P.B. (2015) Mesorhizobium waimense sp. nov. isolated from Sophora longicarinata root nodules and Mesorhizobium cantuariense sp. nov. isolated from Sophora microphylla root

31 32 In total 14 strains of gram-negative, rod-shaped bacteria were isolated from Sophora 33 longicarinata and Sophora microphylla root nodules and authenticated as rhizobia on these 34 hosts. Based on the 16S rRNA gene phylogeny, they were shown to belong to the genus 35 Mesorhizobium, and the S. longicarinata strains were most closely related to Mesorhizobium 36 amorphae ACCC 19665 (99.8 – 99.9%), Mesorhizobium huakuii IAM 14158 (99.8 – 37 99.9%), Mesorhizobium loti USDA 3471 (99.5 – 99.9%) and Mesorhizobium septentrionale 38 SDW 014 (99.6 – 99.8%), whilst the S. microphylla strains were most closely related to 39 Mesorhizobium ciceri UPM-Ca7 (99.8 – 99.9%), Mesorhizobium qingshengii CCBAU 40 33460 (99.7%) and Mesorhizobium shangrilense CCBAU 65327 (99.6%). Additionally, 41 these strains formed two distinct groups in phylogenetic trees of the housekeeping 42 genes glnII, recA and rpoB. Chemotaxonomic data, including fatty acid profiles supported the 43 assignment of our strains to the genus Mesorhizobium and allowed differentiation from the 44 closest neighbours. Results of DNA-DNA hybridisations, MALDI-TOF MS analysis, ERIC45 PCR, physiological and biochemical tests allowed genotypic and phenotypic differentiation of 46 our strains from their nearest neighbouring species. Therefore, the S. longicarinata and S. 47 microphylla strains represent two novel species for which the names Mesorhizobium 48 waimense sp. nov. (ICMP 19557 = LMG 28228 = HAMBI 3608) and Mesorhizobium 49 cantuariense sp. nov. (ICMP 19515 = LMG 28225 = HAMBI 3604), are proposed 50 respectively. 51 52 New Zealand (NZ) Sophora species (tribe Sophoreae in the Fabaceae plant family) are 53 woody trees and shrubs that occur in a variety of habitats (Heenan et al., 2001; Heenan et al., 54 2004). The most recent taxonomic revision of New Zealand Sophora recognized eight 55 endemic species: S. chathamica, S. fulvida, S. godleyi, S. longicarinata, S. microphylla, S. 56 molloyi, S. prostrata and S. tetraptera (Heenan et al., 2001). S. longicarinata is an upright 57 small tree with multiple trunks and main branches or densely branched shrub, characterized 58 by numerous and small leaflets that are distant from each other, uniform in size, dark green, 59 glabrous or with a few appressed hairs, and with distinct petiolules (Heenan et al., 2001). This 60 species is geographically restricted to the northern South Island (Nelson and Marlborough) 61 and is a basicole, predominantly occurring on eroding and unstable bluffs, rock outcrops, and 62 hill slopes derived from marble and limestone (Heenan et al., 2001). Sophora microphylla, 63 also known as small-leaved kowhai (tribe Sophoreae), contains trees up to 25 m high with 64 distant leaflets, a moderate number of appressed leaf hairs, and a distinct divaricating and/or 65 strongly flexuose juvenile phase (Heenan et al., 2001; Heenan et al., 2004). S. microphylla 66 occurs throughout the North and South Island, and is predominantly an inland species. It most 67 commonly grows on alluvial river terraces, flood plains, lake margins, and on hill slopes 68 among loose and rubbly rock (Heenan et al., 2001). 69 70 Previous studies investigating the nitrogen fixing symbionts of Sophora species have revealed 71 the presence of Mesorhizobium root nodule endosymbionts (Weir et al., 2004). However, only 72 a limited number of bacterial isolates were investigated. As part of a continuing study on 73 native New Zealand legumes, and their associated rhizobia, forty-eight strains were isolated 74 from surface sterilized root nodules of Sophora species sampled in natural ecosystems. 75 Sequence analysis showed that all isolates belonged to the genus Mesorhizobium and that they 76 grouped in seven different clusters (Tan et al., 2015). In the present study eight strains 77 originating from S. longicarinata root nodules collected from plants growing on limestone 78 alluvium at Waima River, Marlborough and six strains originating from S. microphylla root 79 nodules collected in Canterbury (Tan et al., 2015), were selected for further investigation 80 using a polyphasic approach. Strain ICMP 19557 and ICMP 19515 have been deposited in 81 the BCCM/LMG bacteria collection (http://www.belspo.be/bccm) and the HAMBI Culture 82 Collection, University of Helsinki, Finland (http://www.helsinki.fi/hambi/). All strains were 83 subcultured on Yeast Manitol Agar (YMA) medium (Vincent, 1970) at 28 °C unless 84 otherwise indicated. For PCR, genomic DNA of all isolates was prepared using the standard 85 Qiagen-Gentra PUREGENE DNA Purification Kit as described previously (Tan et al., 2015). 86 87 The ERIC-PCR fingerprints were obtained as described previously (Versalovic et al., 1994) 88 and analysed using the Phoretix 1D Pro v12.2 software package (Phoretix Ltd, UK). The 89 similarity among the digitised profiles was calculated using the Dice coefficient (Dice, 1945) 90 and an unweighted pair group using arithmetic averages (UPGMA) dendrogram was derived 91 from the similarity matrix. The Dice coefficient is used as a general measure of similarity (if 92 two lanes are identical, Distance (D) = 0 and if two lanes are totally different, Distance (D) 93 =1) but gives more weight to matching bands. Figure S1a and b show the ERIC-PCR 94 fingerprints of the S. longicarinata and S. microphylla Mesorhizobium isolates, respectively. 95 The DNA fingerprints suggest that all S. longicarinata and S. microphylla strains representing 96 two novel species form two separate cluster that could be distinguished from their closest 97 neighbours (Fig. S1a and b). Matrix-Assisted Laser Desorption/Ionization Time-of-Flight 98 mass spectrometry (MALDI-TOF MS) was performed as described previously (Wieme et al., 99 2012). All conditions were exactly as previously described except that YMA growth medium 100 was used to culture the strains prior to protein extraction (Wieme et al., 2012). The MALDI101 TOF MS profiles indicate that the isolates represent different strains that can be distinguished 102 from the closest neighbours (Fig. S2a and b). 103 104 Nearly full-length amplicons for the 16S rRNA gene were obtained for all strains using the 105 primers and conditions described previously by Tan et al. (2015). The resulting 16S rRNA 106 gene sequences were aligned using the MEGA 5 software package and phylogenetic trees 107 were constructed with the Maximum Likelihood (ML) method and Neighbor Joining (NJ) 108 method / Kimura 2 parameter model with G substitutions (Tamura et al., 2011). Bootstrap 109 analysis with 500 replicate data sets was performed to assess the support of the clusters. The 110 overall topologies of the phylogenetic trees obtained with the ML and NJ methods were 111 similar (data not shown). Our strains formed two novel branches within the Mesorhizobium 112 genus (Fig. 1), and group 1 containing strain ICMP 19557 shared sequence similarities 113 of 99.8 – 99.9% with Mesorhizobium amorphae ACCC 19665, 99.8 – 99.9% with 114 Mesorhizobium huakuii IAM 14158, 99.5 – 99.9% with Mesorhizobium loti USDA 3471 115 and 99.6 – 99.8% with Mesorhizobium septentrionale SDW 014, and group 2 containing 116 strain ICMP 19515 shared sequence similarities of 99.8 – 99.9% with Mesorhizobium ciceri 117 UPM-Ca7, 99.7% with Mesorhizobium qingshengii CCBAU 33460, 99.6% with 118 Mesorhizobium shangrilense CCBAU 65327, as determined with the EzTaxon-e server 119 (http://eztaxon-e.ezbiocloud.net/, Kim et al., 2012). GlnII [336 bp], recA [381 bp] and rpoB 120 [840 bp] gene sequence analysis was based on the method described by Tan et al. (2015) and 121 the sequences are deposited in NCBI (Accession numbers in Table S1). The gene sequences 122 were aligned using the MEGA 5 software package (Tamura et al., 2011) and phylogenetic 123 trees were constructed using the ML method, with the Tamura-3-parameter model and G 124 substitutions. Bootstrap analysis with 500 replicates was performed to assess the support of 125 the clusters. Congruence between the different gene sequences was investigated using the 126 partition homogeneity tests (Farris et al., 1994) performed with PAUP software v. 4.0b10 127 (Swofford, 1991). Congruence (p > 0.01) was found between all investigated genes and 128 subsequent concatenation using the software SeaView v. 4.4.3 was performed (Gouy et al., 129 2010). The phylogenetic tree based on the concatenated glnII, recA and rpoB gene sequences 130 of our strains (Fig. 2) revealed two monophyletic clusters supported by high bootstrap values 131 (99.9% – 100%). Levels of gene sequence similarity between group 1 containing strain ICMP 132 19557 and the closest neighbour M. septentrionale SDW 014 was 94.7% for glnII, 94.5 – 133 96.8% for recA and 97.6% for rpoB; and with M. amorphae ACCC 19665, 94.8% for glnII, 134 95.4 – 95.7% for recA and 97.3 97.4% for rpoB. Levels of gene sequence similarity 135 between group 2 containing strain ICMP 19515 and the closest neighbour M. ciceri UPM136 Ca7 was 95.9 – 96.0% for glnII, 96.6 – 96.8% for recA and 97.2% for rpoB; and with M. loti 137 LMG 6125, 89.8 – 89.9% for glnII, 96.6 – 96.7% for recA and 96.5% for rpoB. 138 139 Phenotypic analysis was performed with cells grown on YMA medium at 28 °C unless 140 otherwise indicated. Cells were Gram stained (Vincent, 1970). Cell morphology and motility 141 were observed by phase contrast microscopy. Oxidase activity was detected by immersion of 142 cells in 1% N,N,N′,N′-tetramethyl-p-phenylenediamine solution and catalase activity was 143 determined by flooding a colony with 10% H2O2 and checking for the presence of bubbles. 144 Biochemical tests were performed by inoculating API 20NE and API 20E strips (BioMérieux) 145 and Biolog GENIII MicroPlatesTM (Biolog Inc, CA, USA), according to the manufacturer’s 146 instructions. GENIII MicroPlatesTM were read using the MicroStationTM ID System reader 147 (Biolog Inc, CA, USA). Growth was tested at 28 °C in Yeast Mannitol broth with 1% to 8% 148 NaCl and with pH4 pH9, buffered with acetic acid/sodium acetate (pH4 5), citric 149 acid/Na2HPO4 (pH6 7), NaH2PO4/Na2HPO4 (pH8) or Tris/HCl (pH9). Growth on YMA 150 medium was tested at 4, 7, 15, 20, 25, 28, 30 and 37 °C. Colonies were visible after 48 h 151 growth at 15 – 30 °C on YMA medium. The results of the phenotypic and biochemical tests 152 are given in Table 1 and supplementary Table S2a and b. Most notably for group 1 containing 153 strain ICMP 19557, Biolog GENIII MicroPlatesTM carbon source utilisation positive 154 reactions were recorded for N-Acetyl-Beta-D-Mannosamine, 3-Methyl glucose, citric acid 155 and methyl pyruvate; negative reactions for L-serine and weak positive reactions for pectine. 156 For group 2 containing strain ICMP 19515 positive reactions were recorded for D-saccharic 157 acid and propionic acid; weak positive reactions for citric acid, D-lactic acid methyl ester, 158 methyl pyruvate, alpha-D-lactose, glucuronamide and pectine; and negative reactions for 159 stachyose, N-acetyl neuronimic acid and formic acid. Additional antibiotic susceptibility tests 160 were performed on YMA medium using the antibiotic Sensi-disc dispenser system (Oxoid) 161 with bio-discs (Oxoid) containing ampicillin (10 μg), chloramphenicol (30 μg), erythromycin 162 (15 μg), gentamycin (10 μg), kanamycin (30 μg), and streptomycin (25 μg). All strains were 163 grown on YMA for 72 h prior to testing. The plates were incubated at 28°C and read between 164 two and seven days. All strains from group 1 were resistant to erythromycin, and sensitive to 165 ampicillin, chloramphenicol, gentamycin and streptomycin; all strains from group 2 were 166 resistant to chloramphenicol, erythromycin and kanamycin, and sensitive to gentamycin and 167 streptomycin. 168 169 The whole-cell fatty acid composition was analysed and the fatty acid methyl esters were 170 extracted from cells grown on YMA medium according to the MIDI protocol 171 (http://www.microbialid.com/PDF/TechNote_101.pdf). All characteristics such as 172 temperature and physiological age (overlap area of the second and third quadrant from a 173 quadrant streak) were as in the MIDI protocol. The profiles were generated using an Agilent 174 Technologies 6890N gas chromatograph (Santa Clara, CA USA), identified and clustered 175 using the Microbial Identification System software and MIDI TSBA database version 5.0. 176 Fatty acid profiles are listed in Table 2. The most abundant fatty acids for our strains were 177 C18:1 ω7c (58.9 – 51.4%), C16:0 (26.8 – 15.7%) and C19:0 CYCLO ω8c (15.3 – 4.4%) for 178 group 1 and C18:1 ω7c (37.1 – 44.2%), C16:0 (16.6 – 21.4%) and C19:0 CYCLO ω8c (14.2 – 179 18.9%) for group 2. All strains lacked C20:3 ω6,9,12cis which is characteristic for 180 Mesorhizobium species (Tighe et al., 2000). Additionally, there were noticeable differences 181 between the fatty acid profiles of our strains and most closely related Mesorhizobium type 182 strains (Table 2). 183