A Disjointed Pathway for Malonate Degradation by Rhodopseudomonas palustris

There is interest in understanding how bacteria metabolize malonate because this three-carbon dicarboxylic acid can serve as a building block in bioengineering applications to generate useful compounds that have an odd number of carbons. We found that the phototrophic bacterium Rhodopseudomonas palustris grows extremely slowly on malonate. We identified two enzymes and two TRAP transporters involved in the uptake and metabolism of malonate, but some of these elements are apparently not very efficient. R. palustris cells growing with malonate have the potential to be excellent biocatalysts, because cells would be able to divert cellular resources to the production of value-added compounds instead of using them to support rapid growth. In addition, our results suggest that R. palustris is a candidate for directed evolution studies to improve growth on malonate and to observe the kinds of genetic adaptations that occur to make a metabolic pathway operate more efficiently.

ABSTRACT

The purple nonsulfur phototrophic bacterium Rhodopseudomonas palustris strain CGA009 uses the three-carbon dicarboxylic acid malonate as the sole carbon source under phototrophic conditions. However, this bacterium grows extremely slowly on this compound and does not have operons for the two pathways for malonate degradation that have been detected in other bacteria. Many bacteria grow on a spectrum of carbon sources, some of which are classified as poor growth substrates because they support low growth rates. This trait is rarely addressed in the literature, but slow growth is potentially useful in biotechnological applications where it is imperative for bacteria to divert cellular resources to value-added products rather than to growth. This prompted us to explore the genetic and physiological basis for the slow growth of R. palustris with malonate as a carbon source. There are two unlinked genes annotated as encoding a malonyl coenzyme A (malonyl-CoA) synthetase (MatB) and a malonyl-CoA decarboxylase (MatA) in the genome of R. palustris, which we verified as having the predicted functions. Additionally, two tripartite ATP-independent periplasmic transporters (TRAP systems) encoded by rpa2047 to rpa2049 and rpa2541 to rpa2543 were needed for optimal growth on malonate. Most of these genes were expressed constitutively during growth on several carbon sources, including malonate. Our data indicate that R. palustris uses a piecemeal approach to growing on malonate. The data also raise the possibility that this bacterium will evolve to use malonate efficiently if confronted with an appropriate selection pressure.

IMPORTANCE There is interest in understanding how bacteria metabolize malonate because this three-carbon dicarboxylic acid can serve as a building block in bioengineering applications to generate useful compounds that have an odd number of carbons. We found that the phototrophic bacterium Rhodopseudomonas palustris grows extremely slowly on malonate. We identified two enzymes and two TRAP transporters involved in the uptake and metabolism of malonate, but some of these elements are apparently not very efficient. R. palustris cells growing with malonate have the potential to be excellent biocatalysts, because cells would be able to divert cellular resources to the production of value-added compounds instead of using them to support rapid growth. In addition, our results suggest that R. palustris is a candidate for directed evolution studies to improve growth on malonate and to observe the kinds of genetic adaptations that occur to make a metabolic pathway operate more efficiently.

INTRODUCTION

Malonate is a three-carbon compound that was reported in 1952 to be present in leguminous plants (1, 2). It is also produced in small amounts as a bacterial fermentation product, and the dimethyl and diethyl esters of malonic acid have been produced industrially in large amounts (3). Malonate is a good building block for compounds that have an odd number of carbons. In fact, several bioengineering applications have been reported that incorporate enzymes involved in generating malonyl coenzyme A (malonyl-CoA) as a component of processes for polyketide and flavonoid production in Saccharomyces cerevisiae and Streptomyces, respectively (4, 5). Malonate can also have a negative impact on industrial processes, as it inhibits succinate dehydrogenase and lignin degradation (6, 7).

Numerous bacteria have been reported to grow on malonate as the sole carbon and energy source or, anaerobically, as the sole energy source (8–10). All pathways for malonate degradation involve a decarboxylation step that converts either malonyl-S-acyl carrier protein (malonyl-ACP) or malonyl-CoA to acetate or acetyl-CoA (11–13). When the ACP-dependent pathway operates anaerobically in Malonomonas rubra, malonyl-ACP decarboxylation results in the generation of CO₂, acetate (a nonfermentable carbon source), and a proton motive force that supplies energy (14). When this pathway operates aerobically in other bacteria, it also generates acetate and CO₂, but the acetate is used as a carbon and energy source by well-known pathways, including the glyoxylate shunt, which is a variation of the tricarboxylic acid cycle. This malonate pathway has been characterized in Klebsiella pneumoniae and Acinetobacter baylyi (14, 15). In A. baylyi, malonate metabolism is specified by two adjacent operons, one encoding a transporter and the other encoding the malonate degradation enzymes. In K. pneumoniae and A. baylyi, a LysR regulator activates the expression of malonate degradation genes upon binding malonate (15, 16).

A more straightforward pathway for malonate degradation involves activation of malonate to malonyl-CoA by a malonyl-CoA synthetase named MatB, followed by decarboxylation of malonyl-CoA to CO₂ and acetyl-CoA by malonyl-CoA decarboxylase (MatA) (17). This pathway, first detected in Rhizobium leguminosarum bv. trifolii, is encoded by a three-gene operon, matABC. MatC is a transport protein with similarity to a dicarboxylate carrier protein. MatR, a protein encoded by a divergent regulatory gene, is a member of the FadR subfamily of regulators (18).

In this work on malonate utilization by the phototrophic bacterium Rhodopseudomonas palustris, we found that this bacterium grows very slowly with malonate as the sole carbon source. We thought that perhaps a low growth rate could be advantageous in applications where R. palustris is genetically engineered to serve as a biocatalyst to produce value-added products, as it would have to divert relatively few cellular resources to growth. This motivated us to determine the genetic and physiological basis for the slow growth of R. palustris on malonate. We found that this species has malonate utilization genes homologous to some of those present in Rhizobium, but these genes are unlinked and dysregulated. We also identified two transport systems for malonate.

RESULTS

R. palustris grows very slowly with malonate as the sole carbon source.

For this work, we grew R. palustris photoheterotrophically, that is, anaerobically in light with a nonfermentable compound as the sole carbon source. Under these conditions, light supplies energy for growth via cyclic photophosphorylation. We found that while R. palustris grew on acetate with an expected generation time of 6.4 ± 0.5 h (19), it grew extremely slowly on malonate, with a generation time of 50 ± 3 h (Table 1).

TABLE 1.

Generation times of the wild type, the ΔmatB and ΔmatA mutants, and the complemented and overexpressing strains with malonate, acetate, and succinate as carbon sources^a

Strain or genotype	Generation time (h)
	Malonate	Acetate	Succinate
WT	50 ± 3	6.7 ± 0.1	8.2 ± 0.3
ΔmatB	112 ± 7	7.1 ± 0.1	10.0 ± 0.4
ΔmatA	NG	6.5 ± 0.2	17.8 ± 0.9
WT (pBBR)	49 ± 4	—	—
ΔmatB (pBBR-matB)	56 ± 5	—	—
ΔmatA (pBBR-matA)	51 ± 4	—	—
WT (pBBR-matB)	52 ± 9	—	—
WT (pBBR-matA)	59 ± 1.5	—	—

^aWT, wild type; NG, no growth; —, not detected.

Characterization of the metabolic route of malonate utilization in R. palustris.

R. palustris strain CGA009 has a gene annotated as a malonyl-CoA synthetase gene (rpa0221) and another gene annotated as a malonyl-CoA decarboxylase gene (rpa0560). We named these genes matB and matA, respectively. The MatB enzyme from R. palustris has been purified, its activity studied, and its structure determined (9). Transcriptomic and proteomic analyses of cells grown with malonate as the sole carbon source compared to cells grown with acetate did not reveal significant changes in MatB (RPA0221) transcription or protein abundance (Fig. 1A). However, the abundance of MatA increased approximately 7-fold when malonate was supplied as the substrate (Fig. 1A). As would be expected of slow-growing cells, genes encoding ribosomal proteins and translation elongation factors were downregulated significantly in malonate-grown cells (see Table S1 in the supplemental material). This is likely a consequence of the low growth rate of such cells.

FIG 1.

Relative transcription changes of matB and matA genes or MatB and MatA proteins from transcriptomic and proteomic experiments, respectively (A), and from RT-qPCR analysis (B), for R. palustris growth on malonate relative to R. palustris growth on acetate.

Reverse transcription-quantitative PCR (RT-qPCR) analysis also demonstrated that there were no significant differences in the levels of matB transcription between acetate- and malonate-grown cells. Consistent with our proteomic data, matA was approximately 8-fold more highly expressed in cells grown on malonate than in cells grown on acetate, as measured by RT-qPCR (Fig. 1B). It is possible that the increased expression of matA that we observed in cells grown on malonate was induced by this compound, but at this point we cannot exclude the possibility that the higher level of expression of this gene is linked to the low growth rate on malonate rather than to the compound itself.

To confirm the functions of R. palustris MatB and MatA, His-tagged versions of these proteins were expressed in Escherichia coli and purified. Their catalytic activities were measured by detecting malonyl-CoA and acetyl-CoA using high-performance liquid chromatography (HPLC). The samples of the reaction mixture had peaks at the same retention times as the malonyl-CoA and acetyl-CoA standards (Fig. 2A and B). We next measured the activity of MatB with various substrates, including malonate, acetate, butyrate, succinate, and propionate, and found that the enzyme had significant activity only with malonate. The K_m and k_cat values of MatB obtained using malonate as the substrate were 150 ± 19 μM and 6 ± 0.2 s⁻¹, respectively, which were approximately equal to the values previously reported (9, 20). A previous report showed that this enzyme is also active with methylmalonate, ethylmalonate, and butylmalonate (9).

FIG 2.

Detection of enzymatic activities of MatB (A) and MatA (B) in vitro by HPLC. “Standard” refers to pure malonyl-CoA or acetyl-CoA; “Sample” refers to products of purified MatB or MatA given malonyl-CoA or malonate.

As further confirmation of the involvement of matA and matB in malonate metabolism, we found that a ΔmatB mutant grew extremely slowly on malonate, with a generation time of 112 ± 7 h, and a ΔmatA deletion mutant did not grow at all (Table 1). Even though it grew slowly, the ΔmatB mutant still reached the same final cell density as the wild type. This and its low but measurable growth rate suggest that R. palustris encodes additional CoA synthetases that can convert malonate to malonyl-CoA (21, 22). Crosby et al. (9) reported similar results for the growth of an R. palustris ΔmatB mutant on malonate but found that this strain could not grow at all on methylmalonate. Hence, they designated MatB as a methylmalonyl coenzyme A synthetase (9). The ΔmatB and ΔmatA mutants grew normally on acetate, but the ΔmatA mutant grew more slowly on succinate than the wild type, suggesting that this enzyme might play a role in succinate metabolism.

Transporters of malonate in R. palustris.

Malonate is not freely permeative across cell membranes and requires active transport to enter cells. Numerous dicarboxylate transporters have been reported that can transport malonate. MdcF is a membrane protein that functions as a malonate carrier in K. pneumoniae and belongs to Interpro group IPR004776 (11). A two-subunit malonate-sodium symporter, named MdcLM, is present in A. baylyi (15). An integral membrane protein, MatC, acts as a malonate carrier in R. leguminosarum bv. trifolii (17). Tripartite ATP-independent periplasmic (TRAP) transporters are periplasmically binding protein-dependent transporters that function in dicarboxylate transport (23–28). TRAP transporters use the electrochemical gradient to drive solute transport. The best-characterized system is the high-affinity C₄-dicarboxylate transport system, named Dct, which is composed of a periplasmic binding protein receptor subunit (DctP) and two integral membrane proteins, namely, DctQ and DctM (25). An aerobic TRAP transporter for dicarboxylic acids (Dct), composed of DctA and DctB, and two homologous anaerobic C₄-dicarboxylate membrane transporters (DcuA and DcuB) were found in Escherichia coli (29, 30). MatC from R. leguminosarum bv. trifolii is similar to DcuA and DcuB (17).

R. palustris encodes eight TRAP systems (Fig. 3). In some cases, the dctQ and dctM paralogs are fused, as has been seen in archaea and some bacteria (25). DctP can sometimes be replaced by a TRAP-associated extracytoplasmic immunogenic protein (TAXI). TRAP transporters that contain a TAXI protein always exhibit fusion of the two integral membrane subunits DctM and DctQ (23). There is one such protein set encoded by R. palustris.

FIG 3.

Comparison of TRAP system gene clusters in R. palustris.

In our proteomic analysis, the abundance of DctP encoded by rpa2543 increased 35-fold in cells grown on malonate, suggesting that the TRAP system RPA2541-2543 might be involved in malonate transport. We deleted all the TRAP transporter gene sets except rpa4554 to rpa4556 and found that the Δrpa2047-2049 and Δrpa2541-2543 mutants had significantly lower growth rates with malonate than wild-type cells (Fig. 4). A double Δrpa2047-2049 Δrpa2541-2543 deletion mutant grew extremely slowly on malonate, with a 231-h generation time (Table 2). The Δrpa2047-2049 mutant also grew slowly on succinate, suggesting that this TRAP system transports both succinate and malonate. Based on these results, we can conclude that R. palustris has two TRAP systems that can transport malonate.

FIG 4.

Growth rates of TRAP system mutants and wild type with malonate as the carbon source. Cells were grown photoheterotrophically with light as the energy source.

TABLE 2.

Generation times of TRAP mutants with malonate, acetate, and succinate as carbon sources

Strain or mutation	Generation time (h)
	Malonate	Acetate	Succinate
WT	50 ± 3	6.7 ± 0.1	8.2 ± 0.3
Δrpa1782-1784	54 ± 5	5.9 ± 0.1	9.5 ± 0.1
Δrpa1976-1977	56 ± 4	7.0 ± 0.2	11.3 ± 0.5
Δrpa2047-2049	86 ± 6	6.7 ± 0.1	27.9 ± 1.4
Δrpa3458-3459	58 ± 5	6.7 ± 0.1	15.1 ± 0.3
Δrpa2541-2543	72 ± 6	7.1 ± 0.2	9.6 ± 1.1
Δrpa4509-4510	58 ± 5	6.3 ± 0.1	8.6 ± 0.6
Δrpa4554-4556	47 ± 5	7.2 ± 0.3	13.2 ± 0.6
Δrpa2047-2049 Δrpa4509-4510	231 ± 1	6.8 ± 0.1	22.1 ± 0.5

Regulation of malonate metabolism in R. palustris.

A GntR family transcriptional regulator, encoded by rpa1467, shares 42% identity with MatR from R. leguminosarum bv. trifolii. MatR is a transcriptional repressor, and the transcription of mat genes is derepressed in response to malonate (18). A ΔRPA1467 mutant grew approximately half as fast (85 ± 5 h) on malonate as the wild type, suggesting that RPA1467 could act as a transcriptional activator, but the transcription levels of matB and matA were not significantly different in the Δrpa1467 mutant compared to the wild type growing on acetate alone or on acetate with the addition of malonate. Although rpa1467 does not appear to regulate malonate metabolic genes, we cannot exclude the possibility that it regulates genes for malonate transport.

DISCUSSION

This work shows that R. palustris has elements of malonate metabolism similar to those of other alphaproteobacteria, but unlike in other alphaproteobacteria, the genes involved are not linked and at least some are expressed constitutively. The predicted functions of MatB and MatA in the conversion of malonate to acetyl-CoA were confirmed by a combination of enzyme assays and mutant analysis. We also identified two TRAP transporters encoded by rpa2407 to rpa2409 and rpa2541 to rpa2543 that are likely involved in malonate transport in R. palustris. We found that overexpression of MatB and MatA in wild-type cells did not increase their growth rate (Table 1), indicating that the transport of malonate may be the rate-limiting step in malonate metabolism. We failed to identify a regulator that directly controls matB expression.

Diagrams of the pathway for acetate metabolism and our predicted pathway for malonate metabolism in R. palustris are shown in Fig. 5. Part of the CO₂ produced in central metabolism is fixed by the Calvin-Benson-Bassham (CBB) cycle. We propose that malonate is transported into cells by a TRAP system, converted to malonyl-CoA by MatB, and decarboxylated by MatA to form acetyl-CoA and CO₂.

FIG 5.

Metabolic routes of acetate (A) and malonate (B) in R. palustris.

The slow growth of R. palustris on malonate coupled with a lack of tight regulation of malonate utilization suggests that the malonate degradation pathway has not fully evolved for efficiency in this organism. Some or all of the genes for malonate metabolism, such as the malonyl-CoA decarboxylase gene, could have been acquired by R. palustris via horizontal gene transfer or from a gene duplication event. R. palustris has over 40 CoA synthetase genes, many with overlapping substrate specificities (21, 22), and one of these genes encodes MatB, which exhibits good activity with malonate. Since two different TRAP transporters allow growth on malonate and since the rate of growth on malonate appears to be limited by transport, it seems likely that these transporters may play a “moonlighting” role in bringing malonate into cells. R. palustris is a good candidate for directed evolution studies to improve growth on malonate.

MATERIALS AND METHODS

Strains and growth conditions.

For anaerobic cultivation, R. palustris strains were grown in defined mineral medium (photosynthetic medium [PM]) supplemented with 13.3 mM malonate, 20 mM acetate, or 10 mM succinate in light at 30°C. All R. palustris strains were grown aerobically during genetic manipulation on PM agar supplemented with 10 mM succinate at 30°C. The growth medium was degassed and extensively bubbled with N₂. The medium was dispensed into anaerobic culture tubes in an anaerobic glove box. E. coli strains were grown in LB medium at 37°C and with shaking at 180 rpm. When appropriate, R. palustris was grown with gentamicin and kanamycin at 100 μg/ml and 50 μg/ml, respectively. E. coli cultures were supplemented with gentamicin and kanamycin at 20 μg/ml and 100 μg/ml, respectively (31). All the strains and plasmids used in this work are listed in Table S2 in the supplemental material.

Genetic manipulation of R. palustris.

To generate the suicide plasmid for genetic operations, the upstream and downstream homologous arms of the target genes were amplified using the Phanta Super-Fidelity DNA polymerase (Vazyme Corp.). The two homologous arms were then linked with the linearized pJQ200SK vector using the ClonExpress Ultra one-step cloning kit (Vazyme Corp.). The suicide plasmids were verified by restriction enzyme digestion and sequencing and transformed into E. coli S17-1. The suicide plasmids were transferred from E. coli S17-1 to R. palustris strains by conjugation, and appropriate recombinants were identified by antibiotic resistance screening and sucrose counterselection. All the mutant strains were verified using PCR amplification and sequencing of the resulting PCR product. All primers used are listed in Table S3.

Real-time quantitative PCR.

RNA was extracted from cells using RNAiso Plus (TaKaRa Corp.) according to the manufacturer’s instructions. Reverse transcription assays were carried out using the PrimeScript RT reagent kit with gDNA Eraser (Perfect Real Time) (TaKaRa Corp.). One microgram of total RNA was used for every 20-μl reverse transcription system to obtain cDNA under the following conditions: 42°C for 2 min, 37°C for 15 min, and 85°C for 5 s. The obtained cDNA and SYBR Premix ExTaq (Tli RNase H Plus) (TaKaRa Corp.) were used in the real-time quantitative PCR systems, which were performed with qTOWER³ (Analytic Jena). All the primers used for RT-qPCR analysis are listed in Table S3. Here, the gapdh gene in R. palustris, encoding a glyceraldehyde-3-phosphate dehydrogenase, was used as the reference gene for normalization. Relative expression was calculated using the comparative ΔΔC_T method, and the values were expressed as 2^−ΔΔCT (32). Three independent replicates were performed for all experiments. All the values shown in this work are the mean values of three independent replicates showing fold changes.

Purification of MatB and MatA.

MatB and MatA were expressed and purified as follows. The corresponding genes of R. palustris were amplified by PCR using the primer pairs matB-F/matB-R and matA-F/matA-R (see Table S3), and the PCR products of matB and matA were cloned into the plasmids pACYCDuet-1 and pET28a, respectively. Positive clones were verified by colony PCR and sequencing and transformed into E. coli BL21(DE3), in which the proteins were expressed after induction with IPTG. Next, the proteins were purified using a HisTrap HP column (GE Healthcare), and the concentrations of the purified proteins were determined using the Bradford assay.

HPLC analysis.

The detection of malonyl-CoA and acetyl-CoA was conducted by HPLC using a Thermo Fisher Ultimate 3000 system with a C₁₈ column (250 mm by 4.6 mm, 5 μm; Thermo Fisher Scientific). The concentration of malonyl-CoA was determined utilizing a previously described method with a detection wavelength of 245 nm (33). The concentration of acetyl-CoA was determined as follows. The mobile phase consisted of solvent A (20 mM sodium phosphate, pH 5.0) and solvent B (800 ml of 250 mM sodium phosphate, pH 5.0, mixed with 200 ml of acetonitrile). The gradient elution procedure was as follows: 0 to 5 min, 3.0% B; 5 to 7.5 min, 18% B; 7.5 to 12.5 min, 28% B; 12.5 to 18 min, 40% B; 18 to 20 min, 42% B; after 20 min, 3.0% B. The flow rate was 1 ml/min with a column temperature of 25°C. The concentration of acetyl-CoA was determined by a UV detector at a wavelength of 254 nm.

Proteomic analysis.

Proteomic analysis was conducted as described previously, with modifications (19). Cells were lysed using ultrasonication in 50 mM NH₄HCO₃ buffer. The protein concentration was quantified using Bradford reagent (Sango). Then, 200-μg aliquots of protein were mixed with the same volume of 50 mM NH₄HCO₃ containing 8 M urea and 10 mM dithiothreitol (DTT), followed by incubation at 37°C for 45 min. The reactions were then diluted 4-fold in 50 mM NH₄HCO₃. Protein digestion was carried out overnight with incubation at 37°C with a 1/100 trypsin/protein ratio, followed by a second digestion for 4 h at 37°C with the same trypsin/protein ratio. Digested peptides were desalted in SOLAμ horseradish peroxidase (HRP; 2 mg/1 ml) 96-well plates (catalog no. 60209-001; Thermo Fisher Scientific) and dried in a vacuum centrifuge and the concentration was adjusted to 1 μg/μl before liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis.

Each experimental condition was represented by biological triplicates. All samples were block randomized before analysis on the mass spectrometer. A total of 1 μg of digested peptides was loaded into a trap column (2 cm by 75 μm [inside diameter], packed with 3 μm PepMap C₁₈ stationary phase; Thermo Fisher Scientific) and separated in a capillary column (25 cm by 75 μm [inside diameter], packed with 2 μm PepMap C₁₈ stationary phase; Thermo Fisher Scientific) with the following gradient: 5% to 10% solvent B for 1 min, 10% to 30% B for 45 min, 30% to 45% B for 20 min, 45% to 90% B for 6 min, and a hold for 3 min in 95% B. Solvent A was composed of 0.1% formic acid in water, and solvent B was 0.1% formic acid in 80% acetonitrile. The eluted peptides were directly analyzed in an Orbitrap mass spectrometer (Orbitrap Fusion Lumos; Thermo Fisher Scientific). Full MS scans were collected over a range of 350 to 1,500 m/z and a resolution of 120,000 at m/z 200. The ions with the most intense signals with ≥2 charges were subjected to higher-energy collision dissociation (HCD) with a normalized collision energy of 30 and resolution of 15,000 at m/z 200. Each parent ion was fragmented once before being dynamically excluded for 60 s.

Analysis of proteomic data.

Peptides were identified and quantified with Proteome Discoverer software (2.2). R. palustris ATCC BAA-98 (strain CGA009) sequences were downloaded from the UniProt Knowledgebase. Database searches were performed by SequestHT. The parameter considered for the analysis was oxidation of methionine as a variable modification. A precursor mass tolerance of 10 ppm and fragment mass tolerance of 0.02 Da were used to search the database. Identified peptides were filtered with a 1% false discovery rate at the peptide-spectrum match and protein levels. The “matching between runs” function was enabled for extracting peak areas to detect peptide features that were not identified in all the LC-MS/MS runs.

Protein abundance values from Proteome Discoverer were used to identify significantly changed proteins between conditions. The growth-with-acetate condition was set as the reference.

Transcriptomic analysis.

RNA was extracted from cells using RNAiso Plus (TaKaRa Corp.) according to the manufacturer’s instructions. RNA purity was checked using a NanoPhotometer spectrophotometer (Implen, Westlake Village, CA). RNA concentration was measured using a Qubit RNA assay kit in a Qubit 2.0 fluorometer (Life Technologies, Carlsbad, CA). RNA integrity was assessed using the RNA Nano 6000 assay kit of the Bioanalyzer 2100 system (Agilent Technologies, Santa Clara, CA). A total amount of 3 μg of RNA per sample was used as input material for RNA sample preparation. Sequencing libraries were generated using the NEBNext Ultra directional RNA library prep kit for Illumina (New England Biolabs, Ipswich, MA) following the manufacturer’s recommendations, and index codes were added to attribute sequences to each sample. Briefly, mRNA was purified from total RNA using poly(T) oligo-attached magnetic beads. For prokaryotic samples, rRNA was removed using a specialized kit. Fragmentation was carried out using divalent cations at an elevated temperature in NEBNext first-strand synthesis reaction buffer (5×). First-strand cDNA was synthesized using random hexamer primers and Moloney murine leukemia virus (M-MuLV) reverse transcriptase (RNase H). Second-strand cDNA synthesis was subsequently performed using DNA polymerase I and RNase H. In the reaction buffer, deoxynucleoside triphosphates (dNTPs) with dTTP replaced by dUTP were used. The remaining overhangs were converted to blunt ends via exonuclease and/or polymerase activities. After adenylation of the 3′ ends of DNA fragments, NEBNext Adaptor with a hairpin loop structure was ligated to prepare the sample for hybridization. To preferentially select cDNA fragments that were ∼150 to ∼200 bp in length, the library fragments were purified with the AMPure XP system (Beckman Coulter, Beverly, MA). Next, 3 μl of USER enzyme (New England Biolabs) was used with size-selected, adapter-ligated cDNA at 37°C for 15 min followed by 5 min at 95°C before PCR. Then, PCR was performed with Phusion high-fidelity DNA polymerase, universal PCR primers, and the index primer. Finally, the products were purified (AMPure XP system), and library quality was assessed on the Agilent Bioanalyzer 2100 system. The clustering of the index-coded samples was performed in the cBot cluster generation system using the TruSeq PE cluster kit v3-cBot-HS (Illumina) according to the manufacturer’s instructions. After cluster generation, the library preparations were sequenced on an Illumina HiSeq platform, and paired-end reads were generated.

Analysis of transcriptomic data.

Raw data (raw reads) in fastq format were first processed through in-house Perl scripts. Clean data (clean reads) were obtained by removing reads containing adapters, reads containing poly(N) sequences, and low-quality reads from the raw data. Then, the Q20, Q30, and GC content of the clean data were calculated. All downstream analyses were based on the clean data with high quality. Reference genome and gene model annotation files were downloaded from the genome website directly. Both index building of the reference genome and aligning of clean reads to the reference genome were performed by using Bowtie2-2.2.3 (34). HTSeq v0.6.1 was used to count the read numbers mapped to each gene. Then, the FPKM (the expected number of fragments per kilobase of transcript sequence per million base pairs sequenced) of each gene was calculated based on the length of the gene and read counts mapped to the gene. FPKM considers the effect of sequencing depth and gene length on the read count at the same time and is currently the most commonly used method for estimating gene expression levels (35). Differential expression analysis of two conditions or groups (two biological replicates per condition) was performed using the DESeq R package (1.18.0). DESeq provides statistical routines for determining differential expression in digital gene expression data using a model based on the negative binomial distribution. The resulting P values were adjusted using Benjamini and Hochberg’s approach for controlling the false discovery rate (36). Genes with an adjusted P value of ≤0.05 found by DESeq were assigned as being differentially expressed.

Accession number.

The transcriptomic data have been uploaded to the SRA database with accession no. PRJNA542766.