Metabolic Characteristics of Porcine LA-MRSA CC398 and CC9 Isolates from Germany and China via Biolog Phenotype MicroArrayTM

Livestock-associated methicillin-resistant Staphylococcus aureus (LA-MRSA) is an important zoonotic pathogen, often multi-resistant to antimicrobial agents. Among swine, LA-MRSA of clonal complex (CC) 398 dominates in Europe, Australia and the Americas, while LA-MRSA-CC9 is the main epidemic lineage in Asia. Here, we comparatively investigated the metabolic properties of rare and widespread porcine LA-MRSA isolates from Germany and China using Biolog Phenotype MicroArray technology to evaluate if metabolic variations could have played a role in the development of two different epidemic LA-MRSA clones in swine. Overall, we were able to characterize the isolates’ metabolic profiles and show their tolerance to varying environmental conditions. Sparse partial least squares discriminant analysis (sPLS-DA) supported the detection of the most informative substrates and/or conditions that revealed metabolic differences between the LA-MRSA lineages. The Chinese LA-MRSA-CC9 isolates displayed unique characteristics, such as a consistently delayed onset of cellular respiration, and increased, reduced or absent usage of several nutrients. These possibly unfavorable metabolic properties might promote the ongoing gradual replacement of the current epidemic LA-MRSA-CC9 clone in China with the emerging LA-MRSA-CC398 lineage through livestock trade and occupational exposure. Due to the enhanced pathogenicity of the LA-MRSA-CC398 clone, the public health risk posed by LA-MRSA from swine might increase further.


Biolog PM Assay Performance and Data Evaluation
Materials, chemicals, instrumentation and consumables for the Biolog PM assays were obtained from Biolog Inc., Hayward, CA, USA. The Biolog PM assays were conducted using the 96-well PM1 (carbon utilization), PM2A (carbon utilization), PM9 (osmotic and ionic effects), and PM10 (pH effects) microplates, which represent different metabolic capabilities and/or growth conditions [24]. The experiments were carried out at least in independent duplicates according to the manufacturer's PM procedures for Gram-positive bacteria with slight modifications of the incubation conditions. Bacteria from the frozen stock cultures were grown overnight at 37 • C on BUG + B (Biolog Universal Growth Medium + 5% sheep blood) agar plates. After subculturing for a second time, bacterial cells were removed from the BUG + B plate and transferred into sterile IF-0a GN/GP Base fluid. Uniform cell suspensions with a final cell density of 81% T (transmittance) were combined with the respective inoculating fluid depending on the microplate to be used, according to the manufacturer's recommendations. The recipes varied slightly in the composition of IF-0a GN/GP Base and IF-10b GN/GP Base fluid, the redox dye tetrazolium violet (Dye H), and different additive solutions. The Biolog PM microplates were inoculated with the final cell suspension (100 µL/well) and incubated at 37 • C in the OmniLog incubator. The OmniLog PM system (OmniLog Data Collection software version 2.3.01) recorded the colorimetric data every 15 min for 48 h. In order to detect abiotic "false-positive" reactions, an additional set of control Biolog PM microplates (PM1, PM2A, PM9, PM10) was incubated using the same assay protocol described above but without the bacterial cells.
The final kinetic Biolog PM data were processed and evaluated using OmniLog Data File Converter, OmniLog FM (version 1.20.02), OmniLog PM (version 1.20.02), and Biolog Data Analysis (version 1.7.1.51) software. Kinetic graphs obtained from the colorimetric reaction in each well were designated as "respiration curves" as they display the bacterial cells' respiration over time. Respiration curves obtained after 24 and 48 h of incubation were compared visually between the four investigated groups. Moreover, metabolic substrate utilization was quantified via the parameters: area under the curve (AUC), lag time (LT), maximum height (MH), plateau time (PT) and slope of the respiration curves. Two statistical approaches were applied to obtain possible selections of the most informative substrates and/or conditions tested. Firstly, sparse partial least squares discriminant analysis (sPLS-DA) [47] was employed considering two independent AUC measurements obtained after 24 h of incubation for each isolate and all wells/conditions investigated. sPLS-DA provides variable selection (Lasso) for datasets with a large number of highly correlated variables [47]. The optimal number of selected variables was determined using 10-fold cross validation. This approach revealed that four components containing 30 of the 379 substrates and/or conditions provided optimal discriminatory power to differentiate the four MRSA groups. Components combine similarly behaving variables into groups (instrumental variables). This aids in selecting variables according to recurring patterns and reduces the risk of selecting variables based on randomly occurring differences in 'omics datasets. Based on the selected variables, a heatmap was generated consisting of one row per isolate and one column per substrate. The colored tiles represent the average z-score for each isolate and substrate combination. A hierarchically clustered tree was generated using the Euclidian distances between isolate (rows) and substrate (columns) z-scores. Secondly, sPLS-DA [47] analysis was carried out for each of the parameters LT, MH, PT and slope considering independent duplicate measurements obtained after 24 h of incubation for each isolate and all wells investigated. Parameters were ranked according to variable importance in each of the analysis for every parameter. Here, 40 of the 379 substrates and/or conditions tested were selected based on the lowest combined rank between all four analyses. This combined sPLS-DA variable selection was performed since AUC values can differ due to varying reasons, such as clearly different curves or just minor changes in the maximum height. This analysis favors variables with differences in respiration curves in more than one curve parameter. All of the statistical analyses were performed using the R package mixOmics (version 6.18.1) [48] and R version 4.1.3 (R Foundation for Statistical Computing, Vienna, Austria).

Molecular and Phylogenetic Analyses
Molecular typing results and antimicrobial resistance data are displayed in Table 1. All porcine LA-MRSA from Germany and China displayed spa and SCCmec types that are typically associated with CC9 and CC398, respectively [8,21,49]. An exception is spa type t571, which was found in isolate SHP1 and has so far rarely been reported in MRSA sequence type (ST) 398 isolates [50,51]. However, a recent study has shown that swineassociated MRSA ST398 circulating in a slaughterhouse in China were predominantly associated with spa type t571 [17]. The methicillin-susceptible S. aureus (MSSA) t571 clonal lineage also predominated in swine, human, and environment-associated isolates [17]. Moreover, MSSA of ST398 and spa type t571 have been reported from human infections in Belgium [52], China [53], Colombia [54], France [55,56], Germany [51], the Netherlands [50], and the United States [57]. It is worth noting that several of the affected patients had no or only indirect contact with livestock prior to infection. Interestingly, spa type t571 harbors only one repeat (r24) less than type t034, which is commonly detected in MRSA-CC398.
The antimicrobial resistance data were used for selection of representative isolates for Biolog PM assays as described above. In general, antimicrobial resistance patterns of LA-MRSA from China were more complex than those of LA-MRSA from Germany ( Table 1). All tested isolates harbored several antimicrobial resistance genes, which differed in number and distribution among the isolates. The Chinese isolates displayed higher numbers of antimicrobial resistance genes, especially the CHN-MRSA-CC9. As MRSA, all isolates were resistant to beta-lactams and carried the mecA gene as well as at least one copy of the blaZ gene. In addition, all CHN-MRSA-CC9 and -CC398 were resistant to tetracycline, erythromycin, clindamycin and tiamulin. All GER-MRSA-CC398 furthermore revealed tetracycline resistance. However, all isolates were susceptible to linezolid and vancomycin and the German isolates additionally to florfenicol and tiamulin. The isolates' resistance profiles mirror the selection pressure imposed by antimicrobial agents commonly used in the pig industry. Usage of tetracyclines, tiamulin, florfenicol, as well as macrolides, lincosamides and streptogramins (MLS) is associated with selection of multi-resistant isolates in China [58]. In Germany, MRSA isolates have adapted to selection pressure imposed by tetracyclines by acquiring the respective antimicrobial resistance genes [59].
The cgMLST analysis revealed two different clusters, in accordance with the MLST and CC assignments (Table 1), and 20 varying allelic profiles for the 20 MRSA isolates ( Figure 1, Table S1). Cluster 1 comprised the CC398 isolates and included five rather closely related allelic profiles each from Germany and China. The GER-MRSA-CC398 differed in 14 to 42 and the CHN-MRSA-CC398 in 8 to 241 target genes (Table S1). The two most closely related German (DG29) and Chinese (YN523) CC398 isolates varied in 103 target genes. Cluster 2 represented the CC9 sequences, also assigned to five more closely related allelic profiles each from Germany and China. The GER-MRSA-CC9 varied in 25 to 41 and the CHN-MRSA-CC9 in 43 to 108 target genes (Table S1). The two most closely related German (DG39) and Chinese (DL44) MRSA-CC9 differed in 191 target genes. According to a previous study on the phylogenetic analysis of three MRSA outbreaks applying the SeqSphere+ cgMLST approach, S. aureus isolates displaying 0 to 8 allelic differences should be considered as related; isolates with 9 to 29 allelic differences as possibly related; and those with ≥30 allelic variations as unrelated [60]. Thus, two CHN-MRSA-CC398 seemed to be related, while four GER-MRSA-CC398, four GER-MRSA-CC9 and three CHN-MRSA-CC398 showed varying possible relationships among each other (Table S1). Moreover, the differences in 1559 of 1749 alleles included in the comparison reflected the considerably distant clonal relationship between Cluster 1 and 2. As the isolates SHP1 within CHN-MRSA-CC9, and relatively higher numbers of allelic differences between the groups, the division of isolates into four groups for the Biolog PM assay based on their CC and origin was in accordance with the phylogenetic analysis ( Figure 1, Table S1).
three CHN-MRSA-CC398 showed varying possible relationships among each other (Table  S1). Moreover, the differences in 1559 of 1749 alleles included in the comparison reflected the considerably distant clonal relationship between Cluster 1 and 2. As the isolates SHP1 within CHN-MRSA-CC9, and relatively higher numbers of allelic differences between the groups, the division of isolates into four groups for the Biolog PM assay based on their CC and origin was in accordance with the phylogenetic analysis ( Figure 1, Table S1).

Biolog PM Assay Analysis
If not further specified, all findings described in this section relate to an incubation period of 24 h. Some of the chemical compounds used in the Biolog PM microplates can directly reduce Dye H even when no bacterial cells are added resulting in the formation of purple color [24]. These wells will give abiotic "false-positive" reactions without bacteria present because the dye is chemically reduced. There is no final list of abiotic wells available because the dye reduction depends on the definitive test parameters. Therefore, a set of Biolog PM microplates was run without bacteria. Wells that showed abiotic reactions under the study's experimental conditions are given in Table S2. Comparison of the obtained curves with the reduction kinetics resulting from the Biolog PM assays involving bacterial cells enabled assessment of the "false-positive" reactions. Biological reduction was gradual over time, whereas abiotic reduction occurred rapidly, more like a step function. In addition, biological exceeded chemical reduction in the abiotic wells. In the end, abiotic reactions did not affect further evaluation of the Biolog PM assay data as they were easy to distinguish from biological reduction due to bacterial respiration.

Biolog PM Assay Analysis
If not further specified, all findings described in this section relate to an incubation period of 24 h. Some of the chemical compounds used in the Biolog PM microplates can directly reduce Dye H even when no bacterial cells are added resulting in the formation of purple color [24]. These wells will give abiotic "false-positive" reactions without bacteria present because the dye is chemically reduced. There is no final list of abiotic wells available because the dye reduction depends on the definitive test parameters. Therefore, a set of Biolog PM microplates was run without bacteria. Wells that showed abiotic reactions under the study's experimental conditions are given in Table S2. Comparison of the obtained curves with the reduction kinetics resulting from the Biolog PM assays involving bacterial cells enabled assessment of the "false-positive" reactions. Biological reduction was gradual over time, whereas abiotic reduction occurred rapidly, more like a step function. In addition, biological exceeded chemical reduction in the abiotic wells. In the end, abiotic reactions did not affect further evaluation of the Biolog PM assay data as they were easy to distinguish from biological reduction due to bacterial respiration.

General Metabolic Properties of Porcine LA-MRSA from Germany and China
Overall, the porcine MRSA isolates were able to metabolize a broad variety of macromolecules, including a wide spectrum of different carbon sources. These nutrients comprised sugars (D-galactose, D-trehalose, D-mannose, etc.), sugar alcohols (sorbitol, adonitol, maltitol, etc.) or (di-, tri-, hydroxy-) carboxylic acids (formic acid, succinic acid, citric acid, glycolic acid, etc.). Furthermore, the bacteria utilized the emulsifier tween, a wide range of amino acids, a number of other acids (a-keto-valeric acid, caproic acid, quinic acid, etc.) and nucleosides such as adenosine, thymidine or uridine. S. aureus requires endogenous and/or exogenous sources of biosynthetic precursors, e.g., carbohydrates, amino acids, nucleic acids, for assembly reactions generating DNA, RNA and proteins [61]. Moreover, in S. aureus the de novo synthesis of precursors requires 13 biosynthetic intermediates, which are provided by central metabolic pathways [61]. Central metabolism is both catabolic, i.e., energy-generating, and anabolic, i.e., energy-requiring, and involves glycolysis, gluconeogenesis, the pentose phosphate pathway and the tricarboxylic acid cycle in S. aureus [61]. In line with our findings, it is known that S. aureus is able to catabolize many different macromolecules, e.g., amino acids or nucleic acids, to produce the 13 intermediates, but typically prefers to use carbohydrates when available [61]. The process of carbohydrate translocation into the cytoplasm involves a group of different enzymes/transporters, where specificity for a certain sugar is provided by particular components. However, S. aureus has at least 15 enzyme II components encoded within its genome, which also highlights the pathogen's versatility in carbohydrate utilization, consistent with our results [61]. In some cases, substrate metabolism was slightly delayed (α-keto-butyric acid, N-acetyl-neuraminic acid, 2-hydroxy benzoic acid, etc.), or was reduced compared to other substances (i-erythritol, citraconic acid, D-tartaric acid). Only a few carbon sources could not be metabolized at all by the MRSA isolates, including glyoxylic acid, capric acid and itaconic acid, which emphasizes the remarkable versatility in nutrient utilization.
S. aureus has also evolved a number of cellular stress responses allowing it to persist in different host species and to survive in varying environmental settings [62]. The bacterium's high osmotolerance, for example, supports growth in high-osmolarity body sites, e.g., skin and mucosal surfaces, or in food with high salt concentrations [62]. Consistent with this, the porcine LA-MRSA from Germany and China showed pronounced metabolic activity in the presence of many osmolytes in various and even high concentrations. In several studies, the role of cytoplasmic accumulation of K + and solutes as glutamine or proline in the adaptation to settings with increased osmolarity is discussed [62]. Here, all isolates were resilient to increasing concentrations of potassium chloride, sodium sulfate and ethylene glycol up to 6%, 5% and 20%, respectively. Their metabolism was also invariable to sodium phosphate and ammonium sulfate in all concentrations tested. Except for a minor delay, increasing concentrations of sodium chloride up to 10% had no substantial effect on bacterial metabolic activity. Similar to sodium chloride, rising concentrations of urea up to 7% resulted only in slightly delayed metabolic activity. However, individual substances in certain concentrations could significantly delay, reduce or even inhibit cellular respiration. The combination of sodium chloride 6% and dimethylsulphonylpropionate appeared to significantly decelerate the onset of metabolism more for some isolates and less for others. Moreover, increasing concentrations of sodium formate up to 6% reduced metabolic capacity and notably delayed bacterial metabolism. The presence of sodium lactate at concentrations of 3% to 12% also led to increasing inhibition of metabolic activity. Already at 4%, the metabolic activity of a few isolates was only recognizable during a 48 h incubation period. The isolates also showed reduced metabolism in the presence of sodium benzoate at a pH value of 5.2. The reduced metabolism was apparent at 20 mM, only a few isolates were able to maintain metabolism at 50 mM, and no activity was detected at 100 mM or 200 mM except for single isolates during a 48 h incubation period. In addition, rising concentrations of sodium nitrate had only a minor impact on the extent of metabolic activity, while those of sodium nitrite significantly reduced metabolism.
The extent of bacterial metabolism was also dependent on ambient pH conditions. Values between 5 and 10 did not pose significant difficulties, except for a slight decrease in metabolic activity at acidic values of 5 to 6. In contrast, no metabolic activity was detected at acidic pH values of 3.5 to 4.5. Provision of different nutrients such as amino acids and others at a pH of 4.5 did not affect this observation with the exception of urea. Here, some isolates were able to fully develop their activity, after a period of adaptation, which was mainly visible only over 48 h of incubation. Many studies investigated the effect of pH on the growth of S. aureus, for example to develop strategies to control S. aureus growth and enterotoxin production in food-manufacturing processes. Lanciotti et al. showed that among various bacterial species examined, S. aureus was characterized by the highest sensitivity to pH changes [63]. Considering an interactive effect of pH, water activity, temperature and ethanol concentrations on the pathogen's growth, predicted critical pH values ranged between 5 and 7 [63], which matches the pH values allowing bacterial metabolism observed in our study. A pH sensitivity within the range of 5 to 7 was also shown for S. aureus by Iyer et al. emphasizing the advantages of keeping the skin pH within an acidic range for maintenance of a balanced skin microbiome, including a lower level of S. aureus compared with other commensals [64].
In contrast to a previous study [23], no differences were recognized between isolates belonging to CC398 and CC9 regarding the survival under acidic and hypertonic conditions.  Figure 2 and mainly included sugars, sodium salts and amino acids combined with varying pH conditions. The two independent AUC values considered in each case were reproducible. The branching diagram shows that differentiation of the isolates depending on their CC and origin into the four groups CHN-MRSA-CC9, CHN-MRSA-CC398, GER-MRSA-CC9 and GER-MRSA-CC398 was possible on the basis of the most remarkable differences in metabolic substrate utilization. Moreover, the selected substrates and/or conditions could be assigned to eight differing groups. Regarding the 30 most informative substrates, the heat map suggested that the groups GER-MRSA-CC398, GER-MRSA-CC9 and CHN-MRSA-CC398 harbored more similar metabolic profiles compared to the CHN-MRSA-CC9 ( Figure 2). The Euclidian distance based on z-scores is for every Chinese CC9 isolate larger than that between the isolates of the other three groups. However, evaluation of respiration curves and mean AUC values revealed the following substrates as particularly interesting for the study's research question: pH 4.5 + urea, D-arabitol, pH 9.5 + agmatine, sodium benzoate pH 5.2 50 mM, sodium formate 3%, sodium formate 4%, sodium formate 6%, sedoheptulosan, pH 9.5 + L-glutamine, pH 9.5 + hydroxy-L-proline, pH 9.5 + L-leucine, pH 9.5 + L-isoleucine, and pH 9.5 + L-threonine. The four groups also displayed crucial differences in the metabolism of several nutrients beyond the sPLS-DA selection, which are described in the following sections. Overall, the selection of substrates and/or conditions using this method allowed grouping of the isolates, but a more detailed insight into growth kinetics considering additional parameters to the AUC was required to evaluate metabolic differences between the four groups.

Combined LT, MH, PT, and Slope sPLS-DA Variable Selection
The combined sPLS-DA variable selection also allowed a preselection of the most informative substrates and/or conditions tested. Similar to the AUC sPLS-DA selection, the 40 nutrients mainly included sugars, (amino) acids and sodium salts. In each case, the parameters LT, MH, PT and slope were reproducible and could be assessed in detail. Interestingly, the four isolate groups showed recurring patterns regarding the usage of many substances, which allowed classification of these metabolic differences as described in the following sections. However, for several substrates and/or conditions, the patterns were different across all parameters making classification impossible. Moreover, similar to the AUC sPLS-DA, relevant deviations between the four isolate groups were detected regarding various nutrients beyond the selection, which are also described below.

Relevant Metabolic
Varieties between the Four MRSA Groups 3.2.3.1. Metabolic Differences between the Dominant and Rare Lineage within a CC Different colonization and infection sites within a pathogen's host require unique metabolic pathways due to the varying availability of nutrients. For example, S. aureus would likely have access to lactate, urea and amino acids, but not carbohydrates at skin surfaces, and to serum glucose at deep tissue sites [65]. Moreover, several studies have shown that the metabolic state of S. aureus has an impact on the activity of major virulence factor regulators, proving the linkage of S. aureus' metabolism to its overall pathogenesis [65]. Therefore, it seemed likely that the dominant LA-MRSA CC9 from China and CC398 from Germany might have an advantage in host colonization due to certain metabolic properties in relation to the non-successful pig colonizing isolates.

Figure 2.
Heat map constructed via sPLS-DA using four components to differentiate porcine MRSA isolates from Germany and China based on CC and origin. Two independent area under the curve (AUC) values obtained after 24 h of incubation were considered for each of the isolates displayed on the right side. Based on the most remarkable differences selected nutrients are shown at the bottom. The color key mirrors the extent of metabolic substrate utilization (z-scaled) by the isolates, with red color displaying more successful and blue color less successful usage of a nutrient. The branching diagram provides classification of isolates and substrates into different groups.

Combined LT, MH, PT, and Slope sPLS-DA Variable Selection
The combined sPLS-DA variable selection also allowed a preselection of the most informative substrates and/or conditions tested. Similar to the AUC sPLS-DA selection, the 40 nutrients mainly included sugars, (amino) acids and sodium salts. In each case, the parameters LT, MH, PT and slope were reproducible and could be assessed in detail. Interestingly, the four isolate groups showed recurring patterns regarding the usage of many substances, which allowed classification of these metabolic differences as described in the following sections. However, for several substrates and/or conditions, the patterns were different across all parameters making classification impossible. Moreover, similar Lower AUC values indicated a less successful substrate metabolization by the rare lineages compared with the respective dominant clone within the same CC for the carbon sources phenylethylamine, D,L-octopamine and 3-hydroxy 2-butanone (Figure 3a). More precisely, the combined sPLS-DA revealed that the associated respiration curves of the dominant clones displayed greater MH values (Figure 3b). In addition, beyond the sPLS-DA selections, lower AUC values indicated a less successful substrate metabolization by the rare lineages compared with the respective dominant clone within the same CC regarding the carbon sources acetamide, L-phenylalanine and 2,3-butanediol (Figure 3a).
In contrast to initial expectations, only a few relevant metabolic differences were found between the dominant and rare lineages within a CC. Thus, the dominant lineages might have only single advantages in environments with limited nutrient availability. However, due to S. aureus' versatility in nutrient utilization and since several substrates can be used alternatively in the same essential metabolic pathways, we speculate that these variations did not play a role in the emergence of two different epidemic LA-MRSA clones in swine. Advantages such as an enhanced biofilm forming ability and an increased tolerance to desiccation, which we revealed in a previous study [23], more likely allow the dominant clones to outcompete the non-successful lineages in extreme environments.

Distinctive Metabolic Features of CHN-MRSA-CC9
The CHN-MRSA-CC9 showed varying metabolic capacities regarding several substrates compared with the other three groups. Major differences were detected for the carbon sources sedoheptulosan, xylitol and oxalomalic acid. Here, lower AUC values pointed towards a reduced usage in relation to the other groups. According to the combined sPLS-DA, higher PT and lower slope values were recognized for all three substrates. Moreover, an extended LT was observed in the metabolization of sedoheptulosan and xylitol and lower MH values were identified for sedoheptulosan as well as oxalomalic acid (Figure 4a). Sedoheptulosan was also included in the AUC sPLS-DA selection. Regarding the usage of oxalomalic acid, evaluation of the respiration curves provided additional information. Although Figure 4a indicated that the utilization of this nutrient by the CHN-MRSA-CC9 was the lowest, one isolate of the group could use it better than the others (see Section 3.2.3.6). The GER-MRSA-CC9 also seemed to have more difficulties using the substance compared to the MRSA-CC398. Therefore, the differences between the four groups considering oxalomalic acid could also have been discussed in Section 3.2.3.3. Interesting changes in the further progression of the respiration curves were only evaluable considering those resulting from an incubation period of 48 h (see Section 3.2.3.6).
Furthermore, relevant differences were detected beyond the sPLS-DA selections for the carbon sources m-tartaric acid and D-malic acid. Again, lower AUC values indicated a reduced usage in relation to the other groups (Figure 4b). In contrast, respiration curves and AUC values suggested a comparatively increased metabolization of the substrate 2-deoxy-d-ribose (Figure 4b). The AUC values also pointed towards a distinctly increased metabolization of the substrate 3-0-β-D-galactopyranosyl-D-arabinose by the CHN-MRSA-CC9. Evaluation of the respective respiration curves revealed that only these isolates were able to initiate cellular respiration within an incubation period of 24 h (see Section 3.2.3.6). However, the observation of single non-CHN-MRSA-CC9 isolates also being able to metabolize the substance was only detectable over a period of 48 h (see Section 3.2.3.6).
The reduced metabolic capacity and the delay of metabolism resulting from rising concentrations of sodium formate as mentioned in Section 3.2.1 was recognized for all four groups, but was most distinct for the CHN-MRSA-CC9. Here, the respiration curves displayed lower AUC values, but also an extended LT, lower MH as well as PT values and a greater slope as shown by the combined sPLS-DA (Figure 4b,c). The AUC sPLS-DA selection also suggested sodium formate at concentrations of 3%, 4% and 6% as relevant. As mentioned above, rising concentrations of urea also led to a delay of bacterial respiration, which was notably more distinct for the CHN-MRSA-CC9 compared to the other groups ( Figure 5a). Sodium benzoate 50 mM at a pH of 5.2 was also included in the AUC sPLS-DA selection. As discussed in Section 3.2.1, not all isolates could maintain metabolism under these conditions. While lower respiration curves and the delayed onset of metabolic activity displayed in Figure 5a indicated that isolates belonging to all groups struggled using this substance, no metabolic activity was detected for the CHN-MRSA-CC9. Thus, the CHN-MRSA-CC9 showed an increased sensitivity to rising concentrations of sodium benzoate in relation to the other groups (see Section 3.2.1). Microorganisms 2022, 10, x FOR PEER REVIEW 12 of 25

Distinctive Metabolic Features of CHN-MRSA-CC9
The CHN-MRSA-CC9 showed varying metabolic capacities regarding several substrates compared with the other three groups. Major differences were detected for the carbon sources sedoheptulosan, xylitol and oxalomalic acid. Here, lower AUC values pointed towards a reduced usage in relation to the other groups. According to the combined sPLS-DA, higher PT and lower slope values were recognized for all three substrates. Moreover, an extended LT was observed in the metabolization of sedoheptulosan and xylitol and lower MH values were identified for sedoheptulosan as well as oxalomalic acid ( Figure  4a). Sedoheptulosan was also included in the AUC sPLS-DA selection. Regarding the usage of oxalomalic acid, evaluation of the respiration curves provided additional information. Although Figure 4a indicated that the utilization of this nutrient by the CHN-MRSA-CC9 was the lowest, one isolate of the group could use it better than the others (see In addition, the respiration curves of the CHN-MRSA-CC9 suggested the requirement for a longer adaptation time to the surrounding conditions before onset of nutrient utilization. Not all findings can be shown here, but the observation was particularly evident considering the usage of 2,3-butanedione (Figure 5a) as well as of amino acids and derivatives thereof at basic pH values. The examples given in Figure 5b are representative for all amino acids and derivatives at pH 9.5 and several of them were also emphasized by the AUC sPLS-DA (see Section 3.2.2.1). The availability of amino acids, which can be catabolized to produce the essential biosynthetic intermediates, is of importance for S. aureus, as the bacterium is frequently described as an amino acid auxotroph [61]. However, S. aureus often has the ability to revert to a prototrophic state, which is important for the adaptation to host/environmental niches where nutrients are limited [61]. Moreover, Halsey et al. have shown that certain amino acids serve as central carbon sources for S. aureus critical for growth in medium lacking glucose [66]. This supports a model describing the ability to metabolize secondary carbon sources, such as amino acids, as essential for survival in niches where preferred nutrients, such as glucose, are limited, as for example within a staphylococcal abscess [66]. Thus, the CHN-MRSA-CC9 might be inferior in niche adaptation due to their delayed amino acid catabolism.
the CHN-MRSA-CC9 to the other three lineages during evolution of epidemic clones in swine might explain the more distinct metabolic profile of the MRSA-CC9 from China. The carriage of a larger number of antimicrobial resistance genes has possibly led to a higher fitness cost in the CHN-MRSA-CC9 than in other isolates [67], explaining the reduced metabolic capabilities regarding several substances and the throughout extended adaptation times before onset of cellular respiration. Increased metabolization of 2-deoxyd-ribose and 3-0-β-D-galactopyranosyl-D-arabinose by the CHN-MRSA-CC9 in relation to the other groups may, however, represent a slight advantage for these isolates, but might as well result in notable disadvantages in case of harmful metabolite formation during metabolic processes.   In the previous study, we also suggested that the current epidemic MRSA-CC9 clone in China may be gradually replaced by the MRSA-CC398 lineage due to several advantageous features that allow MRSA-CC398 to persist in the pig host or hostile environment [23]. These included easier integration of foreign antimicrobial resistance genes, improved fitness and adaptation to varying environmental conditions, enhanced virulence in infection models, increased pathogenicity due to the hysA νSaβ gene, better adaptation to acidic and hyperosmotic environments and higher competitiveness compared to MRSA-CC9 [23]. The small disadvantages of the CHN-MRSA-CC9 in metabolic properties described As expected, the cgMLST revealed a closer relationship between the isolates within the two CCs regardless of their geographical origin. Nevertheless, the CHN-MRSA-CC9 showed major metabolic differences to the other three groups. The greater similarities of the CHN-MRSA-CC398, GER-MRSA-CC398 and GER-MRSA-CC9 are also displayed in Figure 2. In a previous study, we observed that the MRSA-CC9 from China and Germany possibly belong to two independent evolutionary lineages, whilst the CHN-MRSA-CC398 and the GER-MRSA-CC398 originated from the same ancestor [23]. Moreover, we speculated that the MRSA-CC398 might have been introduced into China through trading activities [23]. Another recent study also revealed a close evolutionary relationship be-tween Chinese and European or Australian LA-MRSA ST398 [16]. The greater spatial distance of the CHN-MRSA-CC9 to the other three lineages during evolution of epidemic clones in swine might explain the more distinct metabolic profile of the MRSA-CC9 from China. The carriage of a larger number of antimicrobial resistance genes has possibly led to a higher fitness cost in the CHN-MRSA-CC9 than in other isolates [67], explaining the reduced metabolic capabilities regarding several substances and the throughout extended adaptation times before onset of cellular respiration. Increased metabolization of 2-deoxy-d-ribose and 3-0-β-D-galactopyranosyl-D-arabinose by the CHN-MRSA-CC9 in relation to the other groups may, however, represent a slight advantage for these isolates, but might as well result in notable disadvantages in case of harmful metabolite formation during metabolic processes.
In the previous study, we also suggested that the current epidemic MRSA-CC9 clone in China may be gradually replaced by the MRSA-CC398 lineage due to several advantageous features that allow MRSA-CC398 to persist in the pig host or hostile environment [23]. These included easier integration of foreign antimicrobial resistance genes, improved fitness and adaptation to varying environmental conditions, enhanced virulence in infection models, increased pathogenicity due to the hysA νSaβ gene, better adaptation to acidic and hyperosmotic environments and higher competitiveness compared to MRSA-CC9 [23]. The small disadvantages of the CHN-MRSA-CC9 in metabolic properties described in this study may support the replacement of the epidemic Chinese clone by MRSA-CC398 after further dissemination of isolates through livestock trade or human occupational exposure. The delayed onset of cellular respiration of the CHN-MRSA-CC9, combined with the decreased usage of many nutrients as well as possible disadvantages resulting from increased metabolization rates of some substrates might be sufficient for MRSA-CC398 to become established once extensively colonizing the pig host. Despite the metabolic robustness typical for S. aureus [61], the CHN-MRSA-CC9 may be less adaptable to environmental challenges than the MRSA-CC398 clone. The increasing reports of swine-associated MRSA-CC398 isolates from China support the hypothesis of the ongoing MRSA-CC9 replacement process [16,17]. It is important to note that in the future MRSA-CC398 may pose a greater public health threat than MRSA-CC9 due to their enhanced pathogenicity [23]. For example, results from Cui et al. recently suggested that ST398 was a frequent source of MRSA and MSSA infections in the Qinghai province [16]. Li et al. even implied a potential MRSA-ST398 transmission among different countries [17]. They reported an MRSA ST398 isolate obtained from an infected patient in Europe that differed by only 31 SNPs from the airborne dust-associated Chinese isolate in their study [17]. This highlights the need for comprehensive worldwide LA-MRSA monitoring.

Metabolic Diversity Based on the Assignment to a CC
Respiration curves resulting from the metabolization of D-serine and D-arabitol indicated a CC-dependent usage of these nutrients in accordance with the combined sPLS-DA variable selection. Considering D-serine, the MRSA-CC9 isolates displayed lower AUC, MH and slope values, but an extended PT in relation to the MRSA-CC398 isolates ( Figure 6). In contrast, higher AUC, MH and PT values indicated an increased metabolic activity of the MRSA-CC9 compared to the MRSA-CC398 isolates regarding D-arabitol ( Figure 6). The substrate was also included in the AUC sPLS-DA selection.
ments [23]. Moreover, a novel hysA νSaβ gene causing enhanced pathogenicity, which was present in CC398 isolates but absent in MRSA-CC9, might explain why MRSA-CC398 has been reported to be more likely to cause infections [6,23,68]. For this reason, we initially expected to detect multiple differences in the metabolic properties of MRSA-CC9 and MRSA-CC398. However, only single metabolic differences due to CC affiliation were recognized in this study overall.  In a previous study [23] we identified several advantages of MRSA-CC398 isolates towards MRSA-CC9 as they revealed more diverse genome structures, higher tolerance to acids and high osmotic pressure and greater competitive fitness in co-culture experiments [23]. Moreover, a novel hysA νSaβ gene causing enhanced pathogenicity, which was present in CC398 isolates but absent in MRSA-CC9, might explain why MRSA-CC398 has been reported to be more likely to cause infections [6,23,68]. For this reason, we initially expected to detect multiple differences in the metabolic properties of MRSA-CC9 and MRSA-CC398. However, only single metabolic differences due to CC affiliation were recognized in this study overall.

Metabolic Diversity Attributed to the Country of Origin
Considering the carbon sources α-hydroxy glutaric acid-γ-lactone, glycolic acid, βmethyl-D-glucuronic acid, stachyose, γ-amino butyric acid, 4-hydroxy benzoic acid and oxalic acid in the combined sPLS-DA selection, differences in metabolic properties seemed to be associated with the isolates' origin. Throughout, the Chinese MRSA showed lower AUC and MH values as well as an extended LT in comparison with the German isolates, indicating a decreased nutrient usage (Figure 7a). Moreover, as shown by lower AUC values beyond the sPLS-DA selections, the CHN-MRSA-CC9 and CHN-MRSA-CC398 appeared to be less successful in metabolizing α-keto-butyric acid and especially 2-hydroxy benzoic acid compared to the German MRSA (Figure 7b). These findings may suggest an adaptation of the MRSA isolates to the unique character of the Chinese and German pig farm environments.

Other Metabolic Differences between the Groups
Some variations in the metabolic activity between the four groups CHN-MRSA-CC9, CHN-MRSA-CC398, GER-MRSA-CC9 and GER-MRSA-CC398 were individual observations and, thus, not discussed in detail. An example is given in Figure S1. According to the combined sPLS-DA, higher AUC and MH values and a shortened LT indicated a more successful usage of glycine by the GER-MRSA-CC398 compared to the other three groups.
indicating a decreased nutrient usage (Figure 7a). Moreover, as shown by lower AUC values beyond the sPLS-DA selections, the CHN-MRSA-CC9 and CHN-MRSA-CC398 appeared to be less successful in metabolizing α-keto-butyric acid and especially 2-hydroxy benzoic acid compared to the German MRSA (Figure 7b). These findings may suggest an adaptation of the MRSA isolates to the unique character of the Chinese and German pig farm environments. In addition, variable results were detected for certain nutrients within each of the four groups and, therefore, not categorized further. Here, the metabolization of nutrients seemed to be more isolate-dependent. Selected examples displaying cellular respiration in the presence of dulcitol, tricarballylic acid, glycogen, a-methyl-D-glucoside, sodium lactate 3% or sodium nitrite 40 mM are shown in Figure S2. Sodium lactate 3% is representative for sodium lactate 2% to 12% and sodium nitrite 40 mM is representative for sodium nitrite 10 mM to 100 mM.

Metabolic Variations Detectable Only after a 48 h Incubation Period
Some disparities in the isolates' metabolic activities were only detectable after an incubation period of 48 h. For the substance 3-0-β-D-galactopyranosyl-D-arabinose, this has already been mentioned in Section 3.2.3.2 ( Figure 8). In addition, the metabolization of urea at a pH of 4.5 would have been interpreted differently, if only the 24 h interval would have been considered. The 24 h respiration curves suggested that only two GER-MRSA-CC9 were able to initiate metabolic activity. In contrast, the curves obtained after 48 h of incubation revealed that more isolates from several groups were able to utilize urea at a pH of 4.5 after an adaptation period. However, the CHN-MRSA-CC9 were unable to metabolize this substrate under these conditions (Figure 8).

Potential Limitations of the Study
The aim of this study was to conduct a comparative evaluation of the metabolic properties of four MRSA groups with only five isolates per group and thus, 20 isolates in total. This is a relatively small sample size and might limit the transferability of the results for the represented groups. Considering the MRSA-CC398 from China, the isolates investigated were the only ones available to us during the study period. Due to limited resources, we decided to assemble comparable groups. Preliminary experiments confirmed, in general, a uniform metabolic behavior of the isolates within each group. For this reason, it was possible to draw the respective conclusions from the results obtained analyzing this limited isolate collection.
Only four Biolog PM microplate layouts out of ten possible metabolic phenotyping arrays available for bacteria and fungi provided by Biolog Inc. (Hayward, CA, USA) were used for the experiments. The microplate layouts PM1 (carbon utilization), PM2A (carbon utilization), PM9 (osmotic and ionic effects) and PM10 (pH effects) were chosen, because here we expected to observe relevant differences between the four MRSA groups. Cellular stress responses and a high osmotolerance allow S. aureus to persist in different environmental settings. In addition, despite its versatility in nutrient usage, S. aureus is known for typically metabolizing carbohydrates when they are available and the carbohydrate ca- The 48 h values also enabled the identification of changes in curve progression and thus of differences between the groups that otherwise would not have been recognized (Figure 8). According to the 24 h respiration curves, all four groups seemed to have difficulties in using the nutrient β-D-allose, but only the 48 h curves provided information on the further course. It was demonstrated for all groups that substrate utilization was possible after a period of adaptation. However, differing durations of adjustment indicated that the ability was not evenly distributed within the four groups. In general, the CHN-MRSA-CC9 required the longest adaptation time (Figure 8). The reduced usage of oxalomalic acid by the CHN-MRSA-CC9 has already been mentioned in Section 3.2.3.2. The CHN-MRSA-CC9 displayed a slightly delayed onset of cellular respiration and all isolates except one showed only very low levels of metabolic activity in relation to the other isolates. The 24 h respiration curves did not suggest changes in course progression at a later point in time. However, the 48 h respiration curves of four CHN-MRSA-CC9 showed an increase towards the end of the incubation period almost up to the level of the other MRSA-CC9 isolates ( Figure 8).
Finally, the examples referring to urea at pH 4.5, β-D-allose and oxalomalic acid also supported the conclusion that, compared to the other three groups, the CHN-MRSA-CC9 required a longer period of adaptation before the onset of cellular respiration and/or are not able to metabolize several substances at all or only to a lesser extent.

Potential Limitations of the Study
The aim of this study was to conduct a comparative evaluation of the metabolic properties of four MRSA groups with only five isolates per group and thus, 20 isolates in total. This is a relatively small sample size and might limit the transferability of the results for the represented groups. Considering the MRSA-CC398 from China, the isolates investigated were the only ones available to us during the study period. Due to limited resources, we decided to assemble comparable groups. Preliminary experiments confirmed, in general, a uniform metabolic behavior of the isolates within each group. For this reason, it was possible to draw the respective conclusions from the results obtained analyzing this limited isolate collection.
Only four Biolog PM microplate layouts out of ten possible metabolic phenotyping arrays available for bacteria and fungi provided by Biolog Inc. (Hayward, CA, USA) were used for the experiments. The microplate layouts PM1 (carbon utilization), PM2A (carbon utilization), PM9 (osmotic and ionic effects) and PM10 (pH effects) were chosen, because here we expected to observe relevant differences between the four MRSA groups. Cellular stress responses and a high osmotolerance allow S. aureus to persist in different environmental settings. In addition, despite its versatility in nutrient usage, S. aureus is known for typically metabolizing carbohydrates when they are available and the carbohydrate catabolism has been studied most extensively [61]. However, the application of additional Biolog PM layouts, e.g., phosphorus and sulfur sources, might be of interest for possible future studies. It should be noted that the PM5 (biosynthetic pathways) layout is not appropriate for testing staphylococci because of their polyauxotrophy [69].
Another potential limitation might be that no standardized statistical pipeline was available that would allow us to compare Biolog PM assay data not only between several bacterial isolates but also between the four isolate groups. For this reason, the AUC sPLS-DA and the combined sPLS-DA were carried out in an attempt to select the most relevant metabolic variations out of the extensive dataset. However, manual comparison of all respiration curves revealed substrates and/or conditions beyond the preselections as relevant due to notable metabolic variations between the four groups. Therefore, an automated, less work-intensive analysis of the data was not possible, although AUC sPLS-DA and combined sPLS-DA proved to be helpful.
Finally, it is worth noting that bacterial metabolism is a large and complex field. Biochemical processes and the regulatory interactions between them have been partially elucidated, but are still unknown to a large extent [70]. In order to generate insight into metabolic networks, integration of experimental results with published data, as well as accumulated observations and anecdotal tales exchanged among scientists working in this field is required [70]. However, a decline in research focusing on microbial metabolism would appear to make this more difficult [70]. Therefore, a limitation of this study may be that we were not able to discuss all substrates in detail that were mentioned in the context of relevant metabolic differences between the MRSA groups. Despite extensive literature research, which metabolic pathways of S. aureus involving these substrates play a role in fitness and niche competition in a given clinical or environmental setting is not known. It was also not possible to assign the respective substrates to specific groups, other than carbon sources, that would have allowed general conclusions to be drawn. Thus, further research integrating in silico, in vitro and in vivo approaches is required to better understand microbial metabolism in its details [70].

Conclusions
Applying Biolog PM technology, we characterized the metabolic profiles of rare as well as widespread porcine LA-MRSA from Germany and China. Overall, differences between the four LA-MRSA groups were rather small. In contrast to initial expectations, only a few metabolic differences were found between the dominant and rare lineages within the same CC. Therefore, unique metabolic profiles most likely did not play a key role in the formation of the dominant clones CHN-MRSA-CC9 and GER-MRSA-CC398, respectively. Surprisingly, the most frequent and substantial metabolic variations were detected between the CHN-MRSA-CC9 and the other three LA-MRSA groups. In addition to reduced or even absent usage of some substrates, and the increased metabolization of 2-deoxy-d-ribose and 3-0-β-D-galactopyranosyl-D-arabinose, especially the delayed onset of cellular respiration was often noticed. The AUC sPLS-DA heat map also suggested metabolic differences of the CHN-MRSA-CC9 compared to the other groups. These comparatively unfavorable metabolic properties might promote the suspected already ongoing gradual replacement of the current epidemic MRSA-CC9 clone in China by the superior MRSA-CC398 clone. Thus, the risk for public health posed by LA-MRSA originating from swine might rise alarmingly in the future due to the enhanced pathogenicity of the MRSA-CC398 lineage.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/microorganisms10112116/s1, Table S1: Distance matrix of the core genome allelic profiles of the 20 representative porcine MRSA from Germany and China including 1749 of 1861 possible target genes by removing 112 columns with missing values from the comparison table created with the SeqSphere+ software, Table S2: Biolog PM microplate wells that showed abiotic "false-positive" reactions under the study's test conditions, Figure S1: Metabolization of glycine by the four isolate groups, and Figure S2: Selected Biolog PM assay results that were variable within the four isolate groups and, therefore, not categorized further.