The carnitine degradation pathway of Acinetobacter baumannii and its role in virulence

Summary

The opportunistic human pathogen Acinetobacter baumannii can grow with carnitine but its metabolism, regulation and role in virulence remained elusive. Recently, we identified a carnitine transporter encoded by a gene closely associated with potential carnitine degradation genes. Among those is a gene coding for a putative d-malate dehydrogenase (Mdh). Deletion of the mdh gene led to a loss of growth with carnitine but not l-malate; growth with d-malate was strongly reduced. Therefore, it is hypothesized that d-malate is formed during carnitine oxidation and further oxidized to CO₂ and pyruvate and, that not, as previously suggested, l-malate is the product and funnelled directly into the TCA cycle. Mutant analyses revealed that the hydrolase in this cluster funnels acetylcarnitine into the degradation pathway by deacetylation. A transcriptional regulator CarR bound in a concentration-dependent manner to the intergenic region between the mdh gene, the first gene of the carnitine catabolic operon and the carR gene in the presence and absence of carnitine. Both carnitine and d-malate induced CarR-dependent expression of the carnitine operon. Infection studies with Galleria mellonella larvae demonstrated a strong increase in virulence by addition of carnitine indicating that carnitine degradation plays a pivotal role in virulence of A. baumannii.

Introduction

Acinetobacter baumannii is an opportunistic human pathogen which has become an emerging threat in healthcare institutions due to increasing antibiotic resistances (Villegas and Hartstein, 2003; Antunes et al., 2014; Harding et al., 2018). The success of A. baumannii in the hospital environment is based on its outstanding desiccation resistance and persistence in the human host (Dijkshoorn et al., 1987; Wendt et al., 1997; Peleg et al., 2008; Towner, 2009; Zeidler and Müller, 2019). The latter is fostered by its ability to metabolize host-derived carbon and nitrogen sources, allowing survival and thriving in different host niches such as lungs (pneumonia), blood, urinary tract or wounds (Dijkshoorn et al., 2007; Fiester and Actis, 2013). Host-derived carbon sources which are metabolized by A. baumannii and therefore are good candidates to play a role in metabolic adaptation are sugars, alcohols, lipids, amino acids, organic acids, aromatic compounds and quaternary amines, such as carnitine and choline (Camarena et al., 2010; Stahl et al., 2015; Breisch et al., 2018; Hubloher et al., 2020; König et al., 2021). The latter are very abundant in human hosts such as choline is the head group moiety of phosphatidylcholine (PC) and sphingomyelin which form 50%–90% of the phospholipids in the outer leaflet of eukaryotic plasma membranes (Zachowski, 1993). Especially the membranes of lung epithelial cells are choline rich and consist of up to 70% PC (Keller and Ladda, 1979). Carnitine is also very abundant in human host tissues and is found in blood, liver, kidney, brain and heart. Carnitine plays a primary role in the carnitine shuttle in mitochondria where it transports long-chain fatty acids for subsequent fatty acid β-oxidation (Bernal et al., 2007; McCann et al., 2021).

The quaternary amine l-carnitine (hereafter referred to as ‘carnitine’) is used by A. baumannii as carbon and energy source (Zhu et al., 2014; Breisch et al., 2018). Recently, we identified a betaine/choline/carnitine transporter, Aci01347, which mediates the energy-dependent uptake of carnitine but also choline by A. baumannii ATCC 19606 (Breisch et al., 2018). Further studies revealed that Aci01347 is not activated by osmolarity which is consistent with our suggestion that carnitine is used as carbon and energy source (Breisch et al., 2018). The aci01347 gene is closely associated with a potential carnitine catabolic gene cluster. This gene cluster was previously detected by genome mining for carnitine-degrading enzymes in the genome of Acinetobacter ssp. within a human microbiome project (HMP) (Zhu et al., 2014). Two genes of this gene cluster, cntA and cntB, were found to encode a two-component Rieske-type oxygenase/reductase complex (CntAB) which catalyses the degradation of carnitine to trimethylamine (TMA) and malic semialdehyde (Zhu et al., 2014; Massmig et al., 2020). The reaction mechanism of the CntAB complex was elucidated recently (Massmig et al., 2020). The generated malic semialdehyde was suggested to be reduced by a malic semialdehyde dehydrogenase to malate which is suggested to be funnelled into the tricarboxylic acid cycle via a malate dehydrogenase (Zhu et al., 2014). However, it has to be noted that so far there is no experimental evidence for this pathway except for the CntAB mediated oxidation of carnitine. The malate dehydrogenase gene is preceded by an oppositely orientated potential LysR-type transcriptional regulator (LTTR) gene, which is a good candidate to mediate transcriptional regulation of the carnitine catabolic gene cluster (Zhu et al., 2014; Breisch et al., 2018).

In this study we identified an acetylcarnitine hydrolase (Hyd) in A. baumannii ATCC 19606 mediating the conversion of acetylcarnitine to carnitine. Carnitine is further oxidized to d-malate. In contrast to the expectations, d-malate is further decarboxylated to pyruvate and CO₂, catalysed by a malate dehydrogenase (Mdh) encoded by the first gene of the carnitine degradation cluster. Furthermore, we identified an LTTR acting as activator of the carnitine degradation pathway in the presence of the inducer substrates carnitine, d-malate and acetylcarnitine. We show that the genes of the carnitine catabolic pathway form an operon and that the transcriptional regulator CarR binds to the intergenic DNA region between mdh and carR. Moreover, we provide evidence by Galleria mellonella larvae infection studies that the carnitine degradation pathway plays a role in virulence.

Experimental procedures

Bacterial strains and culture conditions

Escherichia coli BL21 STAR was grown in LB medium (Bertani, 1951) at 37°C in the presence of 100 μg ml⁻¹ ampicillin. Acinetobacter baumannii strains were grown at 37°C in LB medium (Bertani, 1951) or in mineral medium that consists of 50 mM phosphate buffer, pH 6.8, and different salts (Zeidler et al., 2017) and 20 mM of acetate, carnitine, acetylcarnitine or d-malate were used as carbon source. 50 μg ml⁻¹ kanamycin was added when appropriate. The growth experiments were repeated three times and the ±S.E.M. is shown. Growth curves were fitted manually.

Markerless mutagenesis of A. baumannii ATCC 19606

To generate Δhyd, Δmdh and ΔcarR mutants, 1500 bp upstream and 1500 bp downstream of the genes were amplified from A. baumannii ATCC 19606 genomic DNA (primer pairs: Suppl. Table 1) and cloned in pBIISK_sacB/kanR (Stahl et al., 2015) using NotI and PstI. Plasmid was transformed in electrocompetent A. baumannii wild type cells. Electrocompetent A. baumannii cells were prepared as described before (Stahl et al., 2015). Electroporation was performed at 2.5 kV, 200 Ω and 25 μF. Transformants were selected on LB agar using 50 μg ml⁻¹ kanamycin and verified by PCR (primer pairs: Suppl. Table 1). Segregation was induced by counter selection using 10% sucrose for 18 h at 37°C followed by plating on LB agar containing 10% sucrose. Single colonies exhibiting kanamycin sensitivity were verified by PCR (primer pairs: Suppl. Table 1).

RNA isolation, cDNA synthesis and bridging PCR

For RNA isolation cells were grown overnight in mineral medium with 20 mM carnitine as carbon source. 5 ml of cell cultures were harvested in the stationary phase by centrifugation (10 min, 4°C, 4700 rpm) and re-suspended in 2 ml Tri-Reagent® (Sigma-Aldrich). After 15 min of incubation at room temperature, 0.2 ml chloroform was added, mixed, and followed by incubation for 15 min at room temperature and centrifugation (15 min, 4°C, 12 000g). The aqueous phase was transferred to a new reaction tube. After addition of 500 μl isopropanol to the aqueous phase and 10 min incubation at room temperature, nucleic acids were precipitated by centrifugation (10 min, 4°C, 12 000g) and re-suspended in 50 μl H₂O. Subsequently, DNA contaminations were removed by digestion with TURBO™ DNase (Invitrogen) and proteins were removed by repeating chloroform treatment described above. RNA was used for reverse transcription using M-MLV Reverse Transcriptase (Promega). For bridging PCR analyses 100 ng of cDNA, DNase digested RNA or genomic DNA was used as templates and Phusion DNA-Polymerase (NEB) for amplification. The resulting PCR products were analysed by agarose gel electrophoresis.

RNA extraction and RNA sequencing

To perform RNA extraction for transcriptomic analysis cells were grown to an OD₆₀₀ of 0.5 in mineral medium with 20 mM carnitine, acetylcarnitine, d-malate or succinate as carbon source and 2 ml of the cultures were harvested by centrifugation (10 min, 4°C, 4700 rpm). Harvested cells were re-suspended in 800 μl RLT buffer (RNeasy Mini Kit, Qiagen) with β-mercaptoethanol (10 μl ml⁻¹) and cell lysis was performed using a laboratory ball mill. Subsequently, 400 μl RLT buffer (RNeasy Mini Kit Qiagen) with β-mercaptoethanol (10 μl ml⁻¹) and 1200 μl 96% (vol./vol.) ethanol were added. For RNA isolation, the RNeasy Mini Kit (Qiagen) was used as recommended by the manufacturer, but instead of RW1 buffer RWT buffer (Qiagen) was used in order to isolate RNAs smaller than 200 nucleotides also. To determine the RNA integrity number the isolated RNA was run on an Agilent Bioanalyzer 2100 using an Agilent RNA 6000 Nano Kit as recommended by the manufacturer (Agilent Technologies, Waldbronn, Germany). Remaining genomic DNA was removed by digesting with TURBO DNase (Invitrogen, Thermo Fischer Scientific, Paisley, UK). The Pan-Prokaryotes riboPOOL kit v4 (siTOOLS BIOTECH, Planegg/Martinsried, Germany) was used to reduce the amount of rRNA-derived sequences (samples 1–3) and the Illumina Ribo-Zero plus rRNA Depletion Kit (Illumina, San Diego, CA, USA) was used to reduce the amount of rRNA-derived sequences of samples 4–12. For sequencing, the strand-specific cDNA libraries were constructed with a NEBNext Ultra II Directional RNA library preparation kit for Illumina and the NEBNext Multiplex Oligos for Illumina (96) (New England BioLabs, Frankfurt am Main, Germany). To assess quality and size of the libraries samples were run on an Agilent Bioanalyzer 2100 using an Agilent High Sensitivity DNA Kit as recommended by the manufacturer (Agilent Technologies). Concentration of the libraries was determined using the Qubit® dsDNA HS Assay Kit as recommended by the manufacturer (Life Technologies GmbH, Darmstadt, Germany). Sequencing of samples 1–3 was performed on the HiSeq4000 instrument (Illumina) using HiSeq® 3000/4000 SBS Kit (50 cycles) and the HiSeq 3000/4000 SR Cluster Kit (Illumina) in single-end mode with 50 bp read length. For sequencing of samples 4–12 the NovaSeq 6000 instrument (Illumina) with the NovaSeq 6000 SP Reagent Kit v1.5 (100 cycles) and the NovaSeq XP 2-Lane Kit v1.5 was used in the paired-end mode and 2× 50 cycles. For quality filtering and removing of remaining adaptor sequences, Trimmomatic-0.39 (Bolger et al., 2014) and a cutoff phred-33 score of 15 were used. The mapping against the reference genomes of A. baumannii ATCC 19606 (Ref: NZ_CP058289.1) was performed with Salmon (v 1.5.2) (Patro et al., 2017). As mapping backbone a file that contains all annotated transcripts excluding rRNA genes and the whole genome of the references as decoy was prepared with a k-mer size of 11. Decoy-aware mapping was done in selective-alignment mode with ‘–mimicBT2’, ‘–disableChainingHeuristic’ and ‘–recoverOrphans’ flags as well as sequence and position bias correction and 10 000 bootstraps. For –fldMean and –fldSD, values of 325 and 25 were used respectively. The quant.sf files produced by Salmon were subsequently loaded into R (v 4.0.5) (R Core Team, 2020) using the tximport package (v 1.18.0) (Soneson et al., 2015). DeSeq2 (v 1.30.0) (Love et al., 2014) was used for normalization of the reads and fold change shrinkages were also calculated with DeSeq2 and the apeglm package (v 1.12.0) (Zhu et al., 2019). Genes with a log2-fold change of + 5/− 5 and a p-adjust value <0.05 were considered differentially expressed. Raw reads have been deposited in the Sequence Read Archive as BioProject PRJNA821016.

Cloning, expression and purification of CarR

To express carR in E. coli BL21 STAR, the gene was cloned into the multiple cloning site of pBAD/HisA using EcoRI and PstI (primer pairs in Suppl. Table 1).

E. coli BL21 STAR harbouring plasmid pBAD/HisA_carR was used to inoculate 1 L of LB medium containing 100 μg ml⁻¹ ampicillin. After reaching an OD₆₀₀ of 0.7–0.8 gene expression was induced for 3 h by addition of IPTG to a final concentration of 1 mM, cells were harvested by centrifugation (8000g for 7 min at 4°C), washed in 50 ml lysis buffer (50 mM Tris, 300 mM NaCl, 10 mM imidazole, pH 8,0) and stored at −20°C.

For purification of CarR, frozen cells were thawed on ice, re-suspended in 15 ml lysis buffer containing DNase and 0.5 mM PMSF and disrupted via French Press (three times, 1000 psi). Cell debris was removed (14 000g, 30 min, 4°C) and cell-free lysate was incubated with 1 ml of Ni-NTA material for 30 min and shaken at 4°C for binding of His-CarR to the Ni-NTA matrix. Column was washed with 10 ml washing buffer 1 (50 mM Tris, 300 mM NaCl, 50 mM imidazole, pH 8.0), 10 ml washing buffer 2 (50 mM Tris, 300 mM NaCl, 70 mM imidazole, pH 8.0) and elution of CarR was performed using 10 ml elution buffer in 1 ml elution steps (50 mM Tris, 300 mM NaCl, 150 mM imidazole, pH 8.0). The protein concentration was determined by Bradford (1976). Protein composition of the different fractions was analysed by electrophoresis through a 12.5% sodium dodecyl sulfate-polyacrylamide gel according to Laemmli (1970) stained with Coomassie (0.5 g L⁻¹ Serva Blue R-250, 100 ml L⁻¹ methanol, 100 ml L⁻¹ glacial acetic acid).

Electrophoretic mobility shift assay

For the EMSA studies, the regulator CarR was expressed and produced in E. coli BL21 STAR cells and purified. The DNA fragments used in the EMSA studies were amplified by PCR using the primer pairs given in the Suppl. Table 1. 233 fmol of the 693 bp fragment spanning the intergenic region between the carR and mdh gene and a non-related 1613 bp DNA fragment were incubated with 0–81 pmol of CarR in a total volume of 20 μl with 1× EMSA reaction buffer [10 mM Tris, 1 mM EDTA, 0.1 mM KCl, 5% (vol./vol.) glycerin, pH 8.0] for 30 min at room temperature. 2 μl of EMSA loading dye [10 mM Tris, 1 mM EDTA, 100 mM KCl, 50% (vol./vol.) glycerin, bromphenolic blue, xylene cyanol, pH 8,0] was added to the reaction and 20 μl was analysed by native polyacrylamide electrophoresis.

Galleria mellonella infection studies

Caterpillars (provided by a local provider) were preselected by melanization, size (400 ± 50 mg) and movement in response to touch. A. baumannii wild type and mutants were grown in LB until an OD₆₀₀ of 1.0–1.5, harvested (10 min, 4700 rpm), washed with phosphate-buffered saline (2.7 mM KCl, 1.5 mM KH₂PO₄, 137.9 mM NaCl, 8.1 mM Na₂HPO₄) and adjusted to a final OD₆₀₀ of 1.5. Twenty caterpillars were used per experiment and per strain. 10 μl (~6 × 10⁶ CFU) of the cell suspension was injected into the last prolegs. As control 10 untreated caterpillars as well as 20 caterpillars were treated with 10 μl of PBS, 10 μl of PBS with 5 mM carnitine, acetylcarnitine, d-malate or acetate. Caterpillars were incubated at 37°C for 5 days and survival was documented every 24 h. All experiments were repeated at least three times and the ±S.E.M. was shown. Significance of survival differences was assessed by t-test.

Results

Characterization of the carnitine degradation gene cluster

Recently, we identified a secondary transporter Aci01347 mediating carnitine uptake of A. baumannii. The transporter gene is located in a gene cluster comprising Rieske-type oxygenase/reductase (cntAB) genes and other genes suggested to play a role in carnitine metabolism (Fig. 1A) (Zhu et al., 2014; Breisch et al., 2018). Upstream of the transporter gene aci01347 is an open reading frame coding for a 370 amino acid protein annotated as malate dehydrogenase (Mdh). This protein shares high amino acid sequence similarities and identities with the tartrate dehydrogenase of Pseudomonas putida (90% similarity/70% identity) and the d-malic enzyme of Rhodobacter capsulatus (90% similarity/72% identity), which are responsible for the oxidative decarboxylation of tartrate to oxaloglycolate and d-malate to pyruvate respectively (Tipton and Peisach, 1990; Martínez-Luque et al., 2001). Upstream of this gene is in opposite orientation a gene coding for a potential LTTR comprising 326 amino acids. This LTTR is 76.8% similar and 46.4% identical to the LTTR of E. coli O157:H7, which is a transcriptional activator of a d-malate dehydrogenase (Lukas et al., 2010). The open reading frame downstream of aci01347 encodes a 324 amino acid predicted hydrolase (Hyd), which displays highest similarities (62.2%) and identities (41.7%) to a lipolytic protein with unknown function from Pseudomonas aeruginosa PAO1. The gene of the putative hydrolase is flanked downstream by the cntA gene, which encodes the carnitine oxygenase. The open reading frame downstream of cntA encodes a 483 amino acid protein which was predicted to encode a malic semialdehyde dehydrogenase (MsaDH). The deduced protein shows highest amino acid sequence similarities and identities of 87.7% and 61.7% respectively, to a succinate semialdehyde dehydrogenase of E. coli O157:H7. The last gene of the carnitine degradation cluster, cntB, has already been identified as carnitine reductase subunit of the CntAB complex (Zhu et al., 2014; Massmig et al., 2020). The finding that CntAB are essential for the conversion of carnitine to TMA together with the similarities of the gene products of closely associated genes (Fig. 1A) suggests that the malic semialdehyde generated by CntAB is further oxidized by the potential MsaDH to malate followed by conversion to pyruvate mediated by a potential Mdh (Fig. 1B). However, it has to be noted that so far only the role of CntAB in carnitine conversion to malic semialdehyde and TMA has been experimentally proven, whereas there is no experimental evidence with respect to the function of other gene products of the carnitine degradation gene cluster. The presence of a potential hydrolase gene (hyd) in the carnitine degradation gene cluster led to the hypothesis, that also acetylcarnitine is used as sole carbon and energy source by A. baumannii.

Fig. 1

Organization of the putative carnitine degradation gene cluster (A) and predicted pathway of carnitine and acetylcarnitine degradation (B).

A. Organization of the carnitine degradation cluster (carR: LysR-type transcriptional regulator; mdh: malate dehydrogenase; aci01347: carnitine transporter; hyd: acetylcarnitine hydrolase; cntA and cntB: Rieske-type oxidoreductase complex; msadh: malic semialdehyde dehydrogenase).

B. Acetylcarnitine is suggested to be converted by the putative hydrolase Hyd to carnitine, which is further degraded by the CntAB complex to malic semialdehyde and TMA. Malic semialdehyde is suggested to be metabolized to malate by the malic semialdehyde dehydrogenase MsaDH followed by the conversion to pyruvate and CO₂ mediated by the putative malate dehydrogenase Mdh.

Acetylcarnitine and d-malate are degraded via the carnitine catabolic pathway

To analyse the role of hyd and mdh in carnitine catabolism markerless Δhyd and Δmdh mutants of A. baumannii were generated using a kanamycin resistance cassette (kanR) as positive selection marker and a levansucrase (sacB) for counter selection on sucrose. The deletion mutants were verified by sequencing. The Δhyd mutant was completely defective in growth with acetylcarnitine as sole carbon and energy source (Fig. 2B), whereas growth with acetate was unaffected (Fig. 2A). The Δhyd mutant was still able to grow with carnitine and exhibited a growth rate of 0.45 h⁻¹ comparable to the growth rate of the wild type (0.56 h⁻¹; Fig. 2C). However, the Δhyd mutant exhibited a prolonged lag phase of 14 h in comparison to the wild type cells which entered the exponential growth phase after 10 h of growth. The Δhyd mutant exhibited a slightly increased growth rate (0.29 h⁻¹) with d-malate (Fig. 2D) in comparison to the wild type (0.22 h⁻¹). Taken together these studies led to the conclusion that hyd encodes an acetylcarnitine hydrolase essential for the conversion of acetylcarnitine to carnitine.

Fig. 2

Growth studies of the A. baumannii ATCC 19606 wild type and the Δmdh, Δhyd and ΔcarR mutants in mineral medium with acetate, acetylcarnitine, carnitine or d-malate as sole carbon and energy source. A. baumannii ATCC 19606 wild type cells (●), Δhyd (◆), Δmdh (■) and ΔcarR mutants (▲) were grown in mineral medium with 20 mM acetate (A), 20 mM acetylcarnitine (B), 20 mM carnitine (C) or 20 mM d-malate (D) as sole carbon and energy source. Each value is the mean of ±S.E.M. of at least three independent measurements. The curves were fitted manually.

Growth experiments with the Δmdh mutant revealed that this mutant was completely defective in growth with acetylcarnitine (Fig. 2B) or carnitine (Fig. 2C) as sole carbon source, whereas growth with acetate was not affected (Fig. 2A). Moreover, the Δmdh mutant exhibited a significantly decreased growth rate with d-malate (0.08 h⁻¹) in comparison to the wild type (0.22 h⁻¹). The Δmdh mutant was unaffected in growth with l-malate as expected due to the presence of an l-malate dehydrogenase in the TCA cycle (data not shown). The impaired growth of the Δmdh mutant with d-malate together with the significant similarities and identities of the Mdh to the d-malic enzyme of R. capsulatus led to the conclusion that mdh encodes a d-malate dehydrogenase essential for the conversion of d-malate to pyruvate plus CO₂, the last step in carnitine catabolism (Tipton and Peisach, 1990; Martínez-Luque et al., 2001). The slow growth of the Δmdh mutant with d-malate as sole carbon source could be due to the presence of racemases leading to the conversion of d-malate to l-malate which can be oxidized via the l-malate dehydrogenase of the TCA cycle.

CarR is an activator of carnitine catabolism

The close association of a potential LTTR gene (carR) suggests a role of this potential regulator in transcriptional regulation of the carnitine gene cluster. To get insights into the role of CarR in regulation of the carnitine degradation pathway, a ΔcarR mutant was generated using the sacB system. The ΔcarR mutant was verified by sequencing. The ΔcarR mutant was completely defective in growth with acetylcarnitine (Fig. 2B) or carnitine (Fig. 2C), whereas growth with acetate was unaffected (Fig. 2A). Growth with d-malate was not completely abolished (Fig. 2D) but the growth rate was significantly reduced to 0.06 h⁻¹ comparable to the reduced growth rate of the Δmdh mutant. The growth defect of the ΔcarR mutant suggests that CarR is the transcriptional activator of the carnitine degradation cluster.

Genes of the carnitine catabolism pathway are organized in an operon

Next, we addressed the transcriptional organization of the carnitine degradation gene cluster by bridging PCR. Therefore, cDNA was used as template synthesized from RNA extracted from carnitine induced A. baumannii cells and primers were used that span the different open reading frames (Suppl. Table 1). The amplified intergenic regions are indicated in Fig. 3A and the detection of the amplified PCR products via agarose gel electrophoresis is shown in Fig. 3B. The results obtained demonstrate that mdh, aci01347, hyd, cntA, msadh and cntB form an operon. Control experiments showed that RNA was free from contamination with genomic DNA. No amplification of the intergenic region was observed in the absence of reverse transcriptase and in the absence of template. Furthermore, amplification of the intergenic region between mdh and carR and of the intergenic region between carR and the upstream located open reading frame HMPREF0010_01344 only led to PCR products with genomic DNA as template. Same holds true for the intergenic region between the cntB gene and HMPREF0010_01352. These findings led to the conclusion that the open reading frames HMPREF0010_01344 and HMPREF0010_01352 as well as the carR gene are not part of the operon of the carnitine catabolic pathway.

Fig. 3

The genes of the carnitine degradation pathway form an operon.

(A) Intergenic regions amplified by bridging PCRs. The intergenic regions are indicated by clamps below the genes and arrows above the genes indicate primer binding sites. Arrows indicate the related PCR analysis. The open reading frames upstream and downstream of the carnitine degradation cluster are designated with HMPREF0010_01344 and HMPREF0010_01352 respectively. 

(B) Electrophoretic analysis of the PCR products from the bridging PCRs. PCR products of the PCRs with genomic DNA as template are labelled with ‘gDNA’, those with cDNA as template are labelled with ‘cDNA’, those with RNA as template are labelled with ‘RNA’ and those without template are labelled with ‘H₂O’. 200 ng of PCR products were separated in a 0.8% (wt./vol.) agarose gel and the 100 bp DNA-ladder (Thermo Fisher) was used as marker.

Acetylcarnitine, carnitine and d-malate induced transcription of the carnitine operon

To get insights into the transcriptional regulation of the carnitine catabolic pathway we cultivated A. baumannii with acetylcarnitine, carnitine, d-malate and succinate as sole carbon and energy source respectively. Genome-wide expression profiling was performed using RNA isolated from cultures grown with these different substrates. These analyses revealed that genes of the carnitine degradation pathway were upregulated in the presence of acetylcarnitine, carnitine as well as d-malate by a log₂-fold change of 4.86–8.87 (Suppl. Table 2). Moreover, the transcriptome analyses revealed that the carnitine operon is transcribed at a basal level also under non-inducing conditions with succinate as carbon source; transcript numbers of 50–500 were detected for the genes of the carnitine catabolic pathway (aci01347: 371 ∓ 11, hyd: 559 ∓ 14, cntA: 420 ∓ 7, cntB: 275 ∓ 22, carR: 94 ∓ 12, mdh: 59 ∓ 2). The basal transcript levels guarantee a very fast adaptation upon availability of the inducer substrates. Interestingly, several genes not obviously linked to carnitine degradation were also significantly upregulated after growth with acetylcarnitine, carnitine or d-malate. One such upregulated gene cluster encodes potential key enzymes for degradation of aromatic compounds via the phenylacetic acid pathway or the homogentisate pathway (Suppl. Table 2). The latter is the central catabolic pathway for the degradation of l-phenylalanine and l-tyrosine. Moreover, several genes are significantly downregulated under these conditions, such as ribosomal genes, cold shock protein genes and many hypothetical proteins (Suppl. Table 3).

CarR binds constitutively to the intergenic DNA region between mdh and carR

Next, we analysed whether CarR binds to the intergenic DNA region between the first gene (mdh) of the carnitine operon and the carR gene and addressed the question of whether DNA binding is inducer substrate-dependent. To this end, electromobility shift assays (EMSA) were performed. Therefore, the regulator gene was cloned with an N-terminal His₆-Tag into the expression vector pBAD/HisA (Suppl. Fig. 1A) and His₆-CarR was produced in E. coli BL21 STAR cells and purified via Ni-NTA affinity chromatography. The deduced mass of the purified protein corresponds to the deduced molecular mass of CarR of 37.03 kDa (Suppl. Fig. 1B). Furthermore, the entire intergenic DNA region between carR and mdh plus 292 bp of the 5′ end of carR and 298 bp of the 5′ end of mdh was amplified via PCR (primer pair in Suppl. Table 1). Next, we addressed whether CarR binds to the 693 bp DNA fragment and indeed, preincubation of purified His₆-CarR resulted in a mobility shift (Fig. 4). Three different protein–DNA complexes were detected and designated complex 1, complex 2 and complex 3 (Fig. 4).

Fig. 4

CarR binds to a 693 bp DNA fragment spanning the intergenic region between mdh and carR.

(A) Binding of CarR to a 693 bp DNA fragment upstream of the carnitine degradation operon was analysed by EMSA studies. Different amounts of protein (0–81 pmol CarR) were incubated for 30 min with 233 fmol of the DNA fragment followed by separation on a 5% (vol./vol.) native polyacrylamide gel. DNA was visualized with ethidium bromide.

(B) The shift of the 693 bp DNA fragment spanning the intergenic region between mdh and carR was analysed in the presence or absence of 1 μg BSA to verify specificity of binding. 1 μg BSA was incubated for 30 min with 233 fmol of the DNA fragment followed by separation on a 5% (vol./vol.) native polyacrylamide gel. DNA was visualized with ethidium bromide.

(C) Binding of CarR to a non-specific DNA fragment was analysed with 233 fmol of a 1613 bp DNA fragment. The DNA was incubated for 30 min with 40.5 pmol CarR or 1 μg BSA followed by separation on a 5% (vol./vol.) native polyacrylamide gel. DNA was visualized with ethidium bromide.

Next, we analysed whether the inducer substrates carnitine and d-malate influence the binding of CarR to the DNA. These studies revealed that preincubation with carnitine or d-malate led to the same complexes as observed in the absence of inducer substrates (data not shown). These data indicate that the transcriptional regulator of the carnitine degradation operon CarR binds to the DNA independently of the presence of the inducer substrates. Binding specificity was analysed using the 693 bp fragment and an excess of BSA instead of CarR and by using a 1613 bp DNA fragment not linked to the carnitine operon for binding studies with CarR. In both cases no shift of the DNA fragments was observed. These results led to the conclusion that binding of CarR to the intergenic region between carR and mdh is specific.

Carnitine degradation plays an important role in infection of G. mellonella larvae

Next, we addressed the role of carnitine metabolism in virulence of A. baumannii by performing G. mellonella caterpillar infection studies with the A. baumannii ATCC 19606 wild type and the ΔcarR mutant in the absence (Fig. 5A) and in the presence (Fig. 5B) of carnitine. In the absence of carnitine, no difference in virulence of the wild type and the ΔcarR mutant was observed, the survival of the G. mellonella larvae after infection with the wild type or the ΔcarR mutant was comparable over 5 days (Fig. 5A). In the presence of carnitine during infection of the G. mellonella caterpillars a significant (p ≤ 0.05) increase in virulence was observed for the wild type cells (Fig. 5B), such as after one day only 25% of the infected caterpillars died in the absence of carnitine (Fig. 5A), and 75% of the larvae died after infection with wild type cell suspensions in PBS with 5 mM carnitine (Fig. 5B). In contrast, the ΔcarR mutant exhibited a significantly (p ≤ 0.05) lower virulence in the presence of carnitine in comparison to the wild type cells, such as the survival of G. mellonella after infection with ΔcarR mutant in the presence of carnitine was comparable to the survival in the absence of carnitine. Here, 60% of the larvae with ΔcarR injection survived after 1 day in the presence of carnitine and 75% in the absence of carnitine. These results suggest that carnitine degradation plays an important role in virulence of A. baumannii.

Fig. 5

Carnitine and acetylcarnitine degradation of A. baumannii plays a role in G. mellonella infection. Survival of G. mellonella caterpillars after infection with (A) the wild type (black solid line) or the ΔcarR mutant (orange dashed line) in the absence of carnitine or acetylcarnitine. (B) The wild type (black solid line) or the ΔcarR mutant (orange dashed line) in the presence of 5 mM carnitine. (C) The wild type or the Δhyd mutant in the presence of 5 mM acetylcarnitine or 5 mM carnitine (wild type with carnitine: grey solid line; wild type with acetylcarnitine: black solid line; Δhyd with carnitine: blue dotted line; Δhyd with acetylcarnitine: orange dashed line). (D) The wild type in the absence (black solid line) or in the presence of 5 mM acetate (blue dotted line) or 5 mM d-malate (orange dashed line). Wild type, ΔcarR and Δhyd cells were grown to stationary growth phase in LB, harvested and washed three times in sterile PBS and finally re-suspended to an OD₆₀₀ of 1.5 in PBS, PBS with 5 mM carnitine or PBS with 5 mM acetylcarnitine. 10 μl (~1 × 10⁶ CFU) was injected into each G. mellonella larvae. Pre- and post-selection was performed as described in Experimental procedures. One representative out of at least three experiments is shown.

TMA formation during carnitine catabolism is most likely important for A. baumannii virulence

The increased killing of G. mellonella in dependence of carnitine degradation could be due to a toxic effect of the TMA formed during carnitine oxidation. To test this hypothesis, we performed G. mellonella infection studies with the A. baumannii ATCC 19606 wild type and the Δhyd mutant in the presence of acetylcarnitine (Fig. 5C), the latter was no longer able to convert acetylcarnitine to carnitine and therefore TMA production is abolished. Acetylcarnitine also increased killing of caterpillars, but this was abolished in the Δhyd (Fig. 5C). When carnitine was supplied to the Δhyd mutant, the effect was restored. These data indicate that metabolism of carnitine or acetylcarnitine is responsible for this effect. Oxidation of both substrates leads to the production of TMA, a potent inhibitor of microbial growth. Degradation of the carnitine pathway intermediate d-malate leads to pyruvate and if indeed TMA production is responsible for the increased killing of G. mellonella in the presence of carnitine or acetylcarnitine the presence of d-malate during G. mellonella infection studies should not lead to this effect. Indeed, the presence of d-malate during infection had no significant effect on survival of G. mellonella larvae (Fig. 5D). The same was observed with acetate. Taken together these findings suggest that TMA formation during degradation of carnitine has an effect on the viability of G. mellonella larvae. However, further studies are required to provide clear evidence that TMA is causing the increased G. mellonella killing.

Discussion

In this study, we have addressed the role of hyd and mdh in carnitine degradation of A. baumannii. We demonstrated that hyd mediates the conversion of acetylcarnitine to carnitine and that mdh is essential for the conversion of the carnitine pathway intermediate d-malate. d-malate is formed by a carnitine oxidoreductase (CntAB) that produces TMA in addition (Zhu et al., 2014). A carnitine degradation pathway via malate and TMA was first discovered in the mid-1960s in Serratia marcescens and up to now has been only further detected in Acinetobacter calcoaceticus, later identified as A. baumannii strain (Kleber et al., 1977; Ditullio et al., 1994; Zhu et al., 2014; Meadows and Wargo, 2015).

Our finding that a Δmdh mutant is significantly impaired in growth with d-malate together with the 90% similarity of the Mdh to the d-malic enzyme of R. capsulatus suggests that the Mdh of A. baumannii is decarboxylating d-malate to pyruvate. Enzymes of this superfamily catalyse the oxidation and subsequent decarboxylation of different d-malate-based substrates such as d-malate, l-tartrate, 3-isopropylmalate and d-isocitrate with different specificities with various substrate recognition sites. To date, four different substrate recognition motifs have been identified (Vorobieva et al., 2014). The isopropylmalate dehydrogenases (IPMDH) which exhibit highest activities with alkylmalates contain the substrate recognition motif RXGXLLXXR (Miyazaki et al., 1993; Dean and Dvorak, 1995; Vorobieva et al., 2014), whereas the isocitrate dehydrogenases contains a specific serine instead of glutamate and asparagine instead of lysine (XXSXNXXXR) (Stokke et al., 2007; Vorobieva et al., 2014). The d-malate/tartrate dehydrogenases (Mdh/Tdh) mediate the conversion of d-malate to pyruvate and CO₂ and of tartrate to dihydroxyfumarate. These Mdh/Tdh enzymes contain the substrate recognition site LXXXLXXXR (Giffhorn and Kuhn, 1983; Martínez-Luque et al., 2001; Vorobieva et al., 2014). In contrast, members of a subcluster of the malate/tartrate dehydrogenases, the TtuC-like proteins, which mediate the conversion of tartrate and not d-malate contain the substrate-binding site (LXXXRXXXC) (Crouzet and Otten, 1995). The Mdh of A. baumannii contains the LXXXLXXXR substrate recognition motif which is consistent with our suggestion that the Mdh from A. baummannii ATCC19606 mediates the conversion of d-malate to pyruvate and CO₂.

The significant similarities of CarR to LTTR and the close association of the carR gene with the carnitine catabolic operon made it a good candidate for a regulator of the carnitine catabolic pathway and indeed mutant studies suggest that CarR is the activator of the carnitine catabolic pathway. LTTR have a helix-turn-helix motif at the N-terminus which is important for DNA binding. Such a helix-turn-helix motif is also present (amino acid 7–66) at the N-terminus of CarR (pfam00126). Also characteristic for LTTR is the PBP2_CrgA_like_9 domain at the C-terminus (CDD:176168) (Lu et al., 2020), consisting of two typical Rossmann folds connected by a linker region in between, which is necessary for substrate binding and dimerization (Lochowska et al., 2001; Ezezika et al., 2007). This domain is also present in CarR spanning the amino acid 95–291. Our finding that the transcription of the carnitine operon is dependent on carnitine or d-malate suggests that d-malate and carnitine binding to CarR leads to conformational changes in the CarR proteins affecting the CarR/DNA complexes thereby leading to binding of the RNA polymerase at the promoter of the carnitine operon followed by gene expression (Maddocks and Oyston, 2008).

Carnitine is abundant in the human host, such as 70 mmol is present in a 70 kg adult (Engel and Rebouche, 1984). This abundance of carnitine in the human host makes it a good substrate for pathogenic bacteria, as shown for P. aeruginosa, which metabolizes carnitine to glycine betaine and further to pyruvate (Wargo and Hogan, 2009). This carnitine degradation pathway is linked to virulence such as glycine betaine induces the expression of different virulence factors such as phospholipase C and phosphorylcholine phosphatase (Lisa et al., 1994; Lucchesi et al., 1995; Wargo et al., 2009). Our finding that the presence of carnitine or acetylcarnitine in G. mellonella infection studies with A. baumannii wild type cells led to significantly decreased survival rates of G. mellonella larvae suggests that degradation of carnitine is also linked to virulence. The decreased survival of G. mellonella larvae in the presence of carnitine might be due to the production of TMA. This is also supported by our finding that mutants defect in carnitine degradation led to G. mellonella larvae survival rates comparable to those after infection with A. baumannii in the absence of carnitine. These findings suggest that TMA which is released during carnitine oxidation might be responsible for the decreased survival rates. This is in line with the finding that TMA is a toxic compound for vertebrates such as shown for Necrophorus orbicollis (Abbott, 1936). Here, injection of TMA in low concentration led to a rapid death of the insects. Furthermore, it has to be noted that increased TMA production in the human host due to carnitine degradation of infecting A. baumannii might also affect the health of the human host since TMA is metabolized in the human liver by an flavin-dependent monooxygenase to TMAO (Miao et al., 2015) which was shown to cause prevalent cardiovascular disease and enhance the risk for myocardial infarcts and stroke (Koeth et al., 2013). Therefore, it is tempting to speculate that carnitine degradation in the human host might not only contribute to metabolic adaptation and virulence of A. baumannii but might also trigger cardiovascular diseases. Recently, we generated gene cluster abundance profiles of different metabolic traits in Acinetobacter species and found that the ability to metabolize carnitine occurs inside and outside the A. calcoaceticus–baumannii (ACB) complex (Djahanschiri et al., 2022). This complex represents a group of closely related human opportunistic pathogens. However, 25 of 31 analysed strains outside the ACB complex exhibiting a carnitine degradation pathway were isolated from infected patients. Moreover, the carnitine cluster is absent in A. calcoaceticus which is the only non-pathogenic species in the ACB clade. Taken together, the presence of the carnitine gene cluster correlates well with the pathogenicity of the isolates.

Transcriptomic analyses revealed that growth of A. baumannii with carnitine, acetylcarnitine and d-malate led to a significant upregulation of the homogentisate and phenylacetic acid pathway genes. The homogentisate pathway is the key catabolic pathway for degradation of phenylalanine, 3-hydroxyphenylacetate and tyrosine (Arias-Barrau et al., 2004). The upregulation of these degradation pathways in the presence of carnitine might promote the adaptation to the human host due to the high abundance of phenylalanine, tyrosine and phenylacetic acid in the host.

Acknowledgements

This study was supported by a grant from the Deutsche Forschungsgemeinschaft through DFG Research Unit FOR2251.

Open Research

Data Availability

Transcriptome data have been deposited in the National Center for Biotechnology Information's (NCBI) Sequence Read Archive (SRA) as BioProject PRJNA821016. All other data of this study are available from the corresponding author upon reasonable request.