Phylogenetic Analyses of pheS, dnaA and atpA Genes for Identification of Weissella confusa and Weissella cibaria Isolated from a South African Sugarcane Processing Factory.

A previous study reported on the isolation of 430 polysaccharide (gum)-producing bacteria from a South African sugarcane processing factory and the identification of isolates by comparative 16S rRNA gene sequencing. A large number of isolates (202) belonged to the genus Weissella and clustered with reference strains of Weissella cibaria and Weissella confusa. In this study, we identified 147 strains as W. cibaria and 55 as W. confusa based on phylogenetic analyses of pheS and dnaA gene sequences of representative isolates. We also included atpA gene sequence analysis of Weissella isolates as potential future phylogenetic marker to differentiate amongst strains of W. cibaria and W. confusa.

High levels of polysaccharides in sugarcane have a marked effect on the recovery of crystal sugar and cause severe losses in the sugar industry [[1]]. These polysaccharides, of which dextran has generally been considered the main problem in the South African sugarcane industry [[2]], arise from microbial degradation, or biodeterioration of the cane after harvesting and during the sugar production process. Dextran, consisting of d-glucose units linked with α-1,6 glycosidic bonds, and α-1,2, α-1,3 or α-1,4 bonds [[4]], is produced from sucrose by Leuconostoc, Weissella, Lactobacillus, Streptococcus and Pediococcus spp. with dextransucrase activity [[5]]. Dextran increases the viscosity of sugar processing streams, leading to higher sucrose inversion losses due to extended boiling times caused by evaporation difficulties and reduction in crystallisation rates [[7]].

There is currently no reliable, rapid, easy and inexpensive method to measure cane deterioration [[9]]. Microorganisms present on sugarcane do not all survive the production process and do not always contribute to the production of exopolysaccharides in the factory [[10]]. A limited number of studies have examined the microbial diversity in different unit operations of sugarcane processing factories [[11]–[13]]. These studies were all constrained by the absence of microbial identification methods with high discriminatory power. Modern approaches to microbial taxonomy provide for a much more accurate and reliable identification of spoilage microorganisms and offer the opportunity to revisit the findings of other researchers.

We have isolated 430 exopolysaccharide (gum)-producing bacteria from seven locations in a South African sugarcane processing factory [[3]]. Of these, 110 isolates were obtained from samples taken during spring (September 2103; temperatures ranged from 10 to 27 °C, with a daily mean relative humidity of 70%) when low concentrations of dextran (< 70 mg/kg) were observed in the produced sugar, and 320 isolates from samples during summer (November 2013; temperatures ranged from 13 to 30 °C, with a daily mean relative humidity of 78%) when high concentrations of dextran (> 500 mg/kg) were reported in raw sugar. Phylogenetic analysis of the partial 16S rRNA gene sequences differentiated the gum-producing bacteria into four genera and nine species [[3]]. A large number of isolates (202) belonged to the genus Weissella and representative strains clustered with reference strains of Weissella cibaria and Weissella confusa. A significant finding of this study was that Leuconostoc mesenteroides was not the only spoilage bacterium isolated from sugarcane and sugarcane processing streams. Historically, L. mesenteroides had been implicated as the main causative agent of biodegraded sugarcane [[14]]. Nel [[15]] challenged the perception that L. mesenteroides is the sole cause of biodegraded sugarcane and advocated the concept of microbial diversity profiling of spoilage bacteria in sugarcane processing to develop strategies to prevent biodeterioration of sugarcane, and/or mitigate the detrimental effects of microbial metabolic products (such as dextran) on sugarcane processing. Bacterial spoilage of sugarcane and dextran production during sugarcane processing remains a problem. Current methods used to prevent microbially-mediated sucrose loss and solutions to reduce the effects of dextran on sugarcane processing are inadequate and do not address the cause of the problem. The unanticipated presence of Weissella spp. on prepared cane and in the factory emphasises the importance of accurately identifying spoilage bacteria in sugarcane processing at the various unit operations for effective microbial control and development of cane deterioration indicators.

Although 16S rRNA gene sequence analysis is still considered important for bacterial identification [[16]], the main disadvantage of the method is its often-insufficient resolution at species level, especially between strains of closely related species. W. confusa and W. cibaria share 99.2% 16S rRNA sequence similarity [[17]]. Hong and Farrance [[18]] showed that the overall performance of the first 500 bp sequence of the 16S rRNA gene, compared to the entire 1500 bp sequence for bacterial identification is very high (> 90%). The authors suggested that, for bacterial identification, the generation of full-length sequence data for the 16S rRNA gene is inefficient and impractical, and that a higher phylogenetic resolution can be obtained by sequence analysis of the first 500 bp of the 16S rRNA gene in combination with additional phylogenetic analyses of housekeeping or other protein-coding genes.

Gene sequence analyses of the phenylalanyl t-RNA synthase alpha subunit (pheS) and RNA polymerase alpha subunit (rpoA) have been used to differentiate amongst closely related lactic acid bacteria of the genera Lactobacillus [[19]–[22]] and Enterococcus [[23]]. Gene sequence analysis of the alpha subunit of ATP synthase (atpA) were used to differentiate amongst species of the genera Leuconostoc [[24]] and Pediococcus [[25]]. However, atpA as molecular marker has not previously been used to differentiate Weissella species. Chelo et al. [[26]] assessed the congruence of evolutionary relationships within the Leuconostoc-Oenococcus-Weissella clade by comparison of 16S rRNA gene, dnaA (encoding chromosomal replication initiation protein), gyrB (encoding DNA gyrase B subunit), rpoC (encoding the beta subunit of the DNA-dependent RNA polymerase) and dnaK (encoding the 70 kDa heat-shock protein) sequence analyses.

The identification of the 202 Weissella strains isolated from a South African sugarcane processing factory involved two steps as suggested by Hong and Farrance [[18]]. Previously, partial 16S rRNA gene sequencing was used to determine the genera of the unknown strains [[3]]. In this study, the 16S rRNA sequencing data are compared to the phylogenetic analyses of the housekeeping genes pheS and dnaA for the identification of the Weissella isolates. We also included the atpA sequence analysis of the representative Weissella isolates as potential future phylogenetic marker to differentiate between strains of W. cibaria and W. confusa.

Samples of crushed sugarcane, and samples from the diffuser sump, juice screen (Dutch State Mines; DSM screen), mixed juice tank (MJ tank), filtrate, mud trough and syrup tank from a South African sugarcane processing factory were collected for the isolation of gum-(polysaccharide) producing bacteria [[3]]. Crushed sugarcane samples (10 g each) were added to 100 ml phosphate buffered saline (PBS, [[27]])and incubated on a rotary shaker (30 °C, 150 rpm) for 1 h. Liquid samples collected from each of the sampling points and PBS-cane suspensions were serially diluted in PBS and streaked onto modified dextransucrase-inducing agar with the following composition: sucrose 100 g/l, peptone 20 g/l, KH2PO4 20 g/l, agar 15 g/l and R-salts (4% MgSO4·7H2O, 4% NaCl, 0.2% FeSO4·7H2O and 0.2% MnSO4·H2O) 5 ml [[28]]. Plates were incubated at 30 °C for 14 to 18 h. Colonies with a glistening and slimy appearance were selected and streaked to purity on modified dextransucrase-inducing agar. From these plates a single colony was inoculated into 5 ml MRS broth (Biolab, Merck South Africa) and the cultures incubated on a shaking incubator (150 rpm) for 14 to 18 h at 30 °C. Cells were harvested (16 000 × g, 25 °C, 2 min), re-suspended in sterile glycerol (200 μl; 50%, v/v) and stored at − 70 °C.

Ten μl aliquots of stock culture was inoculated into 5 ml sterile MRS broth (Biolab, Merck, Modderfontein, South Africa) and incubated for 16 h at 30 °C on a rotary shaker (150 rpm). Cells were harvested (16 000 × g, 25 °C, 2 min) and genomic DNA extracted using the GeneJET Genomic DNA Purification kit (Thermo Scientific, Inqaba Biotechnical Industries, Hatfield Pretoria, South Africa) according to the manufacturer's instructions. Purified DNA was suspended in 50 μl elution buffer and used as template in amplification reactions.

Genomic DNA was used as template to amplify partial sequences of the 16S rDNA, pheS, dnaA and atpA genes, respectively, for all isolates using the primers listed in Table 1. The reactions were carried out in 50 μl reaction mixtures containing 10 pmol of each primer, 200 µM of each deoxynucleoside triphosphate (Thermo Scientific), 10 µl of 5 × One Taq Standard Reaction buffer, 1.25 U One Taq Hot Start DNA polymerase (Thermo Scientific) and 100 ng template genomic DNA. PCR reactions were performed in a programmable thermal cycler (MultiGene OptiMax, Labnet International, Whitehead Scientific, Cape Town, South Africa) with an initial denaturation step (94 °C, 30 s), followed by 30 cycles of denaturation (94 °C, 30 s), primer annealing and elongation (see Table 1). Cycling was completed by a final elongation step (68 °C, 10 min), followed by cooling to 4 °C. The resultant amplicons were purified using the DNA Clean and Concentrator™-25 kit (Zymo Research, Inqaba Biotechnical Industries, Hatfield Pretoria, South Africa) according to the manufacturer's instructions.

Primer sequences and PCR conditions for the partial amplification of the 16S rRNA gene and the housekeeping genes pheS, dnaA and atpA

*Sequencing primer only

16S rRNA, pheS, dnaA and atpA gene sequencing were performed by the South African Sugarcane Research Institute (Mount Edgecombe, South Africa) using BigDye Cycle Sequencing chemistry (Applied Biosystems, Johannesburg, South Africa), according to the manufacturer's instructions. Sequence similarity searches were performed using the Basic Local Alignment Search Tool (BLAST) algorithm [[29]]. Reference 16S rRNA, pheS, atpA and dnaA gene sequences were retrieved from the National Centre for Biotechnology Information (NCBI) (https://www.ncbi.nlm.nih.gov/). Where gene sequences were identical for a group of isolates from the respective sampling times and locations, only one isolate was chosen as representative of the group. GenBank accession numbers for 16S rRNA, pheS, dnaA and atpA gene sequences of representative strains for each sampling time and location, as determined in this study, are listed in Table 2. For phylogenetic inference, five different alignments were created; one corresponding to the 16S rRNA gene sequences of Weissella type strains as obtained from GenBank and four alignments corresponding to the single locus analysis of the 16S rRNA, pheS, dnaA and atpA genes of strains representing different groups of isolates. The 16S rRNA gene sequences were aligned using the SILVA Incremental Aligner software version 1.2.11 [[30]], and the pheS, dnaA and aptA gene sequences were aligned with ClustalW [[31]] as implemented in the BioEdit Sequence Alignment Editor programme [[32]]. A data matrix for each alignment was created for the representative sequences of strains at each sampling location and sampling time. Phylogenetic analyses were conducted using the Molecular Evolutionary Genetics Analysis (MEGA) version 7.0 software [[33]]. The evolutionary histories were inferred using the Neighbour-Joining method [[34]] with the Kimura 2-parameter model [[35]] for the 16S rRNA gene sequence analyses, and the Tamura 3-parameter model [[36]] for the respective housekeeping genes. The strengths of the internal branches of the resultant trees were statistically evaluated by bootstrap analysis [[37]] with 1000 bootstrap replications.

GenBank accession numbers of the sequences as determined in this study for representative Weissella strains for each sampling location

A phylogenetic tree constructed from near full-length 16S rRNA gene sequences (1286 bp) of Weissella type strains is shown in Fig. 1. The phylogenetic relationship of strains representing different groups of isolates, based on partial 16S rRNA gene sequences (752 bp), is shown in Fig. 2. The tree topology of the partial 16S rRNA gene sequences of representative strains (Fig. 2) is in agreement with that obtained for the near full-length 16S rRNA gene sequences of Weissella reference strains (Fig. 1) and as reported by Lee et al. [[38]]. A total of 202 strains, representative of the different isolates, grouped into two distinct clusters (Fig. 2), supported by high bootstrap values (97 and 99%, respectively). The two clusters represent W. cibaria (147 isolates) and W. confusa (55 isolates). 16S rRNA gene sequence similarities from the Weissella type strains (Fig. 1) ranged from 90.2% (Weissella halotolerans/Weissella beninensis) to 99.2% (W. cibaria/W. confusa). Of all the locations sampled, Weissella spp. were isolated from the prepared (shredded) sugarcane, diffuser sump, filtrate and syrup tank.

Graph: Fig. 1Phylogenetic analysis of 16S rRNA gene sequences of Weissella type strains, using Neighbour-Joining method and Kimura's 2-parameter model [ [35] ]. Each dataset had 1285 bp. Bootstrap values (> 50%, 1000 replications) are shown. The bar indicates % estimated substitution per nucleotide position. Enterococcus faecalis DSM 20478 T was used as the outgroup

Graph: Fig. 2Phylogenetic analysis of partial 16S rRNA gene sequences of representative Weissella strains isolated from a South African sugarcane processing factory at times when low (filled circle) and high (filled square) dextran concentrations were observed in the produced sugar. The number of isolates from each sampling time and location is indicated in brackets. The tree was constructed using the Neighbour-Joining method and Kimura's 2-parameter model [ [35] ]. Each dataset had 752 bp. Bootstrap values (> 50%, 1000 replications) are shown. The bar indicates % estimated substitution per nucleotide position. Enterococcus faecalis DSM 20478 T was used as the outgroup

De Bruyne et al. [[39]] reported on the high discriminatory power of pheS gene sequences in comparison to 16S rRNA gene sequences amongst species of Weissella, Leuconostoc and Oenococcus. A phylogenetic tree constructed from partial pheS gene sequences (352 bp) of representative Weissella isolates is shown in Fig. 3. The discriminatory power of pheS as phylogenetic marker for W. cibaria and W. confusa is confirmed by the results obtained, with high bootstrap values (100% in both cases). Weissella isolates from the sugarcane processing factory clustered with W. cibaria (six strains representing 147 isolates) and W. confusa (4 strains representing 55 isolates). The pheS sequence similarities of the type strains ranged from 70.7% (W. cibaria/W. beninensis) to 89.6% (Weisella fabaria/Weissella ghanensis), with W. cibaria/W. confusa sharing an 88.5% pheS sequence similarity.

Graph: Fig. 3Phylogenetic analysis of partial pheS gene sequences of representative Weissella strains isolated from a South African sugarcane processing factory at times when low (filled circle) and high (filled square) dextran concentrations were observed in the produced sugar. The number of isolates from each sampling time and location is shown in brackets. The tree was constructed using the Neighbour-Joining method using Tamura's 3-parameter model [ [36] ]. Each dataset had 352 bp. Bootstrap values (> 50%, 1000 replications) are shown. The bar indicates % estimated substitution per nucleotide position. Enterococcus faecalis LMG 7937 T was used as the outgroup

Contrary to the widespread use of pheS as phylogenetic marker to differentiate Weissella species, dnaA gene sequence analysis has only been used by Chelo et al. [[26]] to evaluate six Weissella type strains as part of a study clarifying the intra- and intergeneric phylogenetic relationships of the Leuconostoc-Oenococcus-Weissella clade. Despite its limited use, the phylogenetic tree constructed from partial dnaA gene sequences of representative strains of Weissella (574 bp, Fig. 4) displayed high bootstrap support (100 and 99%, respectively) for the differentiation of W. cibaria (six strains representing 147 isolates) and W. confusa (four strains representing 55 isolates). The dnaA sequence similarities of the type strains ranged from 59.6% (Weissella kandleri/W. halotolerans) to 100% (Weissella thailandensis/Weissella jogaejeotgali), with 81.4% dnaA sequence similarity between W. cibaria and W. confusa.

Graph: Fig. 4Phylogenetic analysis of partial dnaA gene sequences of representative Weissella strains isolated from a South African sugarcane processing factory at times when low (filled circle) and high (filled square) dextran concentrations were observed in the produced sugar. The number of isolates from each sampling time and location is shown in brackets. The tree was constructed using the Neighbour-Joining method using Tamura's 3-parameter model [ [36] ]. Each dataset had 574 bp. Bootstrap values (> 50%, 1000 replications) are shown. The bar indicates % estimated substitution per nucleotide position. Enterococcus faecalis V583 (29374661), chromosomal location number NC_004668.1:59-1402, was used as the outgroup

The atpA gene has not been previously used as phylogenetic marker for the identification of Weissella species. De Bruyne et al. [[24], [40]] used atpA gene sequence analysis for the identification of Pediococcus and Leuconostoc species. Currently, the only available nucleotide sequence for the atpA gene for Weissella species is that of the W. viridescens type strain. The phylogenetic tree inferred from partial atpA gene sequences (820 bp) from the isolated Weissella strains (Fig. 5) showed the clustering of six strains representing 147 isolates (Cluster 1) and four strains representing 55 isolates (Cluster 2) with high bootstrap support.

Graph: Fig. 5Phylogenetic analysis of partial atpA gene sequences of representative Weissella strains isolated from a South African sugarcane processing factory at times when low (filled circle) and high (filled square) dextran concentrations were observed in the produced sugar. The number of isolates from each sampling time and location is shown in brackets. The tree was constructed using the Neighbour-Joining method using Tamura's 3-parameter model [ [36] ]. Each dataset had 820 bp. Bootstrap values (> 50%, 1000 replications) are shown. The bar indicates % estimated substitution per nucleotide position. The only Weissella atpA gene sequence available was from Weissella viridescens LMG 350 T. Enterococcus faecalis LMG 7937 T was used as the outgroup

This study identified W. cibaria as major contributor of gum-producing bacteria isolated from the prepared (shredded) sugarcane in summer, when high dextran concentrations were reported in the produced sugar. Hot and humid conditions favour rapid accumulation of dextran in harvested cane which is exacerbated by delays in delivering the cane to the factory [[2]]. The dominance of Weissella spp. on prepared cane is significant because these bacteria are not usually associated with deteriorated sugarcane and sugarcane processing factories. Weissella bacteria have been isolated from a diversity of ecological niches, including soil, plants, a variety of fermented foods, as well as from humans and animals [[41]]. The risk of contamination of Weissella bacteria in sugarcane processing factories is therefore quite high. Weissella spp. have the ability to synthesise a variety of polysaccharides and oligosaccharides, which includes high molecular weight, low-branched dextran, as well as levan and inulin, in addition to gluco- and fructo-oligosaccharides and cell-associated ropy polymers [[42]]. These poly- and oligosaccharides may have a severe impact on the quality and quantity of produced sugar, and accurate identification of the spoilage bacteria in sugarcane processing is key for effective microbial control and development of cane deterioration indicators.

Previous studies have shown that the housekeeping genes pheS, dnaA and atpA have high discriminatory power in differentiating various lactic acid bacteria. In this study, single locus analyses of pheS and dnaA gene sequences revealed the clustering of representative Weissella isolates with the type strains of W. cibaria (147 strains) and W. confusa (55 strains) (Fig. 3 and 4). Phylogenetic analysis of the atpA gene showed the same grouping of representative Weissella isolates into two clusters. Previous studies which examined the microbial diversity in sugarcane processing factories [[11]–[13]] were constrained by the absence of microbial identification methods with high discriminatory power. We have shown the potential of phylogenetic analyses of housekeeping genes as alternative identification method for sugarcane processing spoilage microbes.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.