regulations of single enzymes, such as an increase of trimethylamine-N-oxide reductase (torA) enzyme in P. carnosum TMW 2.2149, producing trimethylamine and contributing to alkalization. P. carnosum TMW 2.2021T strain showed higher regulation of cellular stress proteins. In addition, P. phosphoreum strains TMW 2.2103 and TMW 2.2134 also showed higher accumulation of the glutathione-S-transferase and bifunctional glutathionylspermidine amidase/synthase enzymes (involved in redox sensing and protein S-thiolation) (Pai et al., 2011), respectively, in presence of carbon dioxide.

In conclusion, photobacteria do not show a common adaptation to environmental presence of carbon dioxide alone. We suggest that photobacteria do adapt to CO₂/acidification as a response to their own metabolism and presence/absence of O₂, rather than sensing the environmental levels of CO₂. Consequently, the higher concentration of CO₂ might increase the adverse effect on the bacteria and, since no adaptation to increased stress is performed to counteract the detrimental effect of CO₂, the growth is negatively affected. Still, the modified atmosphere (N₂/CO₂ 70/30%) is unable to prevent the growth of photobacteria (Table 3).

The effects of combined CO₂ and O₂ at air-like conditions were determined by the comparison N₂/CO₂_vs._O₂/CO₂/N₂ and considering the effects of aerobic vs. anaerobic conditions without the presence of CO₂. Presence of air-like O₂ concentration when CO₂ is present appears to benefit the growth (μ_max and OD_max) of all strains when compared to the sole presence of CO₂ anaerobically. However, it is only statistically significant for P. carnosum strains, namely, μ_max of TMW 2.2149 and OD_max of both strains.

When O₂ is once again introduced to the gas mixture in the presence of carbon dioxide, similar regulations are observed as when CO₂ was absent (air vs. N₂). There is an enhancement of oxidoreductase activity, transport of iron, and other metals that might be required for the synthesis of cofactors, pyruvate oxidation, synthesis of lipoic acid, TCA cycle and fatty acid degradation for P. carnosum strains, and enhancement of heme utilization proteins, oxidoreductase activity, iron transport, and assembly of iron-sulfur clusters for P. phosphoreum. Additionally, we observed an increase in oxidative stress and cellular stress proteins on P. phosphoreum strains and P. carnosum TMW 2.2149 strain. The induction of oxidative stress response was absent from most strains when the O₂ concentration was still 21%, but it appears enhanced when the comparison is made in presence of CO₂. These results might suggest a synergistic effect between CO₂ and O₂ that emulates the effects of high O₂ concentrations even at low O₂ percentages (21%) when CO₂ is present. The enhanced effect of the lower O₂ concentration might be tied to the suggested disruptive mechanism of action of CO₂ over the cell membrane (Daniels et al., 1985), allowing a faster diffusion of O₂ into the cell, thereby emulating the effects of the higher O₂ concentration.

We also observed a reduction of accumulation of acid-counteracting reactions as a response to aerobic growth even in presence of CO₂, supporting the idea that photobacteria do not sense the environmental levels of carbon dioxide. The enzyme hydroxylamine reductase had a lower accumulation in presence of O₂ for P. carnosum strains, and so did the enzyme carbonic anhydrase for P. carnosum strain TMW 2.2021^T. The lysine decarboxylase, with lower accumulation levels for P. phosphoreum strains in the presence of O₂ catalyzes the proton-dependent decarboxylation of L-lysine to produce the polyamine cadaverine. It also plays a role in pH homeostasis by consuming protons and neutralizing the acidic by-products of carbohydrate fermentation (Moreau, 2007). The enzyme glutamate decarboxylase, also with a lower accumulation for P. phosphoreum strains, is reported as one of the most efficient methods for growth under acidic conditions via production of γ-aminobutyrate (GABA) for Lysteria monocytogenes (Cotter et al., 2005).

On a note on the spoilage potential of the species, amino acid decarboxylases encoded in the genome of each strain were detected under all conditions. Enzymes arginine decarboxylase (L-arginine to agmatine and CO₂), agmatinase (agmatine to putrescine and urea), and glutamate decarboxylase (L-glutamate to GABA and CO₂) were detected on all conditions for all four strains. Additionally, both P. phosphoreum strains accumulated under all conditions the tyrosine decarboxylase (L-tyrosine to tyramine and CO₂) and lysine decarboxylase (L-lysine to cadaverine and CO₂). Results revealed that regardless of the atmosphere used, photobacteria are able to produce a wide range of biogenic amines and contaminate the raw meat upon growth, as previously predicted by transcriptomics analysis before (Höll et al., 2019).

The effects of the increase in O₂ concentration (up to 70%) compared to air-like conditions in the presence of CO₂ were studied by the comparison O₂/CO₂/N₂_vs._O₂/CO₂ and by considering the effects of increased O₂ concentration alone. The increase in O₂ concentration when CO₂ is present significantly impacts the growth of photobacteria by fully inhibiting P. carnosum and decreasing the growth rate and maximum OD of both P. phosphoreum strains (Figure 1 and Table 3).

Response to oxidative stress between high and low O₂ in presence of CO₂ was the same, contrary to observed results between the same conditions in absence of CO₂. These results again support the idea of a synergistic effect of CO₂ and O₂, and that the response of photobacteria to oxidative stress already reaches its peak with low O₂ concentrations rather than with 70%.

The increase of O₂ concentration in the gas mixture induces the accumulation of multiple proteins on both strains of P. phosphoreum when compared to the low O₂ mixture. The response of both strains appears to be the enhancement of most pathways and reactions in the cells, including the respiratory chain, oxidoreductase activity, alternative electron acceptors and donors, pyruvate metabolism, TCA cycle, glyoxylate cycle, fatty acid degradation, and amino acid metabolism. The response observed in both strains suggests that the combination of high O₂ and CO₂ in the gas mixture is enough to override the stress response of the bacteria. While it is not possible to determine the specific response of P. carnosum due to its lack of growth, we suggest that P. phosphoreum enters a state where survival is prioritized. The species upregulates the entire metabolic machinery as a “panic” reaction against extreme environmental stress. The species might trade off the energy required to maintain such a large enzymatic range for the diversification of energy production or several adaptive mechanisms. The observed growth parameters also suggest that this trade-off might allow the photobacteria to survive under high stress conditions such as those derived from the combined action of high O₂ concentration (oxidative), carbon dioxide (osmotic) and their synergia. However, in said cases, growth is severely hindered due to an energetic yield being either very low or null.

We observed, however, that proteins related to the heme utilization and iron-sulfur cluster assembly were significantly less accumulated in conditions with high O₂ and CO₂ compared to low O₂ and CO₂. This is contrary to what is observed between the two conditions in the absence of CO₂. This phenomenon might be an indication that photobacteria are not able to efficiently use O₂ in this gas mixture despite its higher percentage, and therefore do not fully benefit from the higher yield of aerobic metabolism.

Despite previous reports supporting the reduced growth of photobacteria under modified atmospheres in packaged raw meat (Hauschild et al., 2021), it is still a common niche from which these species are isolated (Fuertes-Perez et al., 2019) in high cell numbers of >10⁸ log CFU/g. While we deliberately chose to study specific strains alone in vitro in this study without any interference or bias by a consortium or variations of substrates, differences in observed growth might be due to the presence of other species of meat spoilers or differences in the model used for growth compared to naturally contaminated raw meat. Spoilage species can have an influence by consumption of O₂ and by reduction of part of the stress induced, as in is the case of B. thermosphacta (Kolbeck et al., 2019), or commensal relationships with photobacteria (Hauschild et al., 2021; Hauschild, 2022). Additionally, the model used in this study, due to the limitations in proteomic sample collection, requires planktonic growth with constant shaking reducing the formation of protective strategies, such as biofilms, that modify diffusion of gases to the cells (Flemming, 1993).