Bacterial Endophytes: Exploration of Methods and Analysis of Community Variation

Shu Yi Shen

A thesis submitted in conformity with the requirements for the degree of Master of Science Ecology and Evolutionary Biology University of Toronto

Shu Yi Shen

Master of Science

Ecology and Evolutionary Biology University of Toronto

2013 Abstract

Bacterial endophytes, bacteria residing within plants, play an important role in the growth and development of plants and their ability to thrive under adverse conditions. The endophytes of

Acer negundo, Ulmus pumila and Ulmus parvifolia trees sampled from a hydrocarbon- contaminated site were analyzed for variation between seasons and plant species. Branches from the same trees over a span of 3 seasons were collected and analyzed via culture dependent and culture independent methods. Numerous culture independent approaches were tested, culminating in the development of a new method for the amplification of endophytic bacterial ribosomal DNA that excludes plastid DNA. Community analyses using this new method in combination with T-RFLP showed significant differences between the endophytic communities of different plants species and of the same species growing in different seasons. The proposed technique can be used for the future study of endophytic communities of plants.

ii Acknowledgments

I would like to first and foremost thank my supervisor, Dr. Roberta Fulthorpe, for allowing me to carry out this project under her supervision. I am very thankful for your constant motivation, support and guidance throughout this project and through my moments of panic and irrationality.

I would like to all the past and present members of the Fulthorpe Lab who have helped me along the way: Tony Qian for all your help with programming and being a sounding board for ideas, providing an analytical view to everything in life; Nicole Ricker for your great life advice, knowledge of almost everything there is to know about anything and the calmness you bring to the lab; Rosemary for the fun you brought into the lab and to Rhea Lumactud for all the help you have provided me with everything from going to the field and sampling, helping me process my samples, keeping me company in the late lab nights and helping me understand stats. I will miss the talks and lab karaoke that got me through the long days and nights in the lab.

Specials thanks goes out to other people who have helped with various aspects of this project: Richard Dickinson for helping me identify all the trees species, you still amaze me with your knowledge and ability to identify plants in general and Thanushiga Theivendram for helping me streak out bacteria, I know it wasn't easy but thanks for doing it.

Lastly I would like to thank my family for being there as a constant source of support and encouragement through the ups and downs of this project and my life.

iii

Table of Contents

Acknowledgments ...... iii

Table of Contents ...... iv

List of Tables ...... vii

List of Figures...... ix

List of Appendices...... xx

Chapter 1 An Introduction to Bacterial Endophytes ...... 1

1.1 General Introduction...... 1

1.2 Current Culture Independent Methods for Studying Endophytes ...... 4

Chapter 2 Analysis of Molecular Methods for Post Extraction Exclusion of Plastid Amplicons and Analysis of Endophytic Community...... 9

2.1 Methods...... 9 2.1.1 Database Analysis of Primers ...... 9 2.1.2 Surface sterilization of Acer negundo and Ulmus spp. Branches ...... 10 2.1.3 Isolation of DNA from Acer negundo and Ulmus spp. Branches...... 11 2.1.4 Isolation of Plastid 16S rRNA Amplicons...... 11 2.1.5 Primer Testing on Extracted Plastid Amplicons ...... 12

2.2 Results ...... 13 2.2.1 Database Analysis of Primers ...... 13 2.2.2 Isolation of Plastid Amplicons ...... 15 2.2.3 Analysis of Primer 783R...... 15 2.2.3 Primer Testing on Extracted Plastid Amplicons ...... 16 2.2.4 Preliminary Testing of Plastid Excluding Primers on Extracted Plant Samples ...... 17

2.3 Discussion...... 18

Chapter 3 Physical Separation of Plastid Cells via Differential Centrifugation and Density Gradient Centrifugation...... 21

3.1 Introduction...... 21

3.2 Method ...... 22 iv

3.2.1 Preparation of Plant Macerate ...... 22 3.2.2 Differential Centrifugation of the Macerate ...... 23 3.2.3 Sucrose Density Gradient Centrifugation ...... 23

3.3 Results ...... 25 3.3.1 Differential Centrifugation of the Macerate ...... 25 3.3.2 Sucrose Density Gradient Centrifugation ...... 26

3.4 Discussion...... 29

Chapter 4 Subtractive Hybridization...... 32

4.1 Introduction...... 32

4.2 Methods...... 34 4.2.1 Designing Plastid Specific Probes...... 34 4.2.2 Binding of Magnetic Beads to Phosphorus Dendrimers ...... 34 4.2.3 Immobilization of Oligonucleotide Probes ...... 35 4.2.4 Subtractive Hybridization...... 36

4.3 Results ...... 39 4.3.1 Plastid Specific Probe ...... 39 4.3.2 Subtractive Hybridization...... 40

4.4 Discussion...... 42

Chapter 5 Enzymatic Digestion of Genomic DNA samples ...... 46

5.1 Introduction...... 46

5.2 Methods...... 47 5.2.1 Database Analysis of Restriction Enzymes...... 47 5.2.2 Enzymatic Digest of DNA and Amplification of Bacterial DNA ...... 47 5.2.3 Amplification using DGGE primers ...... 48

5.3 Results and Discussion...... 48 5.3.1 Database Analysis of Restriction Enzymes...... 48 5.3.2 Experimental Analysis of the Restriction Enzymes on Plant Genomic DNA ...... 49

Chapter 6 Analysis of Variation – Culture Dependent Method...... 52

6.1 Introduction...... 52

6.2 Methods...... 53 6.2.1 Sample Collection and Surface Sterilization...... 53 6.2.2 Culturable Endophyte Extraction, Isolation and Identification ...... 54 6.2.3 Community and Statistical Analysis ...... 55

6.3 Results ...... 57 6.3.1 Bacterial Densities, Richness and Diversity ...... 57 6.3.2 Identification of Bacterial Strains ...... 60 6.3.3 Comparison of Community Structures Between Samples...... 69

6.4 Discussion...... 81

Chapter 7 Analysis of Variation – Culture Independent Method...... 88

7.1 Introduction...... 88

7.2 Methods...... 88 7.2.1 DNA Extraction of Endophytic Community...... 88 7.2.2 Enzymatic Digestion, PCR Amplification and TRFLP...... 89 7.2.3 Data Analyses – T-RFLP and Statistical ...... 90

7.3 Results ...... 91 7.3.1 Preliminary Analyses - DGGE Results ...... 91 7.3.1 Analysis of Individual Terminal Restriction Fragments, Richness and Diversity ...... 93 7.3.2 Comparison of Community Structures Between Samples...... 97

7.4 Discussion...... 109

Chapter 8 Conclusions and Recommendations for Further Studies...... 114

Bibliography ...... 119

Appendix A – RDP Database Information ...... 131

Appendix B – Bacterial MspI Restriction Fragment Sizes...... 136

Appendix C – Cultured Bacterial Phyla Abundances...... 140

Appendix D – Hierarchical Clustering ...... 142 D.1 Culture Dependent Analyses – Hierarchical Clustering ...... 142 D.2 Culture Independent Analyses – Hierarchical Clustering...... 152

List of Tables

Table 2.1: Primers and their corresponding sequences tested in the RDP database.

Table 3.1: Specific gravity of various sucrose solutions made in 50 mM Tris-HCl (pH = 7.69), measured at room temperature.

Table 5.1: Summary of analysis of bacterial and land plant plastid sequences in the RDP database with restriction sites for enzymes PvuII and MscI.

Table 6.1: Tree branches collected for analysis.

Table 6.3: Cumulative list of identified bacterial endophytes and their corresponding bacterial phyla, isolated from Acer negundo, Ulmus parvifolia and Ulmus pumila branches from Winter, Summer and Autumn 2012.

Table 6.4: Permutation multivariate analysis of R2A data conducted using adonis after 999 permutations.

Table 6.5: Permutation multivariate analysis of TSA data conducted using adonis after 999 permutations.

Table 7.1: List of forward T-RFs exclusive to certain plant species with  indicating presence of the fragment in the samples from the plant species.

Table 7.2: List of reverse T-RFs exclusive to certain plant species with  indicating presence of the fragment in the samples from the plant species.

vii

Table 7.3: ANOVA analysis on the phylotype richness and the diversity index values of the samples based on the plant species and seasons for the forward and reverse T-RFs data sets.

Table 7.4: Permutation multivariate analysis of forward TRFLP data conducted using adonis after 999 permutations.

Table 7.5: Permutation multivariate analysis of reverse TRFLP data conducted using adonis after 999 permutations.

Table 8.1: Summary of significant differences found in the comparison of the endophytic community of Acer negundo, Ulmus parvifolia and Ulmus pumila based on plant species and seasons sampled using culture dependent and culture independent methods.

Table A.1: Bacterial hits with restriction sites for enzymes PvuII and MscI in the RDP Database (*hits as of August 3, 2012), using sequences of good quality and >1200bp.

Table A.2: List of some of the bacterial species or genera that have restriction sites corresponding to enzymes PvuII and MscI in the RDP database.

Table B.2: Database of cultured endophytic bacterial species from Acer negundo and Ulmus spp. and the corresponding forward terminal MspI restriction digested fragment from the amplicons amplified with primers 27F-1492R.

Table C.1: Percentage of bacterial phyla cultured in each season in R2A and TSA media.

Table C.2: Percentage of bacterial phyla cultured in each plant species in R2A and TSA media.

viii

List of Figures

Figure 1.1: Sample alignment of some endophytic bacteria 16S rRNA sequences, plant plastid rRNA sequences and plant mitochondrion rRNA sequences achieved through the program Geneious. Green arrows found below the sequences indicate the binding sites corresponding to primers 799F and 783R.

Figure 2.1: Graphical representation of the percentage of bacterial sequences and land plant plastid sequences targeted by each primer tested on the RDP database.

Figure 2.2: Agarose gel image of the resultant amplicons after amplification with full 16S rRNA primers, followed by digestion with PvuII enzyme, ran on a 1.5% gel. The samples used for PCR amplification were the plastid DNA (lanes 1-4) of: Acer negundo, Ulmus parvifolia, Ulmus pumila and Arabidopsis thaliana respectively and known bacterial DNA (lane 5). L = 1 Kb plus DNA ladder (Fermentas) and red arrow indicates 1500 bp.

Figure 2.3 – DGGE results of the PCR amplification done with individual versions of primer 783R tested on Acer negundo DNA. The samples are amplicons obtained using different primer sets: 1 - 341FGC and 783RB, 2 – 341FGC and 783RA, 3 –341FGC and 783R (equimolar concentrations of 783RA, 783RB and 783RC) and L – DGGE ladder.

Figure 2.4: Primer testing using a) primers 341FGC-783R and 341FGC-907R, and b) primers 799FGC-1197R on crudely extracted plastid amplicons from 1 – Acer negundo, 2 – Ulmus parvifolia and 3 – Ulmus pumila and L – DGGE ladder. Green letters represent the bands excised and sequenced with their identities as follows: letters A, B and C = Pantoea sp. 16S rRNA, letters D, E and F = plant plastid rRNA, and letters G and H = Pantoea sp. 16S rRNA.

Figure 2.5: Agarose gel image of the resultant amplicons after amplification using a) plastid exclusion primers, 341FGC-MOD783R and b) universal 16S DGGE primers 341FGC-907R on extracted plant genomic DNA from Acer negundo (lane 1-3), Ulmus parvifolia (lane 4-5) and Ulmus pumila (lane 6-7) collected in Autumn 2012, known bacterial DNA (lane 8), negative control (lane 9) and L – 1 Kb plus DNA Ladder (Fermentas).

Figure 3.1: DGGE results of the Acer negundo plant macerate subjected to differential centrifugation with DNA extracted from the pellets present after each centrifugation step amplified using primers a) 341FGC and 907R and b) plastid excluding primers 341FGC- MOD783R. The samples are as follows: lane 1 – original Acer negundo plant macerate, lanes 2- 8 – pellet after centrifugation at 200xg, 500xg, 800xg, 1000xg, 3000xg, 5000xg, respectively, 7 – left over supernatant and L – DGGE ladder.

Figure 3.2: Image showing a) original samples overlaid the sucrose cushion and b) samples after being subjected to centrifugation, originally carried out in 1.5 mL tubes. The tubes are labeled as follows: A - mixed bacteria sample obtained from media plates and B - plant macerate sample.

Figure 3.3: DGGE gel of plant macerate subjected to 27%-36% sucrose gradient, 30% sucrose cushion and 35% sucrose cushion, centrifuged at 4000 rpm for either 1 hour or 1 minute. The top layer, consisting of the top 8 mL of liquid, and the pellet were examined using primers a) 341F-GC-907R and b) using plastid excluding primers 341FGC-MOD783R. The samples were as follows: 1 - original plant macerate, while lanes 2-7 represent the plant macerated centrifuged on various sucrose gradient/cushions: 2 - 27%-36% sucrose gradient supernatant, 3 - 27%-36% sucrose gradient pellet, 4 - 30% sucrose cushion supernatant, 5 - 30% sucrose cushion pellet, 6 - 35% sucrose cushion supernatant, 7 - 35% sucrose cushion pellet, 8 - known plastid rDNA and L - DGGE ladder.

Figure 3.4: DGGE gel of plant macerate subjected to 20%, 23% and 26% sucrose cushion, centrifuged at 4000 rpm for 1 minute, examining the top layer consisting of the top 8 mL of liquid and the pellet amplified with primers 341F-GC-907R. The samples are as follows: lane 1 - original sample, while lanes 2-7 represent the plant macerated centrifuged on various sucrose cushions: 2 - 20% sucrose supernatant, 3 - 20% sucrose pellet, 4 - 23% sucrose supernatant, 5 - 23% sucrose pellet, 6 - 26% sucrose supernatant, 7 - 26% sucrose pellet, 8 - known plastid DNA sample and L - DGGE ladder.

Figure 4.1: Artistic representation of paramagnetic bead attached to phosphorus dendrimer attached to oligonucleotide probe.

Figure 4.2: Simplified schematic illustration of bacterial rDNA separation from plant plastid rDNA.

Figure 4.3: Flow chart simplifying subtractive hybridization protocol.

Figure 4.4: Alignment of plastid and bacterial 16S rRNA sequences labeled according to name of species, order or phyla for plant and bacteria accordingly. Conserved regions in plastid 16S rRNA outlined in orange were used to create oligonucleotide probes.

Figure 4.5: Results of subtractive hybridization conducted with PCR amplicons generated using a test sample (mix equal parts of known bacterial and plastid DNA) as the template, primers 27F and 1492R in a) 5 cycle and 10 cycle and b) 15 cycle PCR reaction. The samples represented in each lane are: 1) Control sample (PCR amplicons with hybridization buffer but no beads/probes), 2) Stripped sample, 3) Hybridized sample for forward probe and 4) Hybridized sample for reverse probe, 5) known bacterial DNA and 6) known plastid DNA. The top band (red arrow) represents bacterial rDNA and lower band (green arrow) represents plastid rDNA.

Figure 5.1: DGGE gel of 341GC-MOD783R PCR amplifications on samples of DNA from 1) original Acer negundo DNA, 2) MscI digested Acer negundo DNA, 3) PvuII digested Acer negundo DNA, 4) original Ulmus parvifolia DNA, 5) MscI digested Ulmus parvifolia DNA, 6) PvuII digested Ulmus parvifolia DNA, 7) original Ulmus pumila DNA, 8) MscI digested Ulmus pumila DNA, 9) PvuII digested Ulmus pumila genomic DNA and L - DGGE ladder.

Figure 6.1: Average culturable bacteria counts obtained per gram of tissue from a) R2A media and b) TSA media per plant species in each season.

Figure 6.2: Averaged species richness and calculated Shannon index values for each plant species in each season per media type. a) Species richness and b) Shannon diversity index for R2A and c) species richness and d) Shannon diversity index for TSA. Figure 6.3: Distribution of known bacterial genera cultured from Acer negundo in the Winter, Summer and Autumn 2012.

Figure 6.4: Distribution of known bacterial genera cultured from Ulmus parvifolia in the Winter, Summer and Autumn 2012.

Figure 6.5: Distribution of known bacterial genera cultured from Ulmus pumila in the Winter, Summer and Autumn 2012.

Figure 6.6: Abundance of bacterial phyla and/or class in the total bacterial isolates obtained from R2A media from all 3 plant species and from all 3 seasons.

Figure 6.7: Abundance of bacterial phyla and/or class in the total bacterial isolates obtained from TSA media from all 3 plant species and from all 3 seasons.

Figure 6.8: Culturable bacterial endophyte abundance based on bacterial phyla and/or class in each season, isolated on R2A and TSA media from all 3 plants.

Figure 6.9: General endophytic bacteria community classified based on phyla and/or class, cultured from each plant species on R2A and TSA media.

Figure 6.10: NMDS plot of bacterial community profiles collected from R2A media separated based on the seasons: a) Winter, b) Summer and c) Autumn. The symbols in the 2 dimensional NMDS plots represent different plant species:  – Acer negundo,  – Ulmus parvifolia and + – Ulmus pumila samples. The stress values for the NMDS plots are as follows: Winter - 0.139, Summer – 0.175 and Autumn – 0.198.

Figure 6.11: NMDS plot of bacterial community profiles collected from TSA media separated based on the seasons: a) Winter, b) Summer and c) Autumn. The symbols in the 2 dimensional NMDS plots represent different plant species:  – Acer negundo,  – Ulmus parvifolia and + – Ulmus pumila samples. The stress values for the NMDS plots are as follows: Winter – 0.162, Summer – 0.121 and Autumn – 0.152.

Figure 6.12: NMDS plot of bacterial community profiles collected from TSA media separated based on plant species: a) Acer negundo and b) Ulmus spp.. The symbols in the 2 dimensional NMDS plots represent different seasons:  – Winter, – Summer and – Autumn. The stress plots for the NMDS plots are as follows: Acer negundo – 0.165 and Ulmus spp. – 0.206.

Figure 6.13: NMDS plot of bacterial community profiles collected from TSA media separated based on plant species: a) Acer negundo and b) Ulmus spp.. The symbols in the 2 dimensional

xii

NMDS plots represent different seasons:  – Winter, – Summer and – Autumn. The stress values for the NMDS plots are as follow: Acer negundo – 0.142 and Ulmus spp. – 0.169.

Figure 6.14: Visual analysis of all the bacterial community profiles collected from R2A media. NMDS plot of the samples based on a) plant species with symbols representing different species:  – Acer negundo,  – Ulmus parvifolia and + – Ulmus pumila samples. b) seasons with symbols representing different seasons:  – Winter, – Summer and – Autumn. The stress value for the NMDS plot was 0.217.

Figure 6.15: Visual analysis of all the bacterial community profiles collected from R2A media. Simplified dendrogram of hierarchical clustering of the samples based on plant species.

Figure 6.16: Visual analysis of all the bacterial community profiles collected from R2A media. Simplified dendrogram of hierarchical clustering of the samples based on seasons.

Figure 6.17: Visual analysis of all the bacterial community profiles collected from TSA media. NMDS plot of the samples based on a) plant species with symbols representing different species:  – Acer negundo,  – Ulmus parvifolia and + – Ulmus pumila samples. b) seasons with symbols representing different seasons:  – Winter, – Summer and – Autumn. The stress value for the NMDS plot was 0.001.

Figure 6.18: Visual analysis of all the bacterial community profiles collected from TSA media. Simplified dendrogram of hierarchical clustering of the samples based on plant species.

Figure 6.19: Visual analysis of all the bacterial community profiles collected from TSA media. Simplified dendrogram of hierarchical clustering of the samples based on seasons.

Figure 7.1: DGGE gel of Acer negundo tree samples collected during Winter, Summer and Autumn 2012. The replicate samples are as follows: lanes 1-3 – branches from tree replicate 1, lanes 4-6 – branches from tree replicate 2, lanes 7-9 – branches from tree replicate 3 and L – DGGE ladder. Some bands were excised and their identities were determined to be as follows: band in red with letter R = Ralstonia spp.

Figure 7.2: DGGE gel of Ulmus parvifolia and Ulmus pumila tree samples collected during Winter, Summer and Autumn 2012. The lanes 1-6 represent Ulmus parvifolia samples with

xiii lanes 1-3 – branches from tree replicate 1 and lanes 4-6 – branches from tree replicate 2. The lanes 10-15 represent Ulmus pumila samples with lanes 10-12 – branches from tree replicate 1, lanes 13-15 – branches from tree replicate 2 and L – DGGE ladder. Some bands were excised and their identities were determined to be as follows: band in red with letter R = Ralstonia spp. and band in green with letter P = Pantoea spp.

Figure 7.3: Averaged phylotype richness and calculated diversity index of each plant species based on their forward and reverse fragments community data obtained in each season. a) phylotype richness and b) diversity index of the bacterial community based on the forward T- RFs data set and c) phylotype richness and d) diversity index of the bacterial community based on the reverse T-RFs data set.

Figure 7.4: Analysis of bacterial community profiles from the forward T-RFs data separated based on the seasons: a) Winter, b) Summer and c) Autumn. The symbols in the 2 dimensional NMDS plots represent different plant species:  – Acer negundo,  – Ulmus parvifolia and + – Ulmus pumila samples. The stress values for the NMDS plots are as follows: Winter - 0.073, Summer – 0.150 and Autumn – 0.186.

Figure 7.5: Analysis of bacterial community profiles from the reverse T-RFs data separated based on the seasons: a) Winter, b) Summer and c) Autumn. The symbols in the 2 dimensional NMDS plots represent different plant species:  – Acer negundo,  – Ulmus parvifolia and + – Ulmus pumila samples. The stress values for the NMDS plots are as follows: Winter – 0.138, Summer – 0.144 and Autumn – 0.211.

Figure 7.6: NMDS plot of bacterial community profiles collected from forward T-RFs data separated based on plant species: a) Acer negundo and b) Ulmus spp.. The symbols in the 2 dimensional NMDS plots represent different seasons:  – Winter, – Summer and – Autumn. The stress plots for the NMDS plots are as follows: Acer negundo – 0.146 and Ulmus spp. – 0.187.

Figure 7.7: NMDS plot of bacterial community profiles collected from forward T-RFs data separated based on plant species: a) Acer negundo and b) Ulmus spp.. The symbols in the 2 dimensional NMDS plots represent different seasons:  – Winter, – Summer and –

xiv

Autumn. The stress values for the NMDS plots are as follow: Acer negundo – 0.205 and Ulmus spp. – 0.184.

Figure 7.8: Analysis of bacterial community profiles from the entire forward T-RFs dataset. NMDS plot of the community profiles based on a) plant species with symbols representing different species:  - Acer negundo,  - Ulmus parvifolia and + - Ulmus pumila samples b) seasons with symbols representing different seasons:  – Winter, – Summer and – Autumn. The stress value for the NMDS plot was 0.205.

Figure 7.9: Analysis of bacterial community profiles from the entire forward T-RFs dataset. Simplified dendrogram of hierarchical clustering generated through Jaccard dissimilarity matrix and Ward’s agglomeration method with separation based on plant species.

Figure 7.10: Analysis of bacterial community profiles from the entire forward T-RFs dataset. Simplified dendrogram of hierarchical clustering generated through Jaccard dissimilarity matrix and Ward’s agglomeration method with separation based on seasons.

Figure 7.11: Analysis of bacterial community profiles from the entire reverse T-RFs data. NMDS plot of the community profiles based on a) plant species with symbols representing different species:  - Acer negundo,  - Ulmus parvifolia and + - Ulmus pumila samples b) seasons with symbols representing different seasons:  – Winter, – Summer and – Autumn. The stress value for the NMDS plot was 0.217.

Figure 7.12: The stress value for the NMDS plot was 0.217. dendrogram of hierarchical clustering generated through Jaccard dissimilarity matrix and Ward’s agglomeration method with separation based on plant species.

Figure 7.13: The stress value for the NMDS plot was 0.217. dendrogram of hierarchical clustering generated through Jaccard dissimilarity matrix and Ward’s agglomeration method with separation based seasons.

xv represent Acer negundo, EC represent Ulmus parvifolia and ES represent Ulmus pumila. Branches outlined in red represent clusters with ≥95% AU values.

Figure D.1b: Hierarchical clustering of bacterial community profiles collected from R2A media separated based on the season: Summer. Dendrograms include approximately unbiased (AU) p- values in red and bootstrap probability (BP) values in blue, with samples labeled with M represent Acer negundo, EC represent Ulmus parvifolia and ES represent Ulmus pumila. Branches outlined in red represent clusters with ≥95% AU values.

Figure D.1c: Hierarchical clustering of bacterial community profiles collected from R2A media separated based on the season: Autumn. Dendrograms include approximately unbiased (AU) p- values in red and bootstrap probability (BP) values in blue, with samples labeled with M represent Acer negundo, EC represent Ulmus parvifolia and ES represent Ulmus pumila. Branches outlined in red represent clusters with ≥95% AU values.

Figure D.2a: Hierarchical clustering of bacterial community profiles collected from TSA media separated based on the season: Winter. Dendrograms include approximately unbiased (AU) p- values in red and bootstrap probability (BP) values in blue, with samples labeled with M represent Acer negundo, EC represent Ulmus parvifolia and ES represent Ulmus pumila. Branches outlined in red represent clusters with ≥95% AU value.

Figure D.2b: Hierarchical clustering of bacterial community profiles collected from TSA media separated based on the season: Summer. Dendrograms include approximately unbiased (AU) p- values in red and bootstrap probability (BP) values in blue, with samples labeled with M represent Acer negundo, EC represent Ulmus parvifolia and ES represent Ulmus pumila. Branches outlined in red represent clusters with ≥95% AU values.

Figure D.2c: Hierarchical clustering of bacterial community profiles collected from TSA media separated based on the season: Autumn. Dendrograms include approximately unbiased (AU) p- values in red and bootstrap probability (BP) values in blue, with samples labeled with M represent Acer negundo, EC represent Ulmus parvifolia and ES represent Ulmus pumila. Branches outlined in red represent clusters with ≥95% AU values.

xvi

Figure D.3a: Hierarchical clustering of bacterial community profiles collected from R2A media separated based on plant species: Acer negundo. Dendrograms include approximately unbiased (AU) p-values in red and bootstrap probability (BP) values in blue, with samples labeled with W representing Winter, S representing Summer and F representing Autumn. Branches outlined in red represent clusters with ≥95% AU values.

Figure D.3b: Hierarchical clustering of bacterial community profiles collected from R2A media separated based on plant species: Ulmus spp.. Dendrograms include approximately unbiased (AU) p-values in red and bootstrap probability (BP) values in blue, with samples labeled with W representing Winter, S representing Summer and F representing Autumn. Branches outlined in red represent clusters with ≥95% AU values.

Figure D.4a: Hierarchical clustering of bacterial community profiles collected from TSA media separated based on plant species: Acer negundo. Dendrograms include approximately unbiased (AU) p-values in red and bootstrap probability (BP) values in blue, with samples labeled with W representing Winter, S representing Summer and F representing Autumn. Branches outlined in red represent clusters with ≥95% AU values.

Figure D.4b: Hierarchical clustering of bacterial community profiles collected from TSA media separated based on plant species: Ulmus spp.. Dendrograms include approximately unbiased (AU) p-values in red and bootstrap probability (BP) values in blue, with samples labeled with W representing Winter, S representing Summer and F representing Autumn. Branches outlined in red represent clusters with ≥95% AU values.

Figure D.5a: Hierarchical clustering of bacterial community profiles collected from forward T- RFs data separated based on the season: Winter. Dendrograms include approximately unbiased (AU) p-values in red and bootstrap probability (BP) values in blue, with samples labeled with M represent Acer negundo, EC represent Ulmus parvifolia and ES represent Ulmus pumila. Branches outlined in red represent clusters with ≥95% AU values.

xvii represent Acer negundo, EC represent Ulmus parvifolia and ES represent Ulmus pumila. Branches outlined in red represent clusters with ≥95% AU values.

Figure D.5c: Hierarchical clustering of bacterial community profiles collected from forward T- RFs data separated based on the seasons: Autumn. Dendrograms include approximately unbiased (AU) p-values in red and bootstrap probability (BP) values in blue, with samples labeled with M represent Acer negundo, EC represent Ulmus parvifolia and ES represent Ulmus pumila. Branches outlined in red represent clusters with ≥95% AU values.

Figure D.6a: Hierarchical clustering of bacterial community profiles collected from forward T- RFs data separated based on the season: Winter. Dendrograms include approximately unbiased (AU) p-values in red and bootstrap probability (BP) values in blue, with samples labeled with M represent Acer negundo, EC represent Ulmus parvifolia and ES represent Ulmus pumila. Branches outlined in red represent clusters with ≥95% AU values.

Figure D.6b: Hierarchical clustering of bacterial community profiles collected from forward T- RFs data separated based on the season: Summer. Dendrograms include approximately unbiased (AU) p-values in red and bootstrap probability (BP) values in blue, with samples labeled with M represent Acer negundo, EC represent Ulmus parvifolia and ES represent Ulmus pumila. Branches outlined in red represent clusters with ≥95% AU values.

Figure D.6c: Hierarchical clustering of bacterial community profiles collected from forward T- RFs data separated based on the season: Autumn. Dendrograms include approximately unbiased (AU) p-values in red and bootstrap probability (BP) values in blue, with samples labeled with M represent Acer negundo, EC represent Ulmus parvifolia and ES represent Ulmus pumila. Branches outlined in red represent clusters with ≥95% AU values.

Figure D.7a: Hierarchical clustering of bacterial community profiles collected from forward T- RFs data separated based on plant species: Acer negundo. Dendrograms include approximately unbiased (AU) p-values in red and bootstrap probability (BP) values in blue, with samples labeled with W representing Winter, S representing Summer and F representing Autumn. Branches outlined in red represent clusters with ≥95% AU values.

xviii

Figure D.7b: Hierarchical clustering of bacterial community profiles collected from forward T- RFs data separated based on plant species: Ulmus spp. Dendrograms include approximately unbiased (AU) p-values in red and bootstrap probability (BP) values in blue, with samples labeled with W representing Winter, S representing Summer and F representing Autumn. Branches outlined in red represent clusters with ≥95% AU values.

Figure D.8a: Hierarchical clustering of bacterial community profiles collected from reverse T- RFs data separated based on plant species: Acer negundo. Dendrograms include approximately unbiased (AU) p-values in red and bootstrap probability (BP) values in blue, with samples labeled with W representing Winter, S representing Summer and F representing Autumn. Branches outlined in red represent clusters with ≥95% AU values.

Figure D.8b: Hierarchical clustering of bacterial community profiles collected from reverse T- RFs data separated based on plant species: Ulmus spp.. Dendrograms include approximately unbiased (AU) p-values in red and bootstrap probability (BP) values in blue, with samples labeled with W representing Winter, S representing Summer and F representing Autumn. Branches outlined in red represent clusters with ≥95% AU values.

xix

List of Appendices

Appendix A – RDP Database Information

Appendix B – Bacterial MspI Restriction Fragment Sizes

Appendix C – Cultured Bacterial Phyla Abundances

Appendix D – Hierarchical Clustering

Chapter 1 An Introduction to Bacterial Endophytes 1.1 General Introduction

Bacterial endophytes were described in literature as early as late 1940’s and the early 1950’s (Hollis, 1951; Tervet & Hollis, 1948). Hallmann et al., (1997) provide one of the most commonly used definitions of bacterial endophytes: bacteria extracted from healthy looking, surface-sterilized plants. However even if the plant in question appears healthy, it is uncertain whether the bacteria isolated is a pathogen, as dependent on certain environmental conditions or due to latent effects the bacteria might become a pathogen. Therefore these endophytic bacteria may be best described as any bacteria that reside within the internal tissues of plants, whether they are active or latent pathogens, symbionts with the plants or simply have a commensal relationship with their hosts.

Endophytes are believed to originate from the epiphytic bacterial communities of the rhizosphere and phylloplane, endophyte-infested seeds or planting materials as well as natural openings or wounds (Hallmann et al., 1997). Typically the endophytic bacterial densities are acropetally distributed; there is a higher density of endophytes in the roots and base of the plants that progressively decreases towards the leaves of the plant (Gardner, 1982; Quadt-Hallmann & Kloepper, 1996). They can be found in the xylem, intercellular spaces, in the endoderm and between and within the cells of the cortex (Gardner, 1982; Patriquin & Döbereiner, 1978). Once inside the plant these bacteria can offer advantages to their hosts including increasing nutrient acquisition, growth and development promotion through mechanisms such as nitrogen fixation, stress tolerance, pathogen and disease resistance, helping the plants to establish under adverse conditions (Adhikari et al., 2001; Cook et al., 1995; Doty et al., 2009; Moore et al., 2006; Ryan et al., 2008; Strobel et al., 2004). The latter is significant in the phytoremediation of contaminated sites as plants enhance the degradation of toxic compounds and organic contaminants of soil such as trichloroethylene (TCE) (Bankston et al., 2002; Newman et al., 1999), 2,4-dichlorophenoxyacetic acid (Germaine et al., 2006), TNT (Taghavi et al., 2005) and hydrocarbons (Phillips et al., 2009; Tesar, Reichenauer, & Sessitsch, 2002; Yousaf et al., 2010).

1 2

Presently a large number of studies in the literature on bacterial endophytes have focused on plants that are of monetary and agricultural importance including rice (Stoltzfus, So, & Malarvithi, 1997), wheat (Conn & Franco, 2004) , soybeans (Okubo et al., 2009), corn (Figueiredo et al., 2009) and potatoes (Garbeva, Overbeek, Vuurde, & Elsas, 2001). There is increasingly more focus on plants for the purpose of finding bioactive compounds (Strobel et al., 2004), as well as for phytoremediation purposes including willows, cottonwood (Doty et al., 2009), aspen (Yrjala & Mancano, 2010) and poplar (Ulrich, Ulrich, & Ewald, 2008). Most of the information obtained about these endophytes pertains to which bacterial genera are present and what genes they possess.

The majority of the information collected about endophytes has been collected via classical culture dependent methods, whereby the surface of the plant is sterilized using chemicals or by flaming, followed by dilution of plant macerates onto media plates and the isolation of the bacteria (Garbeva et al., 2001; Pereira et al., 2011; Rosenblueth & Martínez-Romero, 2006). In most environments, it has been found that the cultivation of bacteria allows only for a fraction of the bacteria in the entire population to be detected (Kent & Triplett, 2002). In soils, sediments and freshwater samples, it has been found that the culturable bacteria represent less than 1% of the total bacterial population present, but it is uncertain if this is the case with endophytes (Amann, Ludwig, & Schleifer, 1995). With the biases associated with the medium and the culturing conditions, this hampers our ability to fully study the endophytic community of plants (Saito et al., 2007; Sun et al., 2008).

With advancements in technology, bacterial endophyte diversity can be studied using rapid, simple, culture-independent molecular approaches with DNA extracted from the interior of the plant. The identification and characterization of these bacterial endophytes has been achieved through the use of the highly conserved 16S rRNA gene that is found in all microorganisms, circumventing the culturing of bacterial endophytes (Olsen et al., 1986; Pace et al., 1986). These analysis have included the use of PCR amplification of 16S rDNA, cloning and sequencing, amplified ribosomal DNA restriction analysis (ARDRA), denaturing gradient gel electrophoresis (DGGE), terminal restriction fragment length polymorphism (T-RFLP) and most recently pyrosequencing (Manter, Delgado, Holm, & Stong, 2010; Pereira et al., 2011; Saito, Ikeda, Ezura, & Minamisawa, 2007; Sun et al., 2008; Wang et al., 2008). The problem with the use of molecular methods to study the endophytes arises from the interference encountered from plant

3 organelles chloroplast and mitochondria that hinder the detection and identification of bacterial endophytes in the community.

The identification of endophytic bacteria provides an understanding of the structure and species composition of plant associated bacterial populations that make up the endophytic community. This in turns allows us to understand which endophytes are beneficial to the plant and determine the ecological role they play in the plant-bacteria interaction. Ultimately this analysis of community can be used to provide more knowledge of endophytic bacterial diversity in plants growing under different conditions including plants exposed to various contaminants and help determine the potential fate of introduced strains into the plants. A key question is whether or not plant species harbour specific endophytes, or if they can act as hosts to any endophytic bacteria. In order to determine the specificity of the plant/bacterial association we need to first understand the determinants of bacterial community variation over time and space. Like other research with endophytes, a lot of the community variation studies involving different factors have focused typically on non-woody, agricultural plants and most of these involved the use of culture dependent methods (Adams & Kloepper, 2002; Kuklinsky-Sobral et al., 2004). Currently there are only a few studies that have looked at the variation of the bacterial endophytes in woody plants - whether variation due to the temperature or season, or variation between plants collected from the same location (Izumi et al., 2008; Mengoni et al., 2009; Mocali et al., 2003). One of these studies was done by Mocali et al. (2003), who looked at the seasonal variations of culturable bacteria isolated from elm stem and root tissues. They were able to find that fluctuations in the bacterial community were influenced by variations in temperature. Izumi et al., (2008), who studied the diversity of predominant endophytes present in Scots pine, silver birch and rowan using culture dependent methods and culture independent methods through DGGE, found that no significant difference was found with respect to the observed communities based on the different tree species. However, for their study, the authors do mention the interference of plant organelle in their DGGE analyses, which might have affected the results and the subsequent conclusions. Another study was done by Mengoni et al., (2009) who looked at variability of bacterial communities associated with the leaves of Alyssum bertolonii, a nickel hyperaccumulator, through the use of a molecular method, T-RFLP, terminal restriction fragment length polymorphism. Mengoni et al., concluded there was a high plant-to-plant variation in the leaf associated endophytic community and despite this they were able to find a

4 core group of bacterial phylotypes that were specific to this plant. With minimal research available on the study of endophytic community variation of trees via their branches using culture independent methods, the following research was undertaken with two objectives in mind. The first objective was to optimize the molecular methods for the culture independent analysis of the endophytic communities. This involved testing the currently used molecular methods employed in the culture independent analysis of other plants.

The second objective was to study the variation of the bacterial endophytic community within and between tree species collected over different season using culture dependent and culture independent methods. The plants to be studied were plants that appeared to be healthy and thriving although they were found growing on a hydrocarbon and trichloroethylene contaminated site. These plants were Acer negundo, Ulmus parvifolia and Ulmus pumila trees; trees typically found in urban environments. By studying these plants from the same site, in close proximity to each other, we can focus on studying the importance of tree species and season to bacterial communities, independent of concerns of different site differences. My work focused on testing the following two hypotheses using three tree species from one site.

1) The endophytic communities of different plant species are host specific and can be distinguished statistically.

2) The bacterial endophytic composition of a tree is influenced by the season.

1.2 Current Culture Independent Methods for Studying Endophytes

Given that culturing conditions used to isolate individual bacterial endophytes hamper the ability to fully detect all the endophytes present in a plant, the use of culture-independent molecular approaches have become a better alternative to study the bacterial diversity. The culture- independent approach entails extracting the DNA from the plant and analyzing it using various molecular approaches. However, because of how the DNA is extracted, there have been problems caused by the interference of plant plastid DNA and mitochondrial DNA, which often outcompete microbial DNA as template DNA for PCR reactions.

The interference is a result of the mitochondria being derived from the endosymbiosis of an α- proteobacterial ancestor, whereas plant chloroplast (also known as plastids), evolved from the endosymbiosis of a cyanobacterium (Barkan, 2011; Gray, Burger, & Lang, 2001; Keeling, 2004; Niklas & Kutschera, 2010). The plastid genome has retained about 100 genes from the cyanobacterial ancestor including genes that encode the RNA polymerase, ribosomal proteins, tRNAs and rRNAs (Barkan, 2011). Even with the reduced plastid genome, the functional domains of the rRNAs are highly conserved and show 65%-80% similarity to eubacterial ribosomal RNAs (Wicke et al., 2011). This conservation makes it hard for current universal PCR primers designed to target the conserved bacterial 16S rDNA regions to differentiate between plastid and bacterial DNA.

A commonly used alternative to overcome the interference by the plant plastid SSU rRNA involves the use of primers that specifically target bacterial 16S rRNA while excluding plastid DNA. This has been done by Chelius & Triplett (2001), who designed primer 799F and Sakai et al. (2004), who modified primer 799F to create primer 783R, a mixture of 3 primers (783RA,783RB and 783RC) that are the reverse complement of 799F with some modifications. Figure 1.1 illustrates a sample alignment of the endophytic bacteria 16S rRNA, plant plastid rRNA and plant mitochondrion rRNA sequences and where the primers 799F and 783R would bind.

A problem encountered with these primers, mainly 799F, was the fact that its ability to exclude plastid DNA was dependent on the genotype of the plant (Rasche & Trondl, 2006). This means that the primers must be tested each time with each plant species in order to ensure it does not amplify plastid DNA. Another problem is its ability to exclude mitochondrial rRNA, as seen in Figure 1.1, both primers also bind to mitochondrial rRNA. The 799F primer and its reverse primer pair amplify mitochondrial fragment that theoretically can be isolated from bacterial amplicons through size separation techniques - since mitochondrial amplicons are ~1.5 X the size of bacterial amplicons (Chelius & Triplett, 2001). If the 799F primer was paired with primer 1492R, the mitochondrial amplicons would be ~1200 bp while bacterial amplicons would

7 be ~738 bp in size. On the other hand with the 783R primer paired with 27F primer, the difference between the mitochondrial amplicons (~830 bp in size) and the bacterial amplicons (~780 bp in size) is not as drastic as that seen with the 799F primer.

An alternative method to the plastid excluding primers has been to use universal 16S primers to amplify both plastid and bacterial 16S rRNA sequences that are later differentiated through the digestion of the PCR products with PvuII, as the enzyme has a restriction site of 5’-CAGCTG-3’ in plastid 16S rDNA sequence that is not found in most eubacterial 16S rDNA genes (Sessitsch et al., 2002; Wang et al., 2008). There has also been another proposed method whereby the plastid interference is minimized with the pre-extraction of the DNA from the plant tissue. This has involved the use of hydrolytic enzymes and differential centrifugation to separate the plant plastid from the bacterial cells (Jiao et al., 2005).

Regardless of which method is chosen to exclude or minimize plant organelle interference the study of the endophytic community in plants can be carried out using similar methods used to study the bacterial communities of other environments such as soil and water. Bacterial 16S rRNA amplicons can be used to generate community fingerprints via two main techniques: DGGE and T-RFLP. Denaturing gradient gel electrophoresis, DGGE, takes the 16S rDNA amplicons of the same length and separates them on a gel based on their different base pair composition - i.e. GC content. This method provides a snapshot of the community, showing who the dominant members are, and allows for the identification of members of the community through the excision and Sanger sequencing of bands.

T-RFLP or terminal restriction fragment length polymorphism, involves the use of fluorescently labeled PCR primers in a PCR reaction that produces fluorescently labeled amplicons. The amplicons are then digested with restriction enzymes that target different sites in different sequences. This generates a mixture of labeled terminal fragments (T-RFs), whose size and abundance in the sample are detected once the sample is passed through a capillary sequencer (Schutte et al 2008). This generates an electropherogram of T-RFs and their sizes, whereby ideally each fragment represents a single operational taxonomic unit or phylotype found in the sample, providing a community profile (Abdo et al., 2006). There is no opportunity for isolation and sequencing of T-RFs but to a limited extent they can be identified by comparison to a

8 database. Both TRFLP and DGGE allow for the monitoring of spatial and temporal changes in the microbial community structure.

Unlike DGGE and T-RFLP that require subsequent analysis to determine the sequence of the amplicons and therefore the identity of individual members of the community, pyrosequencing is an alternative method that allows direct sequencing of amplicons in a sample. Pyrosequencing is a high-throughput technique that allows for the metagenomic analysis of samples, capable of producing approximately106 sequence reads of an average size of 400 bp per run (Manter et al., 2010). This allows for a more in depth analysis into the microbial diversity. The coupling of pyrosequencing with barcoding methods that allow assignment of reads to individual samples, allows many samples to be processed in one run, leading to reduced per-sample costs and a greater feasibility of ecological studies (Dowd et al., 2008). The only problem is that all the plant organelle contamination should be minimized, ensuring that the majority of the amplicons belong to bacterial endophytes and not plant organelle. Currently the endophytic studies that have been carried out using pyrosequencing have resulted in the presence of plant organelle rDNA amongst the reads obtained and in some cases accounting for between ~68% to ~99% of the reads generated (Gottel et al., 2011; Lucero et al., 2011).

Chapter 2 Analysis of Molecular Methods for Post Extraction Exclusion of Plastid Amplicons and Analysis of Endophytic Community

The plastid excluding primers were tested on DNA extracted from Acer negundo and Ulmus spp., the plants to be studied later on, to ensure successful exclusion of these plant organelles. To compare the effectiveness of the primers, universal bacterial primers that would have otherwise been used for DGGE and 16S analysis were tested alongside the plastid excluding primers. DGGE, which relies on the separation of same sized, different sequence fragments based on their base pair composition, was chosen as a fast and effective way to visualize the variety of 16S rRNA detected by the primers and to test for interference/dominance of plastid rRNA.

2.1 Methods 2.1.1 In silico Analysis of Primers

The sequences of the primers to be tested were inputted as probes into the Ribosomal Database Project (RDP) applet (http://rdp.cme.msu.edu/; Cole et al., 2009) to compare the number of bacterial 16S rRNA sequences each primer could target. The primers tested are outlined in Table 2.1. The RDP results were narrowed to only include sequences ≥1200bp and classified as good quality.

9 10

Table 2.1: Primers and their corresponding sequences tested in the RDP database. Prime Name Primer sequence (5’-3’) 27F AGAGTTTGATYMTGGCTCAG 1492R TACCTTGTTACGACTT 341F CCTACGGGAGGCAGCA 907R CCGTCAATTCMTTTGAGTTT 799F AACMGGATTAGATACCCKG

783R A CTACCAGGGTATCTAATCCTG B CTACCGGGGTATCTAATCCCG C CTACCCGGGTATCTAATCCGG 1197R TTGACGTCATCCCCACCTTC

2.1.2 Surface sterilization of Acer negundo and Ulmus spp. Branches

Branches roughly 1.5-2.0 cm in diameter were collected from Acer negundo, Ulmus parvifolia and Ulmus pumila trees from a hydrocarbon-contaminated site in February 2012. Prior to undergoing surface sterilization, the leaves and smaller branches were removed. The branches were subjected to a detergent wash followed by rinse in water and distilled water to remove any dirt found on the surface of the samples. The branches were cut to smaller sections ~ 9 cm in length and were surface sterilized through a series of washes in ethanol and 1.5% bleach. Briefly, the sections were placed in a sterile Falcon tube, vortexed in 70% ethanol solution for 1.5 mins, followed by incubation in the 70% ethanol for 4 mins. The samples were then vortexed in a solution of 1.5% bleach/0.1% Tween 20, after which they were incubated in 1.5% bleach solution for 20 minutes. The sections were rinsed 5X with sterile distilled water to remove any residual bleach, dipped in 70% ethanol and flamed. To assess the effectiveness of the surface sterilization the final wash in sterile water was spread plated and the sterilized surfaces were imprinted onto agar plates.

2.1.3 Isolation of DNA from Acer negundo and Ulmus spp. Branches

The DNA was extracted from ~5-10 g of surface sterilized branch samples through the use of the FastDNA SPIN kit (MP biomedicals) with some modifications. The periderm of each branch sample was removed before homogenization in a single speed Waring Blender (20000 rpm) with 60 mL of 50 mM Tris-HCl buffer solution. The homogenate was then filtered through 8 layers of sterile cheesecloth, before it was centrifuged at 600xg for 5 minutes then subsequently centrifuged at 8000 rpm for 10 minutes to ensure all the bacterial cells in the sample were pelleted. The pellet was resuspended in 200 µL of 50 mM Tris-HCl solution and this solution was placed into the lysing A matrix tubes provided in the spin kit along with 1mL of CLS-TC solution and the addition of 100 µL of PPS (protein precipitating solution), before the entire solution was homogenized in a bead beater for 55 seconds. The extraction was carried out according to manufacturer’s protocol with 2 additional washes with SEW-SM (ethanol) solution.

2.1.4 Isolation of Plastid 16S rRNA Amplicons

In order to quickly isolate plastid DNA from the plants of interest, a method used by Sessitsch et al. (2002) and Wang et al. 2008 was carried out. The genomic DNA extracted from each plant would be amplified with the 27F and 1492R primer to obtain the full 16S rRNA sequence. These amplicons would consist of a mix of plastid and bacterial rDNA. As done by Sessitsch et al. (2002) and Wang et al. (2008), the PCR products from the amplification would be digested with restriction enzyme PvuII. Ideally the product digested would be of plastid origin and result in a fragment ~1300 bp in size, while leaving the bacterial amplicons, ~1500 bp in size, intact.

The extracted plant DNA was used as a template in a PCR reaction using primers that amplified the full 16S rRNA sequence, primers 27F and 1492R. Amplification was carried out in a 20 µL reaction with a final concentration of 0.5 mM of each of the forward primer, 27F and reverse primer, 1492R, 1.5 mM MgCl2, 200 mM of each dNTP, 2.5 units of HotStarTaq Plus DNA polymerase (Qiagen, Canada) and 1 µL of genomic DNA (30 ng/µL – 50 ng/µL). The PCR reaction was carried out in a PTC-200 thermal cycler (MJ Research Inc., CA, USA) using the following program: initial denaturing at 95˚C for 5 minutes followed by 35 cycles of: denaturing at 95˚C for 1 min, annealing at 56˚C for 1 min and extension at 72˚C for 1 min; final extension at 72˚C for 10 mins. Negative (no template) and positive (known bacterial DNA and

Arabidopsis thaliana plastid DNA obtained from Zhao Lab, UTSC) controls were included each time a PCR reaction was performed. The resultant products were visualized on a 1% agarose gel to ensure the PCR reaction worked without any contamination before 10 µL of PCR product was digested with 0.1µL of 10U/µL PvuII restriction enzyme (NEB Canada). The digested products were visualized on a 1.5% agarose gel and the bottom bands (~1300 bp) were cut out of the gel and purified with QIAEX II Gel Extraction Kit (Qiagen, Canada). The purified amplicons should ideally just consist of plastid amplicons.

2.1.5 Primer Testing on Extracted Plastid Amplicons

The gel-purified product was used as a template to test with the following primer pairs: 341FGC and 907R, 341FGC and MOD783R (a modification of primers by Sakai et al., made up of equimolar concentrations of primer 783RA and primer 783RC), and 799FGC and 1197R. The GC represents the GC clamp present in the 5’ end of the 341F and the 799F primers with the sequence 5’-CGCCCGCCGCGCGCGGCGGGCGGGGCGGGGGCACGGGGGG-3’, added to prevent the complete dissociation of the 2 DNA strands in the denaturing gel and allowing for the improved detection of variants in the sequence (Muyzer & Smalla, 1998). The 20 µL PCR reactions (as above) were carried out in a PTC-200 thermal cycler (MJ Research Inc.) with the following conditions: initial denaturing at 95˚C for 5 minutes followed by 35 cycles of: denaturing at 95˚C for 1 min, annealing at 56˚C for 1 min and extension at 72˚C for 1 min; final extension at 72˚C for 10 mins. A portion of the product (~2.5 µL) was visualized on a 1% agarose gel before the remainder was run in a DGGE gel. For the DGGE gel, a 6% polyacrylamide gel with a gradient of 40%-70% denaturing solution was run in a DGGE-2001 Tank (C.B.S. Scientific Co, Del Mar, California) in 0.5 X Tris-acetate-EDTA buffer for 20 hours at 70V and 58˚C. The gel was stained with 15 µL of 0.5 mg/mL of ethidium bromide stain in 300 mL of 0.5X TBE buffer for 30 minutes before visualized in a UV-light box.

Some of the bands present in the DGGE gel were excised and placed into 200 µL PCR tubes with 100 µL of sterile water and incubated at 4˚C for 48 hours. Following the incubation, 1 µL of the sample was used as template and reamplified using primers 341F and 783R following the previously mentioned conditions. After it was ensured the PCR reaction was successful, the samples were PCR purified using the GenElute PCR clean up kit (Sigma-Aldrich, MO, USA)

13 before being sent for Sanger sequencing at The Centre for Applied Genomics (TCAG) (Toronto, Canada).

2.2 Results 2.2.1 In silico Analysis of Primers

The universal 16S primers and the plastid excluding primers were compared through the use of a “probe search” applet in the RDP database to determine how many bacterial 16S sequences the primers targeted. The number of land plant plastid sequences, under the classification Streptophyta, that were detected by the primers was also recorded. Results from the probe search are shown in Table 2.2 and a visual comparison is seen in Figure 2.2. Of all the primers tested, primers 341F and 907R had the highest number of bacterial sequences hits in the database and the highest land plastid sequence hits. With regards to the plastid excluding primers, 783R and 799F, they both targeted a larger number of bacterial 16S rRNA sequences, in comparison to the full-length 16S primers 27F and 1492R. Upon closer inspection the 783R primer actually captured a bit more sequences than 799F, but both of them did not bind to any land plant plastid rRNA sequences.

Table 2.2: Number of hits on the RDP database of good quality and greater than 1200 bp in length 16S rRNA sequences, using universal bacterial 16S rRNA primers and plastid excluding primers as probes. Number of bacterial Number of land plant sequences hit out of plastid sequences hit out Primer 1173481 sequences (% of 1801 sequences (% of of Total) Total) 27F 174800 (14.5%) 133 (7.4%) 1492R 183100 (15.6%) 349 (19.4%) 341F 1098541 (93.6%) 1745 (96.9%) 907R 1072053 (91.4%) 1748 (97.0%) 799F 967563 (82.5%) 0 (0%) 783R A 960333 (81.8%) 0 (0%) B 53395 (4.6%) 0 (0%) C 6625 (0.56%) 0 (0%) 1197R 841034 (71.7%) 54 (3%)

*Hits as of November 9/2012

Figure 2.1: Graphical representation of the percentage of bacterial sequences and land plant plastid sequences targeted by each primer tested on the RDP database.

2.2.2 Isolation of Plastid Amplicons

The crude retrieval of plastid amplicons from a PCR sample was done through the amplification of the plant genomic DNA using full 16S rRNA primers and digesting the amplicons with the restriction enzyme PvuII. Ideally the amplicons of plastid origin were digested to ~1300 bp in size while the bacterial amplicons remained intact at ~1500 bp as seen in Figure 2.2. The lower band in the gel representing the plastid amplicons, were gel purified and used for subsequent testing of primers. This also shows that the use of the universal full 16S rRNA primer pairs (27F and 1492R) resulted in the amplification of plastid rRNA and very minimal to no bacterial rRNA.

2.2.3 Analysis of Primer 783R

Before the different primers were tested on the crudely extracted plastid amplicons, a preliminary analysis was conducted on primer 783R. This primer was made up of 3 different primer sequences (783RA, 783RB and 783RC) combined together in equimolar concentrations. The primers were tested individually and combined together on Acer negundo genomic DNA to determine if there would be differences in the endophytic community detected by the primers used separately and the primers combined together. As seen in Figure 2.3, the amplicons obtained using 783RA differed from those obtained from 783RB and 783R (all 3 primers combined). There was no amplification using primer 783RC, thus the results were omitted. When the bands found in Figure 2.3 were sequenced, the bands obtained in 783RB (lane 1) and

783R (lane 3) corresponded to plant mitochondrial rRNA whereas the one obtained from 783RA (lane 2) corresponded to uncultured Ralstonia sp., a bacterial strain. Based on these results, primer 783RB preferentially amplified mitochondrial rRNA and if all the primers were mixed together this would lead to the preferential amplification of mitochondrial rRNA. Therefore the use of primer 783RB was omitted and MOD-783R was used for subsequent analysis using only primers 783RA and 783RC.

2.2.3 Primer Testing on Extracted Plastid Amplicons

The 2 different plastid excluding primers were combined with different primers; MOD-783R was paired with 341FGC and 799FGC was paired with 1197R to ensure the amplicons would be ~ 600 bp in size, allowing it to be visualized in DGGE gels. These primer pairs along with primers 341F-907R, typical primer pairs used for DGGE, were tested on the extracted plastid amplicons. As seen in Figure 2.4, there were different bands obtained by the different primers. Upon sequencing of the bands it was determined that all of the bands obtained using primers 341FGC-907R corresponded to plant plastid rRNA. However those obtained using primers 341FGC-MOD783R and 799FGC-1197R corresponded to bacteria: Pantoea spp. 16S rRNA. Therefore the use of the plastid excluding primers allowed for the amplification of bacterial

17 rRNA and no plastid rRNA. These results also showed that the digestion of amplicons with the restriction enzyme PvuII resulted in the digestion of plastid amplicons as well as some bacterial amplicons. Based on the in silico analysis results, between the 2 plastid excluding primers, primer pair 341F-MOD783R would be the best choice over 799F-1197R, due to the primer 341F resulting in more bacterial rRNA sequences hits compared to primer 1197R.

a) b)

2.2.4 Preliminary Testing of Plastid Excluding Primers on Extracted Plant Samples

The plastid excluding primers (341FGC-MOD783R) and the universal 16S DGGE primers (341FGC-907R) were preliminarily tested on plant extracted DNA from Acer negundo, Ulmus

18 parvifolia and Ulmus pumila collected in February 2012. The resultant amplicons of the PCR reactions using the same concentration of plant DNA as template with the different primer pairs are shown in Figure 2.5. As seen in Figure 2.5a, there was minimal amplification of bacterial rRNA from the plant samples using the plastid excluding primers. However using the same DNA as template, the chloroplast rRNA was easily amplified from the samples with the universal primers 341F-907R (Figure 2.5b). This indicates that the lack of amplification of bacterial rRNA was not attributed to the presence of inhibitors in the template as the chloroplast rRNA was easily amplified. It could be that the concentration of bacterial endophytes in the template was too low to be properly detected by the plastid excluding primers.

a) b)

2.3 Discussion

Plastid excluding primers were tested on DNA extracted from Acer negundo, Ulmus parvifolia and Ulmus pumila. This was necessary as Rasche & Trondl (2006) showed that the ability of the plastid excluding primers to exclude plastid DNA was dependent on the genotype of the plant plastid. For this reason an initial in silico analysis of these primers was performed to determine how well the primers were able to detect bacterial 16S rRNA in comparison to the universal 16S primers available (27F, 1492R, 341F, 907R and 1197R). Out of the plastid excluding primers, primer 783R targeted more bacterial sequences than 799F, as it is a composite of three sequences. Of the sequences that were targeted by both of the plastid excluding primers, none of them were plastid sequences of land plants (classified under Streptophyta). However, not all the

19 plastid of all the plants in the world have been sequenced and stored in this database. Given that the plastid sequences for Acer negundo, Ulmus parvifolia and Ulmus pumila were not found in the database, the primers have to be experimentally tested on the plastid of these plants to determine if they are able to exclude it or not.

In comparison to plastid excluding primers, the full 16S primers, primers 27F and 1492R, had considerably a lot less bacterial sequences hits in the database. This is explained by the fact that these primers, typically used to sequence the full length bacterial 16S rRNA genes, are highly likely to be trimmed from the raw sequences for quality control prior to submission - hence the 27F and 1492R primers had such low hits in the RDP. Unlike the binding sites for the primers 27F and 1492R that are found at the ends of the 16S rRNA sequence, the binding sites for the primers 341F and 907R are found further away from the ends of bacterial 16S rRNA sequences such that they are less likely to be trimmed for quality control. This does mean that if the sequence was trimmed for quality control, it could potentially lose its binding site for primers 27F and 1492R but still retain the binding site for primers 341F and 907R. This explains why the 341F and 907R primers had more bacterial sequence hits in comparison to 27F and 1492R primers, when the latter primers are the most commonly used.

The results obtained from the in situ analysis of the primers and the testing of the primers showed that the universal primers would preferentially amplify plant organelle rRNA over bacterial rRNA and should probably not be used for the study of bacterial endophyte community. In comparison the plastid excluding primers were able to exclude the plastid of Acer negundo and Ulmus spp. making them ideal for bacterial community analysis.

Although the primers seemed like a good option to visualize the endophytic community, given that there was no plastid amplification, a problem arose with the actual amplification of bacterial DNA from the samples. As seen in the preliminary testing on extracted plant samples in section 2.2.4, when the primers were used there was minimal to no amplification of bacterial rRNA from the sample, thus making it hard to study the endophyte community of the sample. There were no problems with the samples as plastid rRNA could be easily amplified from the sample. I interpret these results to mean that the success of the primers in amplifying bacterial DNA is dependent on the initial template. This problem could potentially be attributed to the minimal amount of endophytes present in the sample or it could be attributed to too much interference by

20 plastid DNA present in the sample. For this reason, to optimize the use of the primers, the initial template would need to have the least amount of plastid contamination possible. To address this problem, in the next chapter I describe three methods I tested to see if they could be used to decrease the amount of plastid contamination and allow for the amplification of the bacterial 16S rRNA. These three methods are classified into 2 categories: pre-extraction and post- extraction methods. The pre-extraction method involves separation of plastid from bacterial cells prior to DNA extraction. The post-extraction methods involve the 2 following methods: 1) the subtractive hybridization of organelle DNA through the use of magnetic beads with organelle specific probes and 2) selective enzymatic digestion of plastid DNA. Each of these methods was tested for its effectiveness at limiting the plastid interference and allowing for the detection of the microbial community.

Chapter 3 Physical Separation of Plastid Cells via Differential Centrifugation and Density Gradient Centrifugation 3.1 Introduction

Ideally the best approach to limit the contamination of plastid in the subsequent DNA extracted would be to physically remove the plastid cells from the bacterial cells after they are released from the plant tissue. This is important given that it is estimated there are up to 300 chloroplasts can be found in one plant cell (Diekmann et al., 2008). The methods tested in this section were adapted from studies that isolated plant chloroplasts through the use of differential centrifugation and density gradient centrifugation techniques. Differential centrifugation is a method of separation that relies on the use of differences in the sedimentation rate of particles due to their size and density. The sample is subjected to a series of centrifugation steps with the centrifugation speed increasing after each subsequent centrifugation. In this process typically particles that have higher sedimentation rate, are bigger in size and density, will be pelleted in the initial centrifugation while the smaller particles will pellet in the later staged centrifugation. This typically results in a crude separation of the particles or cells of interest, with some contamination by different sized particles (Wilson & Walker, 2010). After the differential centrifugation, if the sample is subjected to an isopycnic density gradient centrifugation, a better resolution of the particles is obtained. This type of centrifugation is solely dependent on the density of the particles (and not influenced by size). The sample is overlaid on top of a gradient solution of a certain density. Once this is centrifuged, the particles will reach a position of equilibrium in the gradient where the density of the surrounding buffer matches its own density, allowing for its separation from the other particles (Wilson & Walker, 2010).

These protocols require a chloroplast isolation buffer consisting of ~0.25M-0.3M sorbitol that is used to maintain the osmolarity of the organelles and a combination of BSA, PVP and reducing agents such as cysteine, B-mercaptoethanol and ascorbate that serve as protecting agents from hydrolytic enzymes, phenolic compounds, and other secondary products released from plant vacuoles (Gualberto, Handa, & Grienenberger, 1995). EDTA is also included in the buffer in order to protect the membranes from ion-dependent lipases and phospholipases (Douce et al., 1987). The chloroplast isolation buffer was not used as initial testing with the buffer resulted in

21 22 no amplification of bacteria and plastid rDNA from the DNA extracts. The buffer used instead consisted only of 50 mM Tris-HCl, to maintain the pH of the sample. Sucrose and sorbitol were omitted from the sample as the plant macerates would be stored at 4˚C, and the sucrose would serve as a carbon source for some of the bacteria, causing the community to change. In this case without the sucrose or sorbitol in the buffer to maintain the osmolarity of the cells, there is a higher risk that a larger portion of chloroplast cells could potentially be lysed after the homogenization process.

Sucrose and Percoll solutions are typically used to create the gradient to separate and purify intact plastids and mitochondria. Percoll is a colloidal suspension of PVP coated silica particles with a density of 1.13g/cm3 (Gualberto et al., 1995). The sucrose at higher concentrations becomes more viscous and becomes a hyperosmotic solution, causing the organelles to shrink and increase in density. Percoll, like sucrose, creates self-forming gradient solutions when centrifuged, but unlike sucrose it does not cause organelle dehydration as they move through the gradient.

Ultracentrifuges are needed to carry out the centrifugation, which reduces their feasibility for routine extractions. However there were some papers that have modified the protocols to allow for the use of a normal benchtop swing out centrifuge rotors such as Kley et al. (2010) in the isolation process of intact plastid. A combination of the information obtained from the plastid isolation methods from Kley et al. (2010) along with literature buoyant density of pure bacterial cultures and plastid (peas, spinach, castor beans) were used as guidelines to devise protocols for the separation of plastid from bacteria (Bakken & Olsen, 1983; Miflin & Beevers, 1974).

3.2 Method 3.2.1 Preparation of Plant Macerate

Approximately 20 g of branches collected from Acer negundo trees were surface sterilized and homogenized as previously described in Chapter 2, using 100 mL of 50 mM Tris-HCl buffer. The macerate was then filtered through 8 layers of sterile cheesecloth and split into 3 portions. One portion was kept as is, while the second portion was subjected to differential centrifugation. The third portion of the macerate was centrifuged at 600xg for 5 minutes to remove residual plant debris not removed by the cheesecloth, after which the supernatant was centrifuged at

8000 rpm for 10 minutes to pellet the cells. The resultant pellet was resuspended in 5 mL of 50 mM Tris-HCl buffer and used for subsequent testing with density gradients. The macerates and the resuspended pellet were stored at 4˚C until further use.

3.2.2 Differential Centrifugation of the Macerate

A portion of the filtered macerate was used to test the effectiveness of differential centrifugation to separate plastids from the sample. The following method was a modification of that used by Jiao et al. (2005) except that their protocol involved a bacterial growth enrichment step and use of enzymes not used in this protocol. The macerate was subjected to 6 centrifugation steps at 200xg, 500xg, 800xg, 1000xg, 3000xg, 5000xg (for 10 minutes each), whereby in each step the pellet was collected and the supernatant was transferred to another 50 mL Falcon tube. Ideally the heavier particles would be pelleted in the initial centrifugation steps, while the lighter particles would be pellet towards the later centrifugation steps. The DNA of each pellet and the final supernatant was extracted using the FAST DNA Spin Kit as previously described (Chapter 2.1.2) and used as a template for a PCR reaction using primers 341FGC-907R and 341FGC- MOD783R following the standard PCR conditions. Briefly the PCR reaction conditions were: initial denaturing at 95˚C for 5 minutes followed by 35 cycles of: denaturing at 95˚C for 1 min, annealing at 56˚C for 1 min and extension at 72˚C for 1 min; final extension at 72˚C for 10 mins. The 341FGC-907R primers were used in order to visualize the presence of the plastid DNA in the sample. In order to visualize the bacterial communities the samples were amplified with primers 341FGC-MOD783R. The resultant amplicons were run on a 6% polyacrylamide DGGE gel with a 40-70% denaturing gradient for 20 hours at 70V before being stained with ethidium bromide and visualized in an UV light box.

3.2.3 Sucrose Density Gradient Centrifugation

In order to determine which concentrations of sucrose should be tested with the macerate, the specific gravity of various sucrose concentrations were determined through the use of a hydrometer (VWR, Canada) and are listed in Table 3.1.

Table 3.1: Specific gravity of various sucrose solutions made in 50 mM Tris-HCl (pH = 7.69), measured at room temperature.

% w/w of Specific sucrose gravity 16.7 1.067 17.3 1.070 20.0 1.082 23.0 1.096 25.9 1.108 28.5 1.120 31.0 1.132 33.3 1.144 35.0 1.156

Using the density of bacteria, ranging from 1.035 to 1.093 g/cm3 (Bakken & Olsen, 1983), the density of plastid, ranging from 1.15 to 1.22 g/cm3 (Miflin & Beevers, 1974), along with the information from Table 3.1, sucrose cushions and gradients were prepared. A sucrose cushion consists of solely one concentration of sucrose. Sucrose gradients were created by carefully overlaying 1 mL of each layer of sucrose on top of each other with the more concentrated sucrose solution at the bottom and the lighter one at the top. Initially the sucrose gradients and cushions were carried out in 1.5 mL tubes, with 200 µL of sample overlaid the sucrose gradient and/or cushion. Every 100 µL of the centrifuged sample was analyzed and there was no difference seen in the each fraction; both plastid and bacteria were still present in each layer. The sucrose gradient and cushions were scaled up and carried out in 15 mL Falcon tubes, with the hopes of improving the separation. The ratio of sample to sucrose was 1 mL of sample to 9 mL of sucrose cushion or gradient, The sucrose cushions tested in the 15 mL Falcon tubes were of the following concentrations: 20%, 23%, 26%, 30% and 35% sucrose, with the plant macerate overlaid on top of the cushion and centrifuged in a Sorvall Legend RT centrifuge (Thermo Scientific) with a swing out rotor at 4000 rpm for 1 minute. The sole sucrose gradient tested in the 15 mL Falcon tube consisted of a 27% to 36% sucrose gradient that was overlaid with the plant macerate and centrifuged at 4000 rpm for 1 hour. Two fractions were examined after centrifugation: 1) the supernatant and 2) the pellet. For the supernatant, this involved aspirating 8 mL from the top and centrifuging it at 10000 rpm in a fixed angle rotor for 20 minutes to pellet

25 any cells present in that layer. Originally smaller fractions, consisting of every 2 mL from the top of the supernatant were extracted and analyzed molecularly however there was no amplification from these fractions. Therefore the volume of the supernatant to be analyzed was increased to the top 8 mL of the supernatant.

The DNA from the 2 fractions was extracted using the FastDNA SPIN kit as previously mentioned. In order to analyze the DNA extracted from either the differential centrifugation or the density gradient centrifugation, the samples were amplified in a PCR reaction with primers 341FGC-907R in order to visualize the presence of the plastid DNA in the sample. The resultant amplicons were run on a 6% polyacrylamide DGGE gel with a 40-70% denaturing gradient for 20 hours at 70V before being stained with ethidium bromide and visualized in an UV light box.

3.3 Results 3.3.1 Differential Centrifugation of the Macerate

According to literature, the density of bacteria range from 1.035 to 1.093 g/cm3 (Bakken & Olsen, 1983) while the density of plastid range from 1.15 to 1.22 g/cm3 (Miflin & Beevers, 1974). Since plastid cells have a higher density compared to bacterial cells it would expected that the majority of the plastid cells would pellet first and their concentration would decrease towards the higher centrifugation step. For the bacterial cells, it would be expected that they would pellet and found in higher concentrations after the higher centrifugation speeds. After the differential centrifugation there were no changes seen in the plastid DNA detected in each of the cell pellets compared to the original plant macerate sample. There was still plastid detected in the left over liquid. With the bacterial community DNA detected, Figure 3.1b, there was an improvement in the detection of bacteria in the pellets formed by centrifugation at 200xg and 500xg in comparison to the original sample. However subsequent centrifugation steps led to a decrease of bacteria in the sample and no bacteria was detected in the left over liquid sample.

a) b)

3.3.2 Sucrose Density Gradient Centrifugation

A sample of previously isolated pure bacterial endophyte culture grown on R2A media was tested along side an Acer negundo plant macerate sample spiked with some of the pure bacterial endophyte culture. The plant sample was spiked with some of the pure bacterial culture to increase the concentration of endophytes, allowing for better visualization of their location after centrifugation. Initially testing with the density gradient centrifugation showed that plant macerate overlaid on top of the sucrose cushion or gradient, resulted in a visual separation of the original samples into different layers and fractions in the tube as seen in Figure 3.2. In the pure bacterial sample, there was no pellet at the bottom of the tube but greyish-white band was seen, representing where the bacteria settled in the tube. In comparison, for the plant macerate sample a pellet was present at the bottom of the tube and three layers were noticeable in the tube. There was a matching layer found at a similar position to the bacterial band in the pure bacterial culture. There was another similar looking band found above the green band in the tube,

27 potentially representing the other endophytes on the sample while the green band could represent the chloroplast in the sample.

a) b)

Using the information obtained from the preliminary density gradient centrifugation, the sucrose density gradient centrifugation was initially carried out with a sucrose gradient of 27%-36% sucrose and sucrose cushions consisting of 30% and 35% sucrose solutions. This allowed for the testing of the plant macerate on sucrose solutions with a specific gravity greater than 1.12. Therefore the bacterial cells should remain in the top layer or in solution, while plastid cells should be found only in the pellet. The molecular analyses of the different sucrose concentrations and the different fractions analyzed are shown in Figure 3.3. Results show that both plastid and bacteria are in both fractions analyzed for the sucrose concentrations tested.

Even though the results were not shown for the centrifugation of the plant macerate on the 27%- 36% sucrose gradient, a duplicate sample was run where every 1 mL layer was aspirated and stored into microtubes before a portion of the liquid was spread plated onto R2A media plates.

After incubation all the plates had bacterial growth present, indicating that the bacteria was found throughout all the layers in the gradient even the pellet.

a) b)

With the results obtained above, lower concentrations of sucrose solutions were tested to see if different results would be obtained when the specific gravity of the solution was decreased. The subsequent sucrose solutions tested were 20%, 23% and 26% sucrose cushions, whose specific gravity ranged from 1.082 to 1.108. The molecular analyses conducted on the fractions extracted from each sucrose cushion are shown in Figure 3.3. These samples were only tested for the presence of plastid and not bacteria. Plastid, like the previous results, was found in all the fractions analyzed.

3.4 Discussion

The proposed pre-extraction exclusion of plastid was initially only tested on plant macerate extracted from Acer negundo plants due to its availability. The results showed that the differential and the density gradient centrifugation of the plant macerate did not result in sufficient separation of the plastid cells from bacterial cells such that molecular analysis could detect the minimization of the interference. Unlike the chloroplast isolation protocols that typically involve the use of the leaves from the plants, the plant macerate from the stems were used. It is uncertain if this difference in the type of sample used might or might not affect how well the plastid isolation protocol worked. This is due to the fact that in plant leaves the plastids present are typically chloroplast. However in the stem and branches of the plant, along with

30 chloroplast there are proplastids present, small, colourless to pale green undifferentiated plastids (Possingham, 1980). There are no published data confirming if the density of the proplastids are the same as chloroplasts, so the experiment was carried out with the assumption of similar densities which might or might not be the case. Another factor that might have contributed to the results is the extraction buffer. Unlike the protocols that typically macerated the plant tissue in chloroplast isolation buffer (CIB), the plant tissue was macerate in 50 mM Tris-HCl buffer, as experimentally problems encountered with the use of the CIB with regards to subsequent molecular analysis, therefore it’s use was omitted from methods.

It is inevitable that some chloroplast DNA would be released into the surrounding liquid due to the homogenization in the blender that would cause for plant cells and chloroplast cells to be lysed and DNA released into the homogenate (Aronsson & Jarvis, 2002). Ideally the CIB buffer would allow for the intact chloroplast cells that were released into the homogenate to be protected from the change in osmolarity; it would have allowed a majority of the chloroplast cells to remain intact. Therefore in comparison to using the CIB buffer, the alternate use of the Tris-HCl buffer that only maintained the pH, potentially more chloroplast cells could have been lysed. This might have an effect on how well the subsequent differential centrifugation and density gradient centrifugation experiments worked.

The differential centrifugation of the plant macerate had no effect on the concentration of the plastids, as it was detected in every centrifugation step, but affected the concentration of bacteria in the sample. After all the centrifugation steps, there was no bacteria detected in the left over liquid but plastid was still present. This could indicate one of two things: it could indicate that the bacterial cells might be potentially of higher density than the plastid cells, since they were removed first from the solution or that plastid DNA is found everywhere and bacterial cells cannot be separated from it based on sedimentation rate itself. There is a great likelihood that the latter occurred given the lack of osmolarity protection offered by the Tris-HCl buffer, leading to the potential lyses of some chloroplast cells.

Although the differential centrifugation did not work, the density gradient centrifugation method using sucrose was still tested to see if different results would be obtained using the differences in the density of plastid and bacterial cells. The density gradient centrifugation was run under the assumption that plastid cells had a density between 1.15-1.22 g/cm3 (Miflin & Beevers, 1974),

31 based on literature on the isolation of plastid from peas, spinach and castor beans and bacterial cells had a density that ranged from 1.035 to 1.093 g/cm3 (Bakken & Olsen, 1983). Therefore the plant macerate was tested out on sucrose concentrations whose measured specific gravity was calculated to be less than the density expected of plastid. This indicates that ideally the bacteria cells should remain in solution whereas the plastid cells that are heavier should pellet at the bottom of the tube. However subjecting the plant macerate to various sucrose concentrations and gradients showed that the plastid was not confined to a specific fraction in the tube, but was instead found in all the fractions analyzed. This was also the case with the bacteria that was found, through molecular and culturable methods, to be present throughout the centrifuged sample. This does signify that the density of the bacterial endophytes covers a range of densities that differ from that obtained by Bakken & Olsen (1983). This is probable given that the literature densities are based on pure bacterial cultures from plates, as no studies have been carried out on the density of endophytic bacteria communities that consist of a mix of various bacteria genera.

In the end, it could be that density gradient centrifugation method was not optimized. Future work should test the improvements possible with the use of an ultracentrifuge, as this might potentially lead to a cleaner separation of the bacteria and plastid as they are currently found in all the layers. However, it could be that the density of endophytic bacteria encompasses a range that overlaps with the density of the plastid from Acer negundo. Regardless of the exact reason for the presence of the cells through the gradients, these types of methods may simply be unsuitable for pairing with molecular methods, as enough cells persist in every layer to be detected with PCR. However from the preliminary results obtained from the testing conducted, it does not seem likely that these pre-extraction methods would be ideal to the minimize the amount of plastid contamination without decreasing the amount of bacteria being extracted from the plant, therefore the post-extraction methods were taken into consideration.

Chapter 4 Subtractive Hybridization 4.1 Introduction

Magnetic or paramagnetic beads attached to molecular probes have previously been used successfully in the field of molecular biology to immobilize proteins and other compounds, as well as purifying DNA and selectively enrich DNA from various samples (Archer, Lin, Wang, & Stenger, 2006; Archer & Lin, 2011; Berensmeier, 2006; Franzreb et al., 2006; Galluzzi et al., 2006; Olsvik, Popovic, & Skjerve, 1994). The following chapter describes a modified protocol tested for the purposes of selectively removing organelle DNA from plant tissue DNA extracts. It is based on the use of paramagnetic beads and oligonucleotide probes, as described by Archer et al. (2006) for the selective extraction of genomic DNA, with some modifications. The paramagnetic beads are beads that only become magnetic in the presence of an external magnetic field. The beads are coated with phosphorus dendrimer linkers, onto which oligonucleotide probes are attached. The phosphorus dendrimers allow for a higher loading capacity of probes onto the bead compared to just the probes added straight on the beads while minimizing the steric hindrance. A visualization of this construct can be seen in Figure 4.1. For the purpose of this experiment, the Archer et al. protocol was modified such that paramagnetic beads would have probes that consisted of 60bp oligonucleotides that would bind specifically to plastid DNA.

I tested this method based on the assumption that plastid DNA in the extracted samples is abundant relative to bacterial DNA and inhibits amplification of bacterial DNA. Once these magnetic beads are mixed with the sample, the probes would ideally only hybridize with the plastid rDNA and allow for its subtractive hybridization from the sample through the use of magnets. The end product would be a sample containing only bacterial DNA and no plastid DNA.

32 33

Figure 4.1: Artistic representation of paramagnetic bead attached to phosphorus dendrimer attached to oligonucleotide probe.

Ideally the sample to be used for the subtractive hybridization would be the genomic DNA extracted from each plant. However due to the large size of the genomic DNA and its circular structure, it would create steric hindrance that would affect the hybridization between the probes and the area of interest in the plastid. For this reason, the samples to be used would be PCR amplicons with the original sample as the template, generated using primers 27F and 1492R. This would result in DNA of small enough size that it would not affect the rate of hybridization. Due to PCR involving the exponential amplification of DNA and the majority of the initial template consist of plastid DNA, the number of cycles for the PCR amplification has to be minimized. This number of cycles has to allow for the amplification of the bacterial DNA, while not allowing for too much exponential amplification of plastid DNA.

4.2 Methods 4.2.1 Designing Plastid Specific Probes

In order to generate a probe specifically targeting only plastid sequences while leaving bacterial sequences alone, 85 plastid 16S rRNA sequences from different plant orders were downloaded from the NCBI database along with 10 bacterial 16S rRNA sequences from different phyla. The sequences were stored in the program Geneious (Drummond et al., 2012) and aligned using the CLUSTALW alignment under the default settings. The alignment was checked for regions of conservation within the plastid sequences that differed from bacterial sequences. An in silico analysis on the chosen conserved region was conducted in the RDP database in order to check if it would bind to any bacterial 16S rDNA.

4.2.2 Binding of Magnetic Beads to Phosphorus Dendrimers

The following protocols follow those established by Archer et al. (2006) with some modifications. Paramagnetic beads, 1.0-3.0 mm in size with amino functionalities were purchased from SPHEROTM (Spherotech, IL). Before the beads could be bound to the phosphorus dendrimers solubilized in dichloromethane, a solvent exchange was carried out. Aliquots of 500 µg beads were initially washed 5X with 1 mL of sterile deionized water before they were resuspended in 1 mL of 100% ethanol and incubated at room temperature for 8 hours with rotation. The ethanol served as the transition solvent between the water and the dichloromethane. After the incubation the beads were resuspended in 1 mL of dichloromethane and incubated for 18 hours in a room temperature rotator and the solvent was refreshed every 4 hours. The beads were then washed with 1mL of 100% ethanol, 1 mL of dichloromethane and 1 mL of 100% ethanol before they were transferred to new 0.5 mL tubes. The beads were resuspended in 1 mL of 2% (w/v) phosphorus dendrimer (cyclotriphosphazene-PMMH-12 dendrimer, generation 1.5, Sigma-Aldrich, MO, USA) in dichloromethane, followed by incubation for 8 hours at room temperature in a rotator. Following the immobilization of the phosphorus dendrimers the beads were washed 4X with 1mL of dichloromethane, 3X with 1 mL 100% ethanol before realiquoted into two 0.5 mL tubes. The ethanol supernatant was removed and the beads were soft baked at 55˚C for 15 minutes before they were stored at room temperature.

4.2.3 Immobilization of Oligonucleotide Probes

The probe that I designed and its reverse complement sequence were ordered with a 5’ amine and C12 modification from Biobasic (Ontario, Canada). The 5’ amine allows for the oligonucleotide to bind onto the phosphorus dendrimers, while the C12 modification reduces the steric hindrance during the hybridization of the DNA to the probe. Before the oligonucleotides could be immobilized on to the magnetic beads, the beads were washed in 750 µL of 0.3M sodium phosphate buffer (pH = 9) and incubated in this buffer overnight at room temperature in the rotator. The buffer was refreshed 3X during the incubation period. After the wash, the supernatant was removed and the beads were transferred to new 0.5 mL tubes and incubated with either 450 ng of forward or reverse oligonucleotide probe in 150 µL of sodium phosphate buffer for 8 hours at room temperature in the rotator. The supernatant was collected in order to determine the effectiveness of the binding of the probe. Briefly the concentration of the probes in the supernatant was measured through the use of a NanoDrop 1000 Spectrophotometer (Thermoscientific) and the concentration was compared to the original concentration of probes before the incubation and to control samples prepared with no beads.

The beads were then washed 3X with 150 µL of sodium phosphate buffer before an imine reduction was conducted to prevent any additional unnecessary binding on the phosphorus dendrimers. This was carried out by incubating the beads with a solution of sodium borohydride (Sigma-Aldrich, MO, USA) (12.5 mg NaBH4 in 15 ml of phosphate-buffered saline, pH 7.0 and 5 mL of 100% ethanol) for 15 minutes at room temperature in a rotator. The beads were then washed 3X in 0.2% SDS solution for 1 min and 3X in nuclease free sterile water before transferred to new 0.5 mL tubes. These beads were stored in 0.5 mL of 1:10 Tris-EDTA buffer at 4˚C until further use.

4.2.4 Subtractive Hybridization

Figure 4.2: Simplified schematic illustration of bacterial rDNA separation from plant plastid rDNA.

A brief simplified outline of the protocol for the subtractive hybridization is shown in Figure 4.2. To prepare test materials, primers 27F and 1492R were used to amplify ribosomal fragments from a 1:1 mixture of a known plastid DNA (Arabidopsis thaliana, obtained from Dr. Zhao’s lab, UTSC) and known bacterial DNA (Burkholderia sp. R172). The PCR reaction was carried out following the protocols and conditions described in Chapter 2.1.4. with changes to the reaction cycles from 35 cycles to 5, 10 and 15 reaction cycles. The amplicons were purified using the GenElute PCR clean up kit (Sigma-Aldrich, MO, USA) and the DNA concentration of the purified amplicons was determined using a NanoDrop 1000 UV/VIS Spectrophotometer (Thermoscientific).

Figure 4.3: Flow chart simplifying subtractive hybridization protocol.

A simplified flow chart illustrating the following protocol for subtractive hybridization is shown in Figure 4.3. Before the subtractive hybridization was carried out the magnetic beads with the immobilized forward probes were washed with 150 µL of hybridization buffer (5X SSC, 1%SDS) prewarmed to 70˚C. After the wash, the beads were mixed with ~200 ng of PCR product and hybridization buffer was added to a final volume of 150 µL in a 0.5 mL tubes. The beads were incubated in a 95˚C water bath for 5 minutes before it was placed in a hybridization oven set at 80˚C with the rotator set at 1 for 1 hour and 15 minutes. While the hybridization of the forward probes was taking place, the reverse probes were prepared for hybridization following the previously mentioned protocol. After the incubation with the forward probes, the magnetic beads were separated from the supernatant by placing the tubes in a magnetic stand. This temporarily adhered the beads to the tube wall, while allowing for the supernatant to be aspirated. The supernatant was transferred to the tubes containing the already washed reverse probes and subjected to the same hybridization conditions as previously mentioned. After the hybridization with the reverse probe, the supernatant was removed through the use of the magnetic stand and this was labeled the stripped sample, which should ideally have a reduced amount of plastid rDNA. After the hybridization, the magnetic beads were subjected to a wash that involved washing with 200 µL of wash solution 1 (0.1XSSC/0.3%SDS), followed by 2 washes with 200 µl of wash solution 2 (0.1XSSC/0.03%SDS). Following the wash, the magnetic beads were stripped of the hybridized amplicons through the following steps: 2 washes with 200 µL of stripping buffer (1XSSC/ 1%SDS) and 2 washes with 200 µL of PCR water all in 95˚C water bath for 5 minutes. The magnetic beads were stored then stored at 4˚C in 10:1 Tris-EDTA buffer. The supernatant from the stripping step was collected into 1.5 mL microtubes and labeled as hybridized samples, which ideally should only contain plastid rDNA.

In order for the stripped and the hybridized samples to be used for molecular analysis, the SDS in the samples had to be removed. This was done through the use of the GenElute PCR clean up kit (Sigma-Aldrich, MO, USA) with a couple of modifications. The samples were subjected to an additional ethanol wash and they were eluted in 20 uL of elution buffer instead of 50 uL. The purified samples were amplified using primers 341FGC and 907R using PCR reaction volumes and conditions previously mentioned in Chapter 2.1.5, before they were run in a 6% polyacrylamide DGGE gel with a 40%-70% denaturing solution gradient for 20 hours in 0.5X TAE buffer at 70V and visualized after stained with ethidium bromide.

4.3 Results 4.3.1 Plastid Specific Probe

Based on the alignment of the plastid and bacterial 16 rRNA sequences and the search for a conserved region, the chosen sequence for the plastid specific probe was ’5- CAGAGGATGCAAGCGTTATCCGGAATGATTGGGCGTAAAGCGTCTGTAGGTGGCTT TTCA- 3’. The chosen probe was checked using "probe match" applet in the RDP database, to determine how many bacterial sequences it would bind onto. It was determined that this probe did not have any bacterial sequence hits but targeted 32% of the land plant plastid sequences. This indicated that the chosen probe had the potential to only target plastid rRNA while avoiding bacterial rRNA. The chosen sequence was made into the forward probe and the reverse complement of this forward sequence became the reverse oligonucleotide probe.

4.3.2 Subtractive Hybridization

Ideally to carry out a successful hybridization the size of the target DNA should be smaller than genomic DNA in order to minimize steric hindrance and allow for a more successful hybridization of the sample to the probe. For this reason the sample to be hybridized should be digested into smaller pieces or consist of PCR amplicons. Initially I sought a restriction enzyme that could digest the DNA into smaller pieces without targeting the 16S rDNA. Two restriction enzymes, SwaI and HpaI, were tested in silico and were found to be theoretically suitable; they cut plastid and bacterial DNA only at sites outside16S rRNA sequences. Plant genomic samples along with known bacteria strain and Arabidopsis thaliana plastid samples were digested with these enzymes. Subsequent use of the digested genomic DNA in the subtractive hybridization protocol resulted in no changes in the detected bacterial or plastid DNA concentration in the stripped samples and there was no hybridization of the plastid DNA to the probes.

Therefore subtractive hybridization was carried out on purified PCR amplicons generated through the use of mixed template containing a ratio of 1:1 known plastid rDNA mixed with known bacterial rDNA. The PCR reactions were carried out with varied number of cycle numbers ranging from 5 to 15 cycles, (subsequently referred to as 5 cycle, 10 cycle and 15 cycle PCR product) to test the minimum number of cycles required for the bacterial rRNA to be detected, while minimizing the amount of plastid rRNA being amplified. There were 2 resultant products from the subtractive hybridization protocol: 1) the stripped sample and 2) the hybridized sample. The stripped sample represented the sample that had undergone the subtractive hybridization and should have a decreased amount of plastid rDNA. The hybridized

41 samples represented the amplicons that had bound to the probes and were removed from the sample. There were 2 hybridized samples as one corresponded the DNA removed from forward probes and the other from the reverse probes. The 2 oligonucleotide probes were bound to separate magnetic beads and kept separate from each other during the hybridization to avoid complementary binding between the 2 probes.

In order to visualize the effectiveness of the protocol with each samples, DGGE analysis of PCR products with primers 341FGC-907R was carried out - as seen in Figure 4.4. When the samples were run in a DGGE gel, bacterial rDNA had a lower GC content and lower mobility (top band) while plastid rDNA had higher GC content and a higher mobility (bottom band). The control samples in the subtractive hybridization consists of just the PCR amplicons mixed with the hybridization buffer without the presence of the beads; representing the extent to which bacterial and plastid rDNA are detected in the original sample. There were slightly different results when the subtractive hybridization was carried out using the different samples. With all the trials, the bacterial and plastid amplicons were still present in the stripped samples and in comparison to the control sample, resulted in fainter bands in the DGGE, indicative of a decreased concentration of the initial amplicons. With the 15 cycle PCR product, the hybridized samples only contained plastid amplicons as seen by the presence of the lower band. However as the cycles decreased, there was an appearance of a very faint upper band corresponding to the bacteria used in the test sample. The faint presence of the bacterial bands in the 5 cycle and the 10 cycle hybridized samples indicated that the probes are mainly binding to the plastid rDNA with some minimal binding to the bacterial DNA.

a) b)

4.4 Discussion

Using this method, the probe designed could have been chosen to target either bacterial 16S rDNA or plastid 16S rDNA. When various plastid 16S rRNA sequences from other land plants were accrued and aligned together with bacterial 16S rRNA sequences, there were more regions of conservation found solely amongst the plastid sequences than solely amongst the bacterial sequences. There were many regions of conservation that were shared amongst both plastid and bacterial sequences. This meant that it would be more feasible to design a probe that would specifically target plastid rRNA instead of bacterial rRNA. With subtractive hybridization, Zammatteo et al. (1997), utilized capture probes ranging in length from 56-255 bp immobilized on magnetic microparticles to capture a 435 bp target and found that longer probes correlated with higher immobilization efficiency and had the best subtraction efficiency. Based on this knowledge, the sequence of the probe chosen from a region that would result in a probe at least 56 bp in size. The sequences for the plastid excluding primers were not used as probes given that they were only between 19-21 bp in size and they could not be further extended in size as the

43 region adjacent to the primers were either conserved regions found in both plastid and bacterial rRNA or they were too variable between bacteria. Based on the alignment, within the actual probe sequence there were some positions in the sequence where the bases differed between certain plant plastid rRNA sequences. Ideally in order to account for the specific differences without creating a probe that would be too degenerate and also target bacterial sequences, various versions of the probe that accounted for the 2-3 different bases in the sequence would have needed to be tested and ordered. However since this was just the testing phase and due to limited resources, only one probe sequence was chosen, the one that seemed to bind to the highest number of plastid sequences as possible. This explains why only 32% of the land plant plastid sequences were targeted in the database. The forward probe and the reverse probe (the reverse complement of the forward probe) ordered for the chosen sequence were bound to magnetic beads and kept separate from each other to avoid complementary binding between the 2 probes. Given that the probes were attached to the beads via a phosphorus dendrimers and C12 modification to avoid steric hindrance, it is possible that the probes would have been far enough from the paramagnetic bead that it could have potentially allowed for the bound probe to bind to its reverse complement probe if it was found in the same solution. If the probes were found to be individually successful the next step would be testing if the 2 probes could be mixed in the same hybridization solution, reducing the amount of time required to carry out the hybridization protocol.

For the starting product to be used for the subtractive hybridization random shearing of the genomic DNA was not taken into consideration, as it would disrupt the16S rDNA gene and the extent of the disruption would be uncertain. Ultimately it was decided that in order for the subtractive hybridization to be tested PCR amplicons would be used as the initial sample. Since PCR exponentially increases the initial template with every cycle, the number of cycles used in the PCR reaction to generate the amplicons for hybridization is critical. Initially, by starting off with a small number of PCR cycles in the overall reaction, this would allow for the amplification of both bacterial DNA and plastid DNA. However as the number of cycles increased, exponentially more plastid amplicons would be created per cycle and substantially less bacterial amplicons due to their initial concentration. Therefore in a 35 cycle PCR reaction, the plastid amplicons becomes overly dominant while the bacterial amplicons become more rare. For this reason the hybridization was tested out using PCR amplicons created using 5

44 cycle, 10 cycle and 15 cycle PCR reactions. Along with the presence of plastid rDNA, bacterial rDNA was amplified from the 5 cycle and 10 cycle hybridized samples. The faint presence of the bacterial bands in comparison to the plastid bands indicate that the probes are mainly binding to the plastid rDNA with some minimal binding to the bacterial rDNA. It could be that this was not seen with the 15 cycle product due to the overwhelming presence of plastid amplicons, preventing the binding of the bacterial DNA to the probes. There could be a couple of reasons as to why the probe was binding to the bacterial DNA, which could include the temperature at which the hybridization was carried out at or maybe the hybridization buffer might need to be changed to increase the stringency. Ideally, if time permitted, hybridization conditions (temperature, buffer composition) could have been optimizing using solely plastid and bacterial DNA separately. This is one of my recommendations for future work (see Chapter 8).

With all the samples used, there was no dramatic decrease in intensity in the plastid band in the obtained stripped sample, where the intensity corresponds to the concentration of plastid product amplified. This could indicate that the ratio of magnetic beads/probes to plastid rDNA may have been too low. If this was the case, two solutions are valid - the addition of more beads/probes to remove more plastid rDNA; or a reduction in the number of PCR cycles to reduce the dominance of plastid rDNA in sample. Reduced PCR cycles were attempted but results suggest that the probes were binding bacterial amplicons as well as those of plastids. Although the intensity could be used as a measure of the concentration product present, it has to be taken into account that band intensity is only an effective measure for certain concentration range of amplicons and anything above this range, the intensity stays relatively the same. This does mean that there is still a possibility that the concentration might have decreased between the different samples but due to the final concentration it was hard to distinguish by just using the intensity.

Although only a probe specific to plastid DNA was mentioned, a probe that would specifically target mitochondrial SSU rRNA was also designed. This probe was selected after alignment of known plant mitochondrial sequences in the NCBI database and checked in the RDP database to ensure it would not bind to any bacterial sequence. Although it was only 26 bp in size, this was the longest conserved region found amongst the plant mitochondrial sequences. The sequence of the probe was 5’-AGCCGAGTGACGTGCCAGCGCTACTA-3’ and due to time restraints it was not ordered nor tested on the plant samples.

Even though this is one of the post-extraction methods proposed to decrease the amount of the plastid interference in a sample, more time is needed to test out the protocol and fully establish that it works. There is potential for this method to work but both graduate program time constraints and both the laborious nature and the expense of the method led me to propose a second post-extraction method that involves simply enzymatic digestion of plant DNA prior to amplification of ribosomal genes. The testing of this novel method is described in the next chapter.

Chapter 5 Enzymatic Digestion of Genomic DNA samples 5.1 Introduction

As previously mentioned, there have been studies where universal 16S primers have been used to amplify both plastid and bacterial 16S rDNA sequence, but the amplicons were later differentiated through digestion with the restriction enzyme PvuII (Sessitsch et al., 2002, Jiao et al., 2006 and Wang et al., 2008). This enzyme has a restriction site of 5’-CAGCTG-3’ in plastid 16S rDNA sequence that according to Sessitsch et al. (2002) and Wang et al. (2008) are not found in most eubacterial 16S rDNA genes. If plastid PCR amplicons created using the primers 27F and 1492R are cut with PvuII, products of ~1317 bp and ~131 bp result with the restriction cut occurring downstream of the 27F primer binding site. The same enzyme does not cut bacterial amplicons leaving them at ~1500 bp in size. These amplicons of different phylogenetic origin can be resolved and separated on a gel, and this fact forms the basis of work cited above.

This method seems to be a viable option to removing the plastid DNA contamination, as it is easy to differentiate between the amplicons of plastid origin and those of bacterial origin. The problem with using this method is the fact that the majority of the amplicons that are produced using primers 27F and 1492R are of plastid origin due to the initial dominance of plastid DNA in the sample and its increasing dominance with amplification. Therefore if the genomic plant DNA is used as the template for the amplification using the full 16S primers, and digested later, there would be a very faint to no band present at 1500 bp but a very strong band present of ~1300bp. Therefore, to prevent this problem, I conceived of the idea of predigesting the genomic DNA with PvuII before the PCR amplification with the 16S primers - this should allow for the sole amplification of the bacterial DNA. However as seen in chapter 2, the enzyme PvuII does not solely target plastid DNA but also has restriction sites found in some bacterial DNA (i.e. Pantoea sp.). This does mean that some bacterial species might be under represented in the sample due to fact that they were removed by digestion. For this reason, further research was conducted to determine if there was another enzyme that could be used to differentiate plastid DNA from bacterial DNA.

I researched into other restriction enzymes that could be used to digest plastid 16S rRNA and found that the restriction site for MscI, 5’-TGGCCA-3’ was found in many plastids 16S rRNA

46 47 sequences but few bacteria. The restriction site for this enzyme is typically located upstream of the 1492R primer binding site in plastid 16S rDNA and the digest by this enzyme results in a product that is ~1372 bp and ~119 bp in most plastids. I hypothesized that while some bacteria might also have ribosomal genes with MscI cut sites, they would differ from bacteria with PvuII sites. Hence taking genomic DNA and digesting half the sample with PvuII and the other half with MscI, neither half would result in plastid amplicons but each half would allow for amplification of intact bacterial genes digested in the other. Therefore when the universal full 16S primers are used in a PCR reaction with the mixed digested genomic DNA as the template, the amplicons of ~1500 bp would all be of bacterial origin. This would allow for the better detection of the of the bacterial sequences found amongst the plastid – given that it is hard for the primers to detect the bacterial DNA with the amount of plastid DNA present. The proposed method was tested to see if it would work experimentally.

5.2 Methods 5.2.1 In silico Analysis of Restriction Enzymes

In order to determine how many bacterial sequences actually had restriction sites for the restriction enzymes PvuII and MscI, an in silico analysis was carried out. The restriction sites for the enzymes were tested using the applet "probe match" the RDP database and only the sequences classified as good quality and >1200bp were examined.

5.2.2 Enzymatic Digest of DNA and Amplification of Bacterial DNA

Using the previously extracted Acer negundo and Ulmus spp. DNA from Summer 2012, 10 µl samples of genomic DNA were individually digested with the PvuII and MscI restriction enzymes obtained from NEB (Canada). The samples were digested for 3 hours at 37˚C, followed by incubation at 80˚C for 20 minutes to inactivate the enzymes. One microliter of the digested products were used as templates for PCR reactions using primers 27F and 1492R following the previously mentioned reaction volume and conditions. The PCR products were visualized in 1.5% agarose gels and bands of 1500 bp in size were excised and gel purified with QIAEX II Gel Extraction Kits (Qiagen, Canada).

5.2.3 Amplification using DGGE primers

The gel-purified products were used as templates for the following PCR reaction using primers 341FGC and MOD783R (a modification of primers by Sakai et al., (2004), equimolar concentrations of primer 783RA and primer 783RC). The PCR reaction followed the same reaction volumes and PCR conditions mentioned in Chapter 2. Briefly, 20 µL PCR reactions were carried out in a PTC-200 thermal cycler (MJ Research Inc.) with the following conditions: initial denaturing at 95˚C for 5 minutes followed by 35 cycles of: denaturing at 95˚C for 1 min, annealing at 56˚C for 1 min and extension at 72˚C for 1 min; final extension at 72˚C for 10 mins. The PCR products were checked on 1% agarose gels before the remainder of the PCR products was run in a DGGE gel. The DGGE gel was a 6% polyacrylamide gel consisting of a 40%-70% denaturing solution gradient and it was run in a DGGE-2001 Tank (C.B.S. Scientific Co, Del Mar, California) with 0.5 X Tris-acetate-EDTA buffer, for 20 hours at 70V and 58˚C. The gel was stained in ethidium bromide for 30 minutes before it was visualized under UV light.

5.3 Results and Discussion 5.3.1 In silico Analysis of Restriction Enzymes

The in silico analysis using the bacterial sequences found in the RDP database (http://rdp.cme.msu.edu), demonstrated that these individual enzymes actually have restriction sites present in some bacterial 16S rRNA sequences. The output from this analysis with the number of bacterial hits and the phyla targeted can be found in Appendix A, Table A.1. The results are summarized in Table 5.1. Details of which organisms were found to have restriction sites for the each individual enzyme and both enzymes combined are listed on Appendix A, Table A.2. Restriction sites for both enzymes are found in some important genera known to be endophytes. For instance PvuII sites are found in Methylobacterium spp., Sphingomonas spp., Rhodobacter spp., Bacillus spp. and Rhizobium sp. MscI restriction sites are found in Rhodococcus sp., Kineococcus sp., Clostridium sp., Flavobacterium sp., Curtobacterium sp., and Pseudomonas sp.. As summarized in Table 5.1, there are more bacterial sequences with restriction sites for the enzyme MscI (7.28%) than those with PvuII sites (5.3%). However, there are actually very few bacterial strains that share both restriction enzyme sites (0.71%). Due to this fact, even if the bacterial DNA is digested by 1 enzyme, there is a very low possibility that the other enzyme also digests it. Thus if a sample was split and individually digested with each

49 enzyme then the subsamples were combined together, the bacterial strains that are digested in the PvuII sample are likely to be intact in the MscI sample and vice versa. The plastid on the other hand would be digested by both of these restriction enzymes, thus it would not be amplified in a PCR reaction using the full 16S rRNA primers.

Table 5.1: Summary of in silico analysis of bacterial and land plant plastid sequences in the RDP database with restriction sites for enzymes PvuII and MscI.

Enzyme PvuII MscI PvuII and MscI together Total Bacteria 62244 (5.3%) 84757 (7.22%) 8346 (0.71%) Sequence Hits (out of 1173481) (%) Total land plant 1501 (83.3%) 469 (26.0%) 375 (20.8%) plastid hits (out of 1801) (%)

5.3.2 Experimental Analysis of the Restriction Enzymes on Plant Genomic DNA

To determine if the use of the restriction enzymes would result in the improved amplification of bacterial DNA, they were tested on DNA extracted from Acer negundo, and Ulmus spp. plant samples. These samples were individually digested with MscI and PvuII, then amplified with the full length 16S primers and gel purified, before amplification with primer pair 341FGC- MOD783R to be visualized in a DGGE gel as seen in Figure 5.1.

In Figure 5.1, lanes 1, 4 and 7, correspond to original samples that were used as template for PCR reaction using the plastid excluding primers. It can be seen that there are problems in how well the plastid excluding primers are able to amplify bacterial DNA present in the sample with little to no amplification of bacteria from untreated Acer negundo and Ulmus pumila DNA samples, but there is good amplification of bacterial sequences from the Ulmus parvifolia DNA. All of these samples were extracted with the same DNA extraction kit and the same concentration of DNA, in the range of 20-30 ng/µL, was used as template in the PCR reaction. Previous PCR reactions using the same DNA as template and primers 27F and 1492R were successful in amplifying chloroplast DNA from the sample (results not shown). This shows that

51 the extent to which plant organelle inhibits the amplification of bacterial DNA differs for what might be assumed to be similar samples. It also shows, looking at the results from Ulmus parvifolia, that the digested samples give very similar profiles to the original DNA with a few more bands present in the digested samples than the original sample. This demonstrates that the bacteria present in the sample can still be detected even after the digestion with the restriction enzyme and it does not differ much from the microbial community found in the sample. The reason for the apparent lack of interference from plastid DNA in Ulmus parvifolia samples relative to the samples from the other two tree species is unknown, but could potentially be attributed to more bacterial DNA present in the sample.

A drawback remains with this method in that not all plastids are susceptible to digestion by these enzymes; therefore the susceptibility of the plastid 16S rRNA of the plant being analyzed must be determined before hand. At the same time, the enzymes unfortunately do not target mitochondrial rRNA, thus the mitochondrial interference will still be present. The pre-digestion of the extracted plant DNA with the enzymes PvuII and MscI allows for an elegant and rapid method for vastly improved detection and amplification of bacterial DNA in comparison to "raw” plant tissue DNA but the problem of contamination from the other bacterially derived symbiont remains. Resolutions to this problem will be suggested in Chapter 8.

Regardless this method was applied to the analysis of bacterial community variance in the tree species of interest and compared to the findings based on classical culturable community studies. These results are presented in Chapters 6 and 7.

Chapter 6 Analysis of Variation – Culture Dependent Method 6.1 Introduction

The first objective of this project, to test and optimize the methods to study endophytic community, was carried out in the previous chapters. The following two chapters will focus on the second objective of this project, to study the variation of the bacterial endophytic community between individual trees and tree species. It is key that before any study of endophytes in trees in this site or any other sites are undertaken, an understanding of the extent of the variation that might result due to biotic and abiotic factors is taken into consideration. This will determine how the subsequent sampling and analysis will be carried out, which might include determining when sampling should be conducted and how many samples are necessary to study the endophytes in plants in contaminated sites and phytoremediated sites. This second objective involved testing the following hypotheses using culture dependent and culture independent methods:

1) The endophyte community profiles of trees of different species can be statistically differentiated from each other. i.e. tree endophyte communities are tree specific.

2) The bacterial endophytic composition of a tree is influenced by the seasonal variation such that the endophyte communities are statistically differentiated between seasons.

The following chapter will discuss the culture dependent analysis of the chosen samples. The culture dependent method of analyzing bacterial endophytes involves obtaining the plant macerate and plating it onto agar plates, allowing for the bacterial endophytes to grow. This allows for the subsequent purification of bacterial strains and their identification through Sanger sequencing. However there are biases associated with plating bacteria as a small proportion of bacteria are culturable and not all culturable bacteria grow in the same medium; different bacteria may have different nutrient and growth requirements. Therefore culturing bacteria will only provide an insight into a subset of the bacterial endophytes that can be found in the plants. Although it is only a subset, the actual isolation of bacteria allows for subsequent metabolic and genetic testing and experimentation to be carried out on these endophytes.

For this study the plants to be analyzed were taken from a hydrocarbon-contaminated site. The trees chosen for the study were Acer negundo, Ulmus parvifolia and Ulmus pumila, since they

52 53 were found in close proximity to each other on the contaminated site. Given that stems were easier to access in comparison to the roots of the trees, triplicate branches were selected from each tree species.

6.2 Methods 6.2.1 Sample Collection and Surface Sterilization

Branches of Acer negundo and Ulmus spp. were collected from a hydrocarbon-contaminated site. From each tree, 3 branches of roughly 1.5-2.0 cm in diameter were obtained, and stored at 4˚C for no more than 1 week after collection, before the branches were subjected to surface sterilization procedures as previously mentioned in Chapter 2.1.2. This sterilization involved the use of 70% ethanol wash followed by washes in 1.5% bleach/0.1% Tween 20, 1.5% bleach and rinse in sterile water. The plant samples were collected in February, July and October, representing the seasons Winter, Summer and Autumn respectively. Since the samples were initially collected in the Winter, it was hard to distinguish the branches of the Ulmus parvifolia and Ulmus pumila without the presence of the leaves. The difficulty in differentiating the Ulmus parvifolia tree from the Ulmus pumila tree lead to an uneven number of replicate trees to be collected for each of the Ulmus species in February and July.

Table 6.1: Tree branches collected for analysis.

Tree species Tree Winter Summer Autumn Replicate (February 13/2012, (July 9/2012, (October 23/2012, temperature = - temperature = temperature = 2˚C) 22˚C) 11˚C)

Acer 1 A, B, C E, F, G H, I, J negundo 2 A, B, C E, F, G H, I, J

3 A, B, C E, F, G H, I, J

Ulmus 1 A, B, C E, F, G H, I, J parvifolia 2 A, B, C E, F, G H, I, J

3 A, B, C

Ulmus 1 A, B, C E, F, G H, I, J pumila

2 A, B, C E, F, G

3 A, B, C

6.2.2 Culturable Endophyte Extraction, Isolation and Identification

Isolation of endophytes: Given that the plant macerate generated would be used for both the culture dependent and culture independent analyses, it was key that any epiphytic bacteria, bacteria found on the surface of the plants were removed from the sample. For this reason the periderm of the sterilized branches was removed in order to ensure that bacterial isolates and DNA extracted would belong to true endophytic bacteria and not epiphytic bacteria. The plant tissue was weighed then homogenized in a single speed Waring Blender (20000 rpm) with 60 mL of 50 mM Tris-HCl solution for 1 min. The homogenized plant tissue was filtered through 8 layers of sterile cheesecloth, removing the majority of the plant debris and the liquid macerate was collected. The macerate was centrifuged at 600xg for 5 minutes to remove more plant debris, and then centrifuged at 8000 rpm for 10 minutes to pellet the bacterial cells. The bacterial cells were resuspended in 1 mL of 50 mM Tris-HCl solution, from which a 1/5 dilution was made and 100 µL was plated onto replicate Reasoner’s 2A (R2A) and Tryptic Soy Agar

(TSA) media plates. The remainder of the macerate was stored at -20˚C for later use. The media plates were incubated at 28˚C for a period of 7 days to 1 month. Unique individual bacterial colonies were identified and their individual counts were recorded for each branch processed. The unique bacterial colonies were isolated and purified by restreaking on R2A or TSA.

Identification of isolates: Lysates were created using individual pure colonies added into 100 µL of sterile water and boiled for 8 minutes. One microliter of the boiled lysates were used as templates for amplification of ribosomal 16S gene fragments using primers 27F and 1492R. The PCR amplifications were carried out in a PTC-200 thermal cycler (MJ Research Inc.) with the following conditions: initial denaturing at 95˚C for 5 minutes followed by 35 cycles of: denaturing at 95˚C for 1 min, annealing at 56˚C for 1 min and extension at 72˚C for 1 min; final extension at 72˚C for 10 mins. Once the PCR reactions were ensured to be successful with proper amplification of each sample and no contamination in the negative control, the amplicons were PCR purified using the GenElute PCR clean up kit (Sigma-Aldrich, MO, USA). The purified amplicons were sent to TCAG (Toronto, Canada) for Sanger sequencing. The obtained sequences were submitted to the RDP database and BLAST (NCBI) in order to determine the potential identity of the bacteria. The identity of the bacteria was based on minimum 99% similarity to database 16S rRNA sequences.

6.2.3 Community and Statistical Analysis

All of the following community and statistical analyses were conducted through the use of the functions and packages created for the program R version 2.15.2 (R Core Team, 2012). With each sample, the bacterial counts from each type of media were recorded and converted to colony forming units (CFU) per gram of plant tissue. This data was then converted to % abundance values for subsequent analyses.

After the total bacterial counts, the species richness and the species diversity, calculated using the Shannon index, were determined for each sample using the % abundance data. This was done through the functions specnumber and diversity found in the ‘vegan’ package (Oksanen et al., 2012). The homoscedascity and the normality of each variable was checked through the use of Levene’s test with the function levene.test in the ‘lawstat’ package (Noguchi et al., 2009) and Shapiro’s test with the function shapiro.test in the ‘stats’ package (R Core Team, 2012). If the data was determined to be homogeneous and normally distributed, ANOVA (analysis of

56 variation) tests using the function aov in the package ‘stats’ were used to test for significant differences in bacterial counts, richness or diversities between tree species and seasons. If differences tested by ANOVA were significant (p < 0.05) further post hoc analysis was conducted using Scheffe’s test using scheffe.test in the package ‘agricolae’ (De Mendiburu, 2012). However if the data was not homogeneous and/or not normally distributed, it was transformed and its homoscedascity and normality was retested. If the data was still not homogeneous and/or normally distributed, Kruskal-Wallis test, a nonparametric ANOVA test, was conducted using function kruskal.test in the ‘stats’ package. If the results from this test indicated significant differences (p < 0.05), a subsequent Wilcoxon signed-rank test was conducted to ensure that the comparison conducted was significant using the function pairwise.wilcox.test in the ‘stats’ package and to identify where the differences were found.

The culturable endophytic community compositions were compared between trees and seasons using non-metric multidimensional scaling (NMDS) analysis of the species abundance (counts) data. NMDS allows for the ordination of samples that are similar to each other, with the generated stress value representing the goodness of fit of the final plot. Stress values >0.2 indicates the plot is close to random, stress values <0.2 and >0.1 indicates a useful 2 dimensional picture and plots with a stress value <0.1 correspond to an ideal ordination (Rees et al., 2004). The NMDS plots were created through the use of the function metaMDS in the ‘vegan’ package, with the number of dimensions set to 2 and the maximum number of random starts set to 20 (Oksanen et al., 2012). In these analyses the data was converted to a dissimilarity matrix using the Jaccard index (function "dist") that only took into consideration presence or absence of the bacterial genera in the branch samples. This removed any errors that might have occurred in the tallying of the abundance of each unique bacterial colony and at the same time allowed for the focus to be solely on the members of the community and not their abundance.

Hierarchical agglomerative clustering analysis was also conducted on the datasets in order to visualize the similarities of the endophytic community structures of the samples. This was done through function hclust in the ‘stats’ package, with the Ward agglomeration method using the same Jaccard distance matrix previously generated. The pvclust function in the package ‘pvclust’ (Suzuki & Shimodaira, 2011) was then used on the results from the hierarchical clustering. This function assessed the uncertainty of the hierarchical clustering through the generation of AU (approximately unbiased) p-values and BP (bootstrap probability) values

57 generated using multiscale and normal bootstrap resampling (Suzuki & Shimodaira, 2011). These values indicate how well the clusters or edges are supported by the data, with the AU value considered by the creators of the program to be more accurate than the BP value.

The last analysis conducted was a permutation multivariate analysis of variance between the endophytic bacterial community structure of each sample using the function adonis in the ‘vegan’ package, using Jaccard pairwise distances, and permutations set to 999. This revealed any significant differences between the community structure of the samples with respect to season or tree species. However if the p-value was significant, the outputs generated by this function could not be subjected to further post-hoc analysis. For this reason, in cases where there was significance found between the factors (season or plant species), the levels (either the 3 seasons or the 3 plant species) were individually compared to each other and the comparison was considered significant if p-value <0.017 representing a Bonferroni correction on the p-value of the original analysis done by adonis.

6.3 Results 6.3.1 Bacterial Densities, Richness and Diversity

The total number of culturable bacteria obtained from each media and each season per plant species ranged from 103 to 107 colony-forming units (cfu) per gram of tissue. The average cfu g- 1 of tissue found in each tree species per seasons are illustrated in Figure 6.1. The two different media used to culture the endophytes, R2A and TSA, resulted in similar average cfu g-1 of tissue. As seen in Figure 6.1, higher culturable bacterial counts were obtained in the Autumn in comparison to the other seasons for all the plant species.

a) b)

Figure 6.1: Average culturable bacteria counts obtained per gram of tissue from a) R2A media and b) TSA media per plant species in each season.

Once the number of culturable bacteria obtained from each bacteria genus was converted to % abundance data, the species richness and diversity was determined. The average species richness and diversity calculated for each plant species per seasons are shown in Figure 6.2. Unlike the total bacterial counts, there were differences between the two media with higher species richness and species diversity found in the R2A in comparison to TSA media.

a) b)

c) d)

Statistical analyses were conducted on the total culturable counts, species richness and species diversity and show in Table 6.2. There were significant difference found in the total culturable counts between the samples based on the season they were collected from, on both TSA (Kruskal-Wallis χ2 = 18.9024, p < 0.05) and R2A media (Kruskal-Wallis χ2 = 26.3108, p < 0.05). Subsequent pair wise comparison testing revealed that the total counts values obtained in the Autumn were significantly higher than those obtained in the Winter and Summer for both media. However, there were no significant differences detected between the total counts with respect to the plant species. The statistical tests on the species richness and diversity index also revealed no significant difference in these values based on season the plants were collected from and based on the plant species they belonged to.

Table 6.2: Result of statistical analysis conducted on the total bacterial counts, species richness and Shannon diversity index for each media with the comparisons based on either plant species or season collected. Data Media Comparison between Comparison between Plant Species Seasons F-Value or P-value F-Value or P-value +Kruskal- +Kruskal- Wallis chi- Wallis chi- squared squared Total R2A +2.58 0.27 +26.31 <0.05 Bacterial TSA +1.78 0.41 +18.90 <0.05 Counts Species R2A 1.75 0.18 0.96 0.81 Richness TSA 0.55 0.75 1.09 0.58 Shannon- R2A 0.31 0.74 0.52 0.60 Diversity TSA 0.22 0.81 0.14 0.87 Index

6.3.2 Identification of Bacterial Strains

The use of two different media resulted in the isolation of different subsets of the culturable bacteria - 30% of the bacteria types isolated were obtained only on R2A, 32% appeared only on TSA and 38% on both media. All the bacteria that could be isolated, for which I could obtain good sequence data are shown in Table 6.3. The list excludes a number of bacteria types that initially grew on spread plates from the plant macerates but could not be further isolated and purified or sequenced, which I estimate was the case for 30% of the bacteria types, but accounted only between 6% and 13% of total isolated bacteria from R2A and TSA respectively. Regardless, the unidentified bacteria were still recorded and added to the data under a code name instead of a bacterial name and treated as bacterial genera so their presence was accounted for in the density, richness and diversity data. A total of 31 bacteria genera were cultured from Acer negundo and Ulmus spp. combined, which have been previously identified as bacterial endophytes and soil bacteria. The specific bacterial genera isolated from each plant species in each season are shown through Venn diagrams in Figures 6.3-6.5. There were some bacteria

61 genera that were only cultured from certain plant species such as Variovorax spp. and Amnibacterium spp. isolated only from Acer negundo and Rhizobium spp. and Rathayibacter spp. isolated only from Ulmus parvifolia. Typically similar bacteria genera were isolated from the 3 plant species, with differences in the season from which they were isolated. At the same time there were some bacterial genera that were isolated in all the plant species, throughout all 3 seasons, such as: Bacillus spp., Curtobacterium spp., Frigoribacterium spp. Methylobacterium spp., Paenibacillus and Sphingomonas spp..

All of the bacteria genera identified belonged to one of the following phylum: Actinobacteria, Firmicutes, Bacteroidetes, Deinococcus-Thermus and Proteobacteria. Under the phylum Proteobacteria, bacteria specifically belonging to the class of Alphaproteobacteria, Betaproteobacteria and Gammaproteobacteria were found. The abundance of each bacterial phylum cultured in each media is shown in Figure 6.6 and Figure 6.7. The majority of isolates were from the phylum Actinobacteria, with 61.7% and 62.9% of total bacterial count in R2A media and TSA media respectively. Between the 2 media, similar bacteria phyla were cultured in each media and their abundances were about similar. There were some exceptions such as the bacteria from the phylum Deinococcus-Thermus isolated only in R2A but not TSA, and with regards to the Proteobacteria, a higher percentage of Alphaproteobacteria was isolated in R2A whereas a higher percentage of Gammaproteobacteria was isolated in TSA. There were also more unknown bacteria, bacteria that could not be further isolated, purified and sequenced, isolated using TSA in comparison to R2A.

For each media, the total endophytic bacteria community obtained in each season and the general endophytic community isolated in each plant species based on bacteria phyla are shown in Figures 6.8 and 6.9. The actual percentages for the abundance of each bacterial phylum in each season or plant species are shown in Appendix C, Table C.1 and C.2. Looking at the abundance of bacterial phyla in each season, there are noticeable changes in the cultured endophytes such as the drastic decrease of Firmicutes and Gammaproteobacteria isolated from Winter into Summer and Autumn and an increase of Bacterioidetes bacteria cultured in the Autumn. With regards to the general endophytic community obtained from each plant species, similar bacterial phyla and class were obtained from Acer negundo and Ulmus spp. with Deinococcus-Thermus bacteria cultured only from Ulmus parvifolia.

Bacterial Phyla Bacterial Genera Actinobacteria Agrococcus sp. Kocuria sp. Amnibacterium sp. T, A Microbacterium spp. Arthrobacter sp. Nocardioides sp. Brevibacterium sp. Patulibacter sp.* Curtobacterium sp. Plantibacter sp. Friedmanniella spp. Pseudoclavibacter spp. T Frigoribacterium spp. Rathayibacter sp. B Geodermatophilus sp. Rhizobium sp. T, B Kineococcus sp. Sanguibacter spp.

Firmicutes Bacillus spp. Paenibacillus spp. Staphylococcus spp.

Bacteroidetes Chryseobacterium sp.

Deinococcus-Thermus Deinococcus sp.R

Proteobacteria Alphaproteobacteria Mesorhizobium sp.R Methylobacterium spp. Paracoccus sp. Sphingomonas spp. Betaproteobacteria Variovorax sp. R, A Gammaproteobacteria Pseudomonas spp. Stenotrophomonas sp. Xanthomonas spp. *Not previously mentioned in literature as endophyte T – Isolated only from TSA media R – Isolated only from R2A media A – Isolated only from Acer negundo B – Isolated only from Ulmus parvifolia

Figure 6.3: Distribution of known bacterial genera cultured from Acer negundo in the Winter, Summer and Autumn 2012.

Figure 6.4: Distribution of known bacterial genera cultured from Ulmus parvifolia in the Winter, Summer and Autumn 2012.

Figure 6.5: Distribution of known bacterial genera cultured from Ulmus pumila in the Winter, Summer and Autumn 2012.

Figure 6.6: Abundance of bacterial phyla and/or class in the total bacterial isolates obtained from R2A media from all 3 plant species and from all 3 seasons.

Figure 6.7: Abundance of bacterial phyla and/or class in the total bacterial isolates obtained from TSA media from all 3 plant species and from all 3 seasons.

Figure 6.8: Culturable bacterial endophyte abundance based on bacterial phyla and/or class in each season, isolated on R2A and TSA media from all 3 plants.

Figure 6.9: General endophytic bacteria community classified based on phyla and/or class, cultured from each plant species on R2A and TSA media.

6.3.3 Comparison of Community Structures Between Samples

The compiled community profile data was analyzed through the use of hierarchical clustering, non-metric multidimensional scaling and a permutation multivariate ANOVA based on dissimilarity matrices generated. The bacterial community structure of the branches was first analyzed by grouping the data based on the season and plant species for each media type as seen in NMDS plots shown in Figures 6.10-6.13 and the hierarchical clustering shown in Appendix D, Figures D.1-D.4, before the entire data sets were analyzed together and shown in Figures 6.14-6.19. Looking at the samples collected from each season, the NMDS plots for the R2A data showed some separation by plant species with Acer negundo samples separating from Ulmus parvifolia samples, but this differentiation was not as obvious in TSA data. In the hierarchical clusters, there was some clustering of Acer negundo and the Ulmus spp. samples. However when the data was separated based on plant genera and analyzed, there were clusters and separation of the Autumn samples versus those from the Summer and Winter in the NMDS plots. In the hierarchical clustering dendrograms created for these data sets there are small groups created based on the season but for the most part the samples from different seasons could not be distinguished from each other.

In the NMDS plots of all the data together (Figures 6.14 to 6.19) there were noticeable groupings of the samples based on season, such that those from the Autumn formed one group and those from Summer and Winter formed the second group. This was not the case with the plant species as there was no clear separation of the Acer negundo samples from the Ulmus spp. samples. In the hierarchical analyses, there were some small clusters of samples that grouped according to the month or to the plant species they belonged to. However these small clusters were found throughout the dendrogram in different branches and there were no clear clustering of all the samples from one month or one species.

In order to determine if the groupings between the samples based on season or plant species were statistically distinguishable, permutation multivariate analyses was carried out on the data. The results for these various multivariate analyses are shown on Tables 6.4 and 6.5. For the R2A isolates there was a significant difference between the communities seen in each season (F- value =3.1552, R2 = 0.0910, p-value < 0.05). There appeared to be a significant difference between R2A communities from different plant species. However upon further analyses where

70 the individual plant species were compared to each other, only the culturable endophytic community of Ulmus pumila differed from those of Ulmus parvifolia and Acer negundo. For the TSA isolates, there were only significant differences between the different sampling seasons (F- value =2.0497, R2 =0.0611, p-value < 0.05) and no significant differences between the samples of different plant species. This significant difference was attributed to differences of the Winter samples from the other sampling seasons. In both media an interaction was present between the plant species and the seasons, indicating there were significant differences between the endophytic communities of samples collected from the same plant species if the season was taken into consideration and vice versa.

Winter R2A Summer R2A

a) b)

Autumn R2A

Winter TSA Summer TSA

a) b)

Autumn TSA

– Ulmus pumila samples. The stress values for the NMDS plots are as follows: Winter – 0.162, Summer – 0.121 and Autumn – 0.152.

R2A Acer negundo R2A Ulmus spp.

a) b)

Acer negundo TSA Ulmus spp. TSA

a) b)

Figure 6.13: NMDS plot of bacterial community profiles collected from TSA media separated based on plant species: a) Acer negundo and b) Ulmus spp.. The symbols in the 2 dimensional NMDS plots represent different seasons:  – Winter, – Summer and – Autumn. The stress values for the NMDS plots are as follow: Acer negundo – 0.142 and Ulmus spp. – 0.169.

R2A Data

Plant species Season

a) b)

R2A Data – Plant Species

Figure 6.15: Visual analysis of all the bacterial community profiles collected from R2A media. Simplified dendrogram of hierarchical clustering of the samples based on plant species.

R2A Data – Seasons

Figure 6.16: Visual analysis of all the bacterial community profiles collected from R2A media. Simplified dendrogram of hierarchical clustering of the samples based on seasons.

TSA Data

Plant species Season

a) b)

TSA Data– Plant Species

Figure 6.18: Visual analysis of all the bacterial community profiles collected from TSA media. Simplified dendrogram of hierarchical clustering of the samples based on plant species.

TSA Data – Seasons

Figure 6.19: Visual analysis of all the bacterial community profiles collected from TSA media. Simplified dendrogram of hierarchical clustering of the samples based on seasons.

Table 6.4: Permutation multivariate analysis of R2A data conducted using adonis after 999 permutations.

Factor F Model R2 P-value Individual analysis (*=significantly different) Winter* Season 3.16 0.09 <0.05 Summer* Autumn* Acer negundo Plant 1.64 0.05 <0.05 species Ulmus parvifolia Ulmus pumila* Seasons: Plant 1.86 0.10 <0.05 species

Table 6.5: Permutation multivariate analysis of TSA data conducted using adonis after 999 permutations.

Individual analysis Factor F Model R2 P-value (*=significantly different)

Winter * Seasons 2.05 0.06 <0.05 Summer Autumn Plant 1.32 0.04 0.1079 species

Season: Plant 1.38 0.08 <0.05 species

6.4 Discussion

Acer negundo, Ulmus parvifolia and Ulmus pumila are trees that are commonly found in urban settings. The study of the individual endophytes found in these trees and the comparison of the endophytic communities between them will provide an idea of what type of bacteria are found in these trees and whether there are bacteria that are common to all these trees. Understanding if there are significant variations in the community in these trees attributed to the plant species or to the season they are sampled from are key to developing a sampling method to allow for the future research of the endophytes in trees from this site and other contaminated sites.

From the analyses carried out on the collected data statistically significant differences were found in the total number of culturable endophytes isolated and the culturable community structure based on the season the samples were collected. However comparing the samples based on species of trees they belonged to, there was only a significant difference in the community isolated using R2A with Ulmus pumila from that of Acer negundo and Ulmus parvifolia. There were no significant differences in the total number of cultured bacteria among the different tree species. This indicates that care should be taken in choosing the media upon which the endophytic bacteria would be cultured on as this might have an effect on the variability detected in endophytes cultured. It also indicates that any future studies comparing endophytes of different trees or even plant species must take into consideration plant samples from the same season to minimize variability.

In this culture dependent analysis, bacteria that were able to grow on the isolation media after the plant macerate was spread plated onto agar plates were taken into consideration and used for the bacterial endophyte community analyses. The total number of culturable bacteria obtained from each media and each season per plant species ranged from 103 to 107 cfu g-1 of tissue. This range is similar to what has been found by Bacon & Hinton (2006) through their bacterial endophyte literature review where they found that normal endophytic concentration ranged from 102 to 106 cfu g-1 of plant tissue with concentrations as high as 107 cfu g-1 in some cases.

As mentioned between the sampling seasons there were significantly higher total bacterial counts in the Autumn in comparison to the Winter and Summer. At the time of sampling, the outside temperature was as follows: Winter – -2˚C, Summer – 22˚C and Autumn – 11˚C. It would be expected that the warmer temperature experienced in the Summer should be better suited for the growth of bacteria, however similar results were obtained by Pillay & Nowak (1997) who found higher culturable endophytic bacterial populations at 10˚C in comparison to 20˚C and 30˚C in tomato plants. It is possible the higher bacterial densities were due to changes in the growth phase, nutrient and carbohydrate content within the plants that accompany the changing seasons. This has been shown with soybeans where the density of the cultured endophytes was largest in the senescence stage followed by the florescent stage and the smallest during the vegetative stage (Kuklinsky-Sobral et al., 2004). As deciduous trees found in a temperate climate, the trees Acer negundo and Ulmus spp. would be adapted to the different seasons and would be able to withstand the winter conditions. Starting at the beginning of spring

83 and into the summer these trees would undergo flowering, bud flush, leaf expansion and wood formation and end with growth cessation and bud set by the beginning of autumn. In autumn these trees would undergo senescence, cold acclimatization, hardiness and enter dormancy that would last over the entire winter (Jansson & Douglas, 2007). The cold acclimatization and hardening that generally occurs in autumn involves an increase in the soluble sugars (i.e. fructose, glucose, sucrose), proteins, amino acids, organic acids and other substances with a decrease in the tissue water content and starch storage in the plant (Cox & Stushnoff, 2001; Li, Junttila, & Palva, 2004; Renaut, Hoffmann, & Hausman, 2005). These changes allow for the reduced osmotic potential within the trees, lowers the freezing point of the sap and reduces the probability of ice crystal formation in a cell (Hinesley et al., 1992; Santarius, 1973). The high concentrations of sugar in the trees would remain throughout the autumn until early winter, where it would start to decrease and by spring and summer would reach its lower concentrations, as the sugars would be used for flowering, bud flush, leaf expansion and wood formation (Landhausser & Lieffers, 2003). The soluble sugar present in the trees would serve as carbon sources for the bacteria, allowing for their proliferation and growth, such that the cycling of sugar concentration present in the plant would result in changes in their density over the seasons. With the presence of higher soluble sugars in the autumn, this might explain the higher bacterial densities detected in this season over the others. The Winter samples were collected mid-winter, where the concentrations of sugars and nutrients would not be as high as Autumn, thus the total bacterial densities isolated would be lower than Autumn. With the variety of soluble sugars and other ions present in the plant changing between the seasons, this could change the endophytic community that could be supported with different endophytes being able to remain metabolically active based on the resources available. This might explain why a significant difference was found in the culturable endophytic community between seasons with respect to isolates from media. This result is similar to the one found by Mocali et al., (2003) who noted that variations in temperatures associated with different seasons had an effect on the culturable endophytic community of elms.

Since each plant would have different nutritional structure, metabolism, concentrations and varieties of sugars, it would be expected that the culturable endophytes would differ between the different plant species. This has been seen in other cultured based studies of endophytes in agricultural plants whereby the plant genotype and even cultivar affects the colonization of

84 endophytes, such as in cotton plants (Adams & Kloepper, 2002), soybean (Okubo et al., 2009) and peas (Elvira-Recuenco & Van Vuurde, 2000). However, the only significant difference in the culturable communities between the plant species was found in the R2A culturable endophytes, with the endophytic community obtained for Ulmus pumila being significantly different from Acer negundo and Ulmus parvifolia. It is possible that this significant difference could reflect the fact there was a smaller number of replicates present from Ulmus pumila in comparison to the other plant species for the Winter and Summer. It is also possible that the limitations associated with identifying the bacteria along with culturing bacteria, including the media biases and culturing condition biases, could have led to inadequate detection of differences between the bacteria endophyte communities between the different plants. It has to be taken into account that the culturable bacteria were only classified to the level of genus and not species or strains. A more in-depth analysis of the cultured endophytes that involves genotyping the bacterial endophytes is warranted and might potentially result in significant differences between the endophytic communities of different plant species. Potential further research would involve studying if the bacterial genera that are found in all the seasons, are of the same bacterial strain, or if the tree species all harbour the same strains.

For this study two different types of media were used to obtain information on the culturable bacterial endophytes: the Reasoner’s 2A (R2A), a low nutrient media and Tryptic Soy Agar (TSA), a nutrient rich agar. These represent the extremes of the nutrient spectrum and should allow for the detection of a variety of microorganisms. The two media have different carbon sources, concentrations and salt contents. R2A includes peptone, casamino acids, starch, yeast extract, dextrose and pyruvate as the main carbon sources, while in TSA the carbon sources are soy peptone and casein (Zimbro et al., 2009). Between the two media, R2A has more complex and diverse carbon sources while TSA provides a saltier environment with less variety of carbon sources. This might explain why a higher bacterial richness and diversity was detected with R2A in comparison to TSA and why the use of this medium allowed for the detection of the significant difference between the bacterial community of the plant species examined. As shown in the results, culturable bacteria do not all necessarily grow in the same type of media regardless of the fact that they are found inside the same environment. It is expected that the media used did not contain a sufficient variety of carbon sources, and proper salt concentrations to allow for the growth of greater variety of endophytes present, but instead only allow for the

85 detection of similar bacteria in each plant. This might have led to detection of no significant differences in the culturable endophytes between Acer negundo and the Ulmus parvifolia. This does raise the question of how many other types of bacteria would have been isolated if different media that consisted of alternative carbon sources, salt concentrations were used. Another factor to take into consideration with the culture method is the fact all the plated macerates were incubated at 28˚C, typical temperature used to incubate soil and endophytic bacteria, instead of at the temperature they were initially sampled from. Incubating at 28˚C might have allowed any endophytic bacteria present in the plant with that optimum temperature to grow on the plates. This might not necessarily reflect the true, metabolically active bacteria in the plant during the Winter and Autumn, when the temperature was -2˚C and 11˚C respectively. Ideally future research should involve comparing cultured bacterial endophytes from the Winter and Autumn plant samples incubated at the same temperature as the sampling temperature and plates incubated at 28˚C, to determine if there would be a drastic difference in the detected endophytic bacteria.

As part of the culturable analysis of the endophytes, another limitation to this study was that many bacteria present in the original spread plate of the macerate could not be recultured. This is a commonly occurring phenomena when attempting to culture bacteria from various environments (Stewart, 2012). It is possible that this lack of subsequent growth could be attributed to absence of something in the new media that was originally present in the plant macerate or the growth of the bacteria requires it to be growing in the presence of another bacteria (Stewart, 2012; Thomas & Soly, 2009). This might require the addition of sterilized plant extracted onto of the media plates to allow for the growth of the endophytes. In some cases it is also possible that the morphology of the bacterial colony inhibits the bacteria from being picked up by a simple loop for re-streaking. If this is the case excision of whole colonies and vigorous disaggregation of the component cells followed by serial dilution or spread plating would be required for re-isolation and purification of these colony types. Regardless, the unidentified bacteria were still recorded and added to the data under a code name instead of a bacterial name and treated as bacterial genera so their presence is accounted for in the density, richness and diversity data.

Even with the limitations of the media used, the bacteria isolated from the plant tissues corresponded to known bacterial genera that have been previously isolated from the internal

86 tissues or phyllosphere of other crops and trees (Lodewyckx & Vangronsveld, 2002; Qin et al., 2012). The exception was Patulibacter sp. which has been previously only been isolated from soil samples (Takahashi et al., 2006). The majority of the identified cultured bacteria endophytes were classified as under the phylum Actinobacteria. The dominance of Actinobacteria in cultured bacterial endophytes has been previously seen in the study of endophytic communities of poplar trees (Ulrich et al., 2008) and corn (Chelius & Triplett, 2001). It is uncertain the reason for the abundance of bacteria from this phylum, but previous studies on Actinobacteria endophytes has focused on their roles as biological control agents (i.e. Streptomyces sp. and Nocardiodes sp.) (Cao et al. 2005; Coombs, Michelsen, & Franco, 2004), plant growth enhancers (Igarashi et al 2002) and producers of novel natural products (Qin et al 2011). Although typically Actinobacteria are known to be spore-formers, whereby they form dormant, non-reproductive bodies, the majority of the Actinobacteria isolated from these trees were classified as non-spore formers with the exception of Geodermatophilus spp.. There are other spore formers that were isolated from these trees, mainly the Bacillus spp. from the phylum Firmicutes. It is possible that the Firmicutes were present in the plant as spores in the Winter and culturing the spread plates at 28˚C allowed for them exit the dormant stage and proliferate, which might explain their higher abundance in the Winter. Ultimately regardless of the seasons, it is uncertain if the bacteria cultured were metabolically active within the plant tissue or in a dormant state. It is possible that these bacteria do not contribute to the plant well being and could potentially be free-living bacteria that entered the plant by chance and remained inside the plant not serving a purpose inside the plant (Mengoni et al., 2009). Further research needs to be conducted in order to determine if the isolated Actinobacteria endophytes or the other endophytes play an important role in the plants and potentially explain their abundance in the culturable endophytes.

Given that the culturable endophytes are only a portion of the total amount of endophytes found inside the plant, culture independent analyses need to be carried out in the samples in order to consider the entire endophyte population in the tree tissues. Therefore based on the culturable results, it can be tentatively concluded that there was a significant difference between the communities of some plant species when R2A media was used to culture the endophytes, supporting part of the first hypothesis. There also was a significant difference in the

87 communities depending on the season the plants were sampled, therefore supporting the second hypothesis.

Chapter 7 Analysis of Variation – Culture Independent Method 7.1 Introduction

As previously mentioned, there are biases associated with culturing bacteria that result with only a portion of the total endophytic community being detected. Through the extraction of DNA from the samples, this allows for all the endophytes present in the plant to be potentially detected, depending on the subsequent molecular method chosen for analysis. As described in Chapter 5, the pre-digestion of the genomic DNA before amplifying with the full 16S primers was ideal at allowing for the improved amplification of the bacterial DNA in a sample with plastid DNA dominating the population. This method combined with the analysis of terminal restriction fragment length polymorphisms (T-RFLP) that results in a mix of various terminal restriction fragments (T-RFs) per sample, allows for generation of the bacterial endophytic community profiles in the plant samples. The T-RFs correspond to different bacterial phylotypes present in the community. These culture independent analyses were carried out on the bacterial endophytic community of the Acer negundo, Ulmus parvifolia and Ulmus pumila to determine if the results obtained with the culture dependent methods in Chapter 6 would still be valid when bacteria that could not be cultured were taken into consideration. This would help confirm if there were variations between the endophytic community of trees based on plant species they were sampled from or from the season they were collected.

7.2 Methods 7.2.1 DNA Extraction of Endophytic Community

Plant samples and processing methods were the same as outlined in Chapter 6. After plant macerates were subsampled for culturable bacteria, their remainders were used to extract total tissue DNA through the FastDNA SPIN Kit (Mp Biomedicals) following manufacturer’s instructions with a couple of modifications. The modifications included the addition of 100 µL of protein precipitation solution (PPS) solution to help remove the plant proteins present in the sample before the macerate was homogenized in a bead beater and the addition of 2 more washes of SEW-SM solution (ethanol wash) to increase the purity of the DNA through the

88 89 removal of salts and water soluble impurities. DNA was eluted in the elution buffer provided in the kit and used as template for subsequent molecular testing.

7.2.2 Enzymatic Digestion, PCR Amplification and TRFLP

For each sample, 10 µL of the genomic DNA was individually digested with 0.1 µL of 10 U/µL PvuII and 0.2 µL of 5 U/µL MscI (NEB Canada). The samples were digested for 3 hours at 37˚C, followed by incubation at 80˚C for 20 minutes to inactivate the enzymes. Preliminary community analyses were conducted using DGGE whereby for each sample, 2 µL of each digested genomic DNA were combined and used as template in the PCR amplification reaction using primers 27F and 1492R, the amplicons gel purified, and used as template for a PCR reaction with primers 341F and MOD783R before visualized in a DGGE gel as mentioned in Chapter 5.2.3. It was determined that the gels generated from DGGE were not quantitative enough to allow for determination of significant variation and statistical analysis between the samples. Therefore it was determined that T-RFLP analysis would be conducted on the samples, allowing for appropriate detection of variation between the samples and generate reproducible and consistent results between the samples. For T-RFLP analyses, for each sample, 2 µL of each digested genomic DNA were combined and used as template in the PCR amplification reaction using a forward primers labeled with 5’-fluorescein amidite dye (27F-FAM from LifeTechnologies, Canada) and a reverse primer labeled with 5’–hexachlorofluorescein dye(1492R-Hex from LifeTechnologies, Canada). The PCR reaction was carried out in a 20 µL reactions with a final concentration of 0.5 mM of the forward primer and reverse primer, 1.5 mM MgCl2, 200 mM of each dNTP, 2.5 units of HotStarTaq Plus DNA polymerase (Qiagen, Canada) and 1 µL of genomic DNA (30 ng/µL – 50 ng/µL). The conditions in which the PCR was carried out in a PTC-200 thermal cycler (MJ Research Inc.) were as follows: initial denaturing at 95˚C for 5 minutes followed by 35 cycles of: denaturing at 95˚C for 1 min, annealing at 56˚C for 1 min and extension at 72˚C for 1 min; final extension at 72˚C for 10 mins. Three microliters of the PCR product was run in 1.0% agarose gel to ensure successful amplification. In early experiments, putative mitochondrial amplicons were avoided by gel purifying only the 1500 bp amplicons from these PCR reactions. This was due to in silico analysis showing mitochondrial amplicons from 27F-1492R ranging from ~1680 bp to ~1950 bp in comparison to bacterial amplicons of ~1500 bp in size. However upon further testing, the gel

90 purification of only the 1500 bp product still resulted in the presence of mitochondrial amplicons and decreased amount of bacterial products. Therefore, the gel purification step was omitted and the amplicons generated from the PCR reaction were used directly.

The amplicons were digested with 0.20 µL of 5 U/µL MspI restriction enzyme (ThermoScientific). MspI was chosen for the T-RFLP analysis based on its power to resolve the terminal restriction fragments of my collection of sequenced culturable endophyte isolates – after comparison to restriction enzymes RsaI, TaqI and HhaI. The PCR amplicons were digested for 3 hours at 37˚C followed by inactivation through incubation at 80˚C for 20 minutes. In order to analyze the terminal fragments generated, the digested PCR amplicons were sent to the Agriculture and Food Laboratory (AFL) at the University of Guelph. At AFL the samples were PCR purified before terminal fragments were sized using an ABI 3730 autosequencer with a fluorescence detector, detecting the terminal fragments in the sample labeled with the 2 different fluorescein dye used.

7.2.3 Data Analyses – T-RFLP and Statistical

The data obtained from AFL consisted of a list of forward and reverse terminal restriction fragment sizes (T-RFs) and their associated fluorescence signals (peak heights and areas) for each sample. Only fragments between 60-1200 bp with fluorescence signals greater than 100 units were included in the output from AFL. This data provided us with community composition information, where each T-RF represents a phylotype and the peak heights and areas are measures of abundance. I chose to use peak heights as the proxy for abundance to minimize the amount of background noise detected. The background noise could result in very small peaks that differ by less than 1 bp, and could have been merged together to form a broad peak and included in the peak area (Kitts, 2001). With the peak height, only the fluorescence of each T- RF is accounted for and if background noise were present, its peak height in comparison to true T-RFs would be relatively small and could be removed through trimming.

The Microsoft Excel macro Treeflap (Rees et al., 2004), obtained from http://www.wsc.monash.edu.au/~cwalsh/treeﬂap.xls, was used to round the fragment sizes to the nearest one base pair and to align the fragments of the same size from different samples, generating a cohesive table that contained the different fragments sizes and their relative heights in each sample. The heights were converted to % abundance numbers by dividing the individual

91 heights by the sum of all heights for each sample to normalize across all the different samples and account for variations in the DNA concentration of the template available for PCR, as well as the concentration of digested product loaded into the autosequencer. Fragments in the dataset in the range of 335-338 bp for the forward fragments or 164-168 bp for the reverse fragments were assumed to represent mitochondrial fragments and omitted from the data. Fragments sizes that had less than 1% abundance or that appeared in less than three samples were removed to account for any background noise generated during the PCR reaction, sequencing artifacts or rare members of the communities. The resulting data are referred to below as the “trimmed data”. In order to determine the potential bacterial identities of the forward and reverse T-RFs, a list of endophytic bacteria from the RDP database and its corresponding T-RF size after digestion with MspI was created, shown in Appendix B.1. MspI T-RF lengths for the bacterial sequences obtained from the isolated endophytes (as described in the chapter 6) were also calculated and are shown in Appendix B.2. The list of corresponding reverse T-RFs for the cultured endophytes could not be generated due to lack of good quality sequence from the 1492R end.

Using the trimmed T-RFLP data, the phylotype richness and the Shannon diversity indices were determined for both the forward and reverse fragments datasets. The phylotype community structures were also analyzed using NMDS plots, hierarchical clustering and permutation multivariate analysis of variance using methods shown in Chapter 6.2.3.

7.3 Results 7.3.1 Preliminary Analyses - DGGE Results

A preliminary analysis of the communities of Acer negundo and Ulmus spp. samples using DGGE gels are shown in Figures 7.1 and 7.2. As seen in the Figures, there appear to be dominant bands that correspond to a specific phylotype that appears in all the Fall and all the Winter for both plant genera examined. When some of the bands were cut and sequenced, it was determined that the bands corresponded to Ralstonia spp. and Pantoea spp.. Clearly DGGE is sensitive to the most dominant bacterial phylotype present in the sample such that it preferentially amplifies that phylotype. Therefore outside of detecting the dominant phylotype, it is hard to detect other members of the community shown in the faint bands present in the gel, making this method the ineffective to visualizing the entire endophytic community of a plant

92 sample. There are other problems such as difficulty normalizing between samples that are run on different gels that affect how subsequent comparisons and statistical analyses will be carried out that T-RFLP was chosen to be analyze the endophytic communities of the samples.

Figure 7.2: DGGE gel of Ulmus parvifolia and Ulmus pumila tree samples collected during Winter, Summer and Autumn 2012. The lanes 1-6 represent Ulmus parvifolia samples with lanes 1-3 – branches from tree replicate 1 and lanes 4-6 – branches from tree replicate 2. The lanes 10-15 represent Ulmus pumila samples with lanes 10-12 – branches from tree replicate 1, lanes 13-15 – branches from tree replicate 2 and L – DGGE ladder. Some bands were excised and their identities were determined to be as follows: band in red with letter R = Ralstonia spp. and band in green with letter P = Pantoea spp.

7.3.1 Analysis of Individual Terminal Restriction Fragments, Richness and Diversity

The T-RFLP analysis of the MspI digested bacterial amplicons derived from the tree samples resulted in 107 different forward T-RFs ranging in size from 61 bp to 1103 bp and 93 different reverse T-RFs ranging in size from 60 bp to 501 bp. The forward T-RFs refers to the terminal fragments that were generated with the forward 27F primer labeled with the 5’-fluorescein amidite dye and the reverse T-RFs refer to the terminal fragments generated with the reverse primer 1492R labeled with 5’–hexachlorofluorescein dye. With the sizes of T-RFs reflecting the differences in the sequence polymorphisms in bacterial 16S rRNA, T-RFs in a T-RFLP analysis corresponds to different phylotypes found in the samples. Amongst the forward T-RFs, five were common to all the branches sampled: 61 bp, 119 bp, 128 bp, 488 bp, and 612 bp. Some forward T-RFs were only associated with certain plant species (Table 7.1). With regard to the reverse T-RFs, only 4 T-RFs were common to all the branches and they were as follows: 60 bp, 114 bp, 146 bp, and 149 bp. Like the forward T-RFs, there were some reverse fragments that

94 were associated with certain plant species and they are listed on Table 7.2. In order to identify the bacteria that the T-RFs may correspond to, the size of the fragments were compared to the list of known bacterial T-RFs in Appendix B, Tables B.1-B.2. Upon comparison of the known bacterial sequences to the obtained T-RFs, there were no exact matches found, thus the identity of these bacteria that correspond to the T-RFs remains unknown.

Table 7.1: List of forward T-RFs exclusive to certain plant species with  indicating presence of the fragment in the samples from the plant species. Forward T-RF Plant species

size Acer negundo Ulmus parvifolia Ulmus pumila 161  163  

268   274   289  

324   340   377/378 

384  

398  417   485   611 

670  1103  

Table 7.2: List of reverse T-RFs exclusive to certain plant species with  indicating presence of the fragment in the samples from the plant species. Reverse T-RFs Plant species size Acer negundo Ulmus parvifolia Ulmus pumila 63   90   100   108  116   121   122   124  142   156  174/175/176  197   200   277   331   333  341  344   477   501 

The T-RFs present in each sample were used to calculate the phylotype T-RFs richness and diversity, with the average of phylotype richness and diversity for each plant species in each season illustrated in Figure 7.3. There are differences between the data obtained from the forward T-RFs and the reverse T-RFs that reflect the amount of sequence polymorphisms found in the bacterial 16S rRNA gene. In general more phylotype richness and diversity was detected

96 with the forward T-RFs data in comparison to the reverse T-RFs data. ANOVA tests were used to test differences in species richness and diversity values and the results are shown in Table 7.3. When the species richness values for the samples were analyzed, a significant difference was found in the reverse T-RF profiles of the Acer negundo and Ulmus parvifolia samples (F-value = 2.87, p-value = <0.05). With regards to calculated diversity index using the Shannon index, there was a significant difference in the forward T-RFs profiles of the samples collected in the Winter and Summer but not in the Autumn (F-value = 3.69, p-value = 0.031).

a) b)

c) d)

Table 7.3: ANOVA analysis on the phylotype richness and the diversity index values of the samples based on the plant species and seasons for the forward and reverse T-RFs data sets. Data Fragments Comparison between Comparison between Dataset Plant Species Seasons F-Value P-value F-Value P-value Phylotype Forward 2.87 0.06 2.13 0.13 Richness Reverse 3.58 <0.05 0.95 0.39 Diversity Forward 0.07 0.93 3.69 <0.05 Index Reverse 1.57 0.22 0.68 0.51

7.3.2 Comparison of Community Structures Between Samples

The bacterial community profiles generated by T-RFLP were preliminarily compared by performing NMDS analyses and hierarchical clustering on the data based on the season they were sampled from as seen in Figures 7.4-7.7 and Appendix C, Figures C.5-C.8. The NMDS plots and dendrograms show some clustering and some separation of the Acer negundo samples from the Ulmus spp. samples in each season. These species differences are most pronounced in the Winter samples, and the Acer samples are more distinguishable than the two Ulmus species.

When the data was grouped according to plant species and the seasons were compared, the NMDS plots and dendrograms showed that clusters of Summer samples were somewhat distinguishable from Winter and Autumn samples for the 2 plant genera.

Following the preliminary look at the samples based on the season and plant genera, the datasets were analyzed in their entirety as seen in Figures 7.8-7.13. With all the data together, the NMDS plots do not show separate distinguishable clusters between the samples from Acer negundo and those from Ulmus spp., with the forward T-RFs data but do show some clusters with the reverse T-RFs data. There are also some clear separations of the Acer negundo samples from the Ulmus spp. samples in the hierarchical clustering dendrograms. In order to further determine if the groupings between the samples based on the plant species were significant, permutation multivariate analyses (adonis) were carried out and results shown in Tables 7.4 and 7.5. The results obtained for the forward T-RFs matched the results obtained from the reverse T-RFs. A

98 significant difference was found between the endophytic community profiles of the Acer negundo samples in comparison to the Ulmus parvifolia and Ulmus pumila samples. However there was no significant difference between the endophytic community of Ulmus parvifolia and Ulmus pumila.

With regards to their sampling months, the samples from the same months did not appear to form cohesive and distinguishable groups in the NMDS plots created using the reverse T-RFs data, but with the forward T-RFs data, the Summer samples formed separate clusters from the Autumn and Winter samples. In the hierarchical clustering there were some clustering of the samples based on the season they were collected, but these clusters were found throughout the dendrogram interspersed amongst different branches. Subsequent permutation multivariate analysis conducted on the data showed that there was a significant difference found between the community profiles of the samples collected in the Summer compared to those collected in Winter and Autumn in both the forward and reverse T-RFLP datasets. The results also indicated an interaction was present between the plant species and the season collected, suggesting that any seasonal differences are plant specific or plant differences are seasonally specific.

Winter Forward T-RFs Data Summer Forward T-RFs Data

a) b)

Autumn Forward T-RFs Data

100

Winter Reverse T-RFs Data Summer Reverse T-RFs Data

a) b)

Autumn Reverse T-RFs Data

101

– Ulmus pumila samples. The stress values for the NMDS plots are as follows: Winter – 0.138, Summer – 0.144 and Autumn – 0.211.

Acer negundo Forward T-RFs Data Ulmus spp. Forward T-RFs Data

a) b)

102

Acer negundo Reverse T-RFs Data Ulmus spp. Reverse T-RFs Data

a) b)

103

Forward T-RFs Data

Plant species Season

a) b)

104

Forward T-RFs Data – Plant Species

105

Forward T-RFs Data – Seasons

106

Reverse T-RFs Data

Plant species Season

a) b) Figure 7.11: Analysis of bacterial community profiles from the entire reverse T-RFs data. NMDS plot of the community profiles based on a) plant species with symbols representing different species:  - Acer negundo,  - Ulmus parvifolia and + - Ulmus pumila samples b) seasons with symbols representing different seasons:  – Winter, – Summer and – Autumn. The stress value for the NMDS plot was 0.217.

107

Reverse T-RFs Data – Plant Species

108

Reverse T-RFs Data – Seasons

109

Table 7.4: Permutation multivariate analysis of forward TRFLP data conducted using adonis after 999 permutations.

Factor F Model R2 P-value Individual analysis (*=significantly different) Season 3.16 0.09 <0.05 Winter* Summer* Autumn* Plant species 3.96 0.11 <0.05 Acer negundo* Ulmus parvifolia Ulmus pumila Season:Plant 2.06 0.10 <0.05 species

Table 7.5: Permutation multivariate analysis of reverse TRFLP data conducted using adonis after 999 permutations. Factor F R2 P-value Individual analysis Model (*=significantly different) Season Winter* 2.30 0.0682 <0.05 Summer* Autumn* Plant species Acer negundo* 3.7873 0.10733 <0.05 Ulmus parvifolia Ulmus pumila Season: Plant species 1.6741 0.08668 <0.05

7.4 Discussion

The culture independent analysis with the use of the enzymatic predigestion in combination with T-RFLP allows for an in-depth study of the entire endophytic community, culturable and nonculturable bacteria, as the community DNA was extracted from each of the samples. This analysis like the culture dependent analysis showed there were significant differences between

110 the endophytic communities of the samples based on the season they were collected from. However unlike the culture dependent analysis, where only the endophytic community of Ulmus pumila differed from the other plant species based on data collected on one media type, with this analysis the results for both forward and reverse data sets showed significant differences between the communities of Acer negundo and Ulmus spp. With regards to different results obtained from the culture dependent analyses, it has to be taken into consideration that the culturable endophytes only represent a portion of the endophytes present in the plant. Only certain types of bacteria will be able to grow on the media and they are selected for out of the entire sample, such that similar bacteria would always be isolated in the media plates regardless of the plant species. At the same time, the identification of the bacterial endophytes solely on the basis of 16S rRNA analysis is not sufficient to distinguish bacteria of the same genus but different species. This creates the illusion that the endophytic communities of different plants are similar when in reality this might not be the case. Therefore between the two methods, community analysis of bacterial endophytes would be better conducted through culture independent methods, which generate more reliable and consistent results, unlike the culture dependent methods whose results vary dependent on culturing conditions. However the two methods could complement each other in the study of endophytes such that individual endophytic members could be analyzed and isolated for further experimentation through the culture dependent method, while overview of the community could be generated through the use of culture independent method. The results of this analysis, does indicate that for any study involving the endophytic community of trees, it is necessary to do comparisons solely based on trees from the same species and trees from the same season in order to minimize the variation that results due to these factors.

As previously mentioned, the community profiles in both the forward and reverse T-RFs datasets of the different plant species were found to be different from each other. Further analysis showed that the samples from Acer negundo were different from those of Ulmus spp. However there were no significant differences detected between the endophytic communities of Ulmus parvifolia and Ulmus pumila samples. Typically in other tree studies other such as with poplar, the diversity of endophytic bacteria via culture independent analyses were found to be influenced by the plant genotype (Moore et al., 2006). The differences in the community due to the plant genotype might be potentially associated with the difference in plant metabolism,

111 nutrients and carbon in the plants and the genetic content of the plant. With different plant species, the carbon in plants can be found in the form of different nonreducing sugars, oligosaccharides, sugar alcohols, starch and other polysaccharides (Bloom, Chapin III, & Mooney, 1985). The ability of the endophytes to use different carbon sources would influence which bacteria are able to grow and thrive, thus influence the community structure. At the same time, it has been suggested that the plant host plays an active role in the colonization of the endophytes by attracting specific bacteria, this might involve communication through the release of certain compounds via their roots (Compant et al., 2005; de Weert et al., 2002; Lemanceau et al., 1995), or by enhancing or diminishing their colonization in the plant, through plant defense response and generation of phytohormones such as jasmonic acid and ethylene (Miché et al., 2006; Shah, 2009; Vargas et al., 2003). Therefore it might just be that between the 2 Ulmus spp. analyzed, the carbon sources and the defense response in the plant were similar, such that similar endophytic communities were detected.

In the analysis of the samples based on the seasons, there were significant variations in the community profiles in both the forward T-RFs and reverse T-RFs datasets. This finding was supported by the culturable analysis done in Chapter 6 with the 2 different media. The results obtained support the second hypothesis that endophytic communities of plants are affected by seasonal variations and there are statistical differences between the microbial profiles obtained from different seasons. Like the significant differences found in the total number of bacteria cultured from each season, this seasonal variation in communities could be attributed to the seasonal changes in the concentration of nutrients and carbohydrate as explained in Chapter 6.

Although differences were detected between the endophytic communities of the different samples based on the plant genera it belonged to and the season it was collected, upon examination of the individual T-RF fragments, there were T-RFs that were found in every single branch analyzed both in the forward and reverse data sets. With the sizes of T-RFs reflecting the differences in the sequence polymorphisms in bacterial 16S rRNA, one T-RF in a T-RFLP analysis could belong to more than one bacterial species or genera, thereby underestimating the diversity (Abdo et al., 2006; Schütte et al., 2008). Therefore these T-RFs could potentially represent a core group of bacteria phylotypes that are found in the 3 tree species examined. The corresponding bacterial genera or species of the obtained T-RFs could not be determined as the sizes did not match any fragments in the list of known bacteria T-RFs or my collection of

112 endophytes. It is possible that the isolated bacteria made up a very small percentage of the total number of bacterial endophytes in the sample. Therefore when molecular analyses were conducted, they were underrepresented in the sample, detected at very low levels and mistakenly removed as background noise in the T-RFLP analysis. This is a common problem found when doing both culture dependent and culture independent analysis where the culturable community generated greatly differs from the community generated through molecular means (Reiter & Sessitsch, 2006; Ulrich et al., 2008). This difference could be potentially attributed to biases associated with PCR including preferential amplification by certain primer pairs and amplification efficiency of the template DNA.

Lastly, it is likely that the obtained T-RFs sizes might need to be adjusted to account for differences associated with the migration of the fragments due to the fluorescein dye attached to the DNA (Schütte et al., 2008). According to Schütte et al. (2008) different fluorophores can affect how the fragments travel in the DNA sequencers. The ladder used to standardize the fragment sizes is typically attached to a ROX fluorescein dye, which differs from the 6-FAM, and HEX fluorescein dye attached to the bacterial 16S rRNA fragments. The differences in the migration of the fragments caused by the fluorophores does not result in a constant discrepancy across all fragment sizes but instead the discrepancy is based on certain fragment sizes (Schütte et al., 2008). This might explain why none of the T-RFs corresponded to the uncultured Rahnella spp. and Pantoea spp. found to be dominant phylotypes in the preliminary DGGE analysis. Therefore in this case, a sample containing a mix of known bacterial 16S rRNA, for which the expected T-RF size was already calculated, should have been analyzed along side these samples. With this control, the discrepancies in the migration would be accounted for and potentially allowed for a more accurate identification of the T-RFs obtained.

All of the analyses that were carried out studied the endophyte community of branches from the 3 tree species, which represent a portion of the total endophytic community of the entire plant. It has to be taken into consideration that these plants taken from the same site were analyzed with the assumption that they were at similar growth stages; they were planted at the same time. There could potentially other factors such as plant growth stage that might influence the community of the endophytes but not considered in this study. The results obtained here are still significant and relevant in further studies involving other plants and improving phytoremediation, the use of plants to remediate a contaminated site.

113

In the end the culture independent analysis of the endophytic community of Acer negundo and Ulmus spp. shows that there are significant differences in the community composition based on the plant genera and the seasons the plants are analyzed.

Chapter 8 Conclusions and Recommendations for Further Studies

All of the analyses and results obtained in this research project were undertaken with two objectives in mind. The first objective was to optimize the methods for the molecular analysis of bacterial endophytes in plants while minimizing the contamination from plastid DNA. The second objective was to study the variation of the endophytic community structure of different tree species from different seasons found in the same site. Understanding the variation in the endophytic community allows for optimal planning of sampling strategies for the future studies of ecological patterns and effects of contaminants on endophytic community structures.

Through the testing of different molecular methods it was concluded that literature plastid excluding primers were valid primers to use for the amplification of bacterial DNA while excluding the amplification of plastid DNA of Acer negundo, Ulmus pumila and Ulmus parvifolia plants. However there was minimal amplification due to the interference by plastid DNA and low densities of endophytic bacteria in the tissues. Although a variety of pre- extraction and post-extraction methods to limit the interference of plastid DNA were tested, only one method was found to be effective, I determined that the pre-digestion of the genomic DNA with restriction enzymes PvuII and MscI, which digest plastid DNA while leaving the majority of the bacterial DNA intact, allowed for the improved amplification of endophytic DNA in the chosen samples. This method was used in the subsequent community analyses and could be applied in any future culture independent analysis of other plant endophytic communities.

The second objective was carried out using Acer negundo, Ulmus pumila and Ulmus parvifolia trees found on a hydrocarbon contaminated site. This research revealed the extent of the variation generated by plant genotype and season on the community of endophytes and allowed me to look for any evidence of host specificity, i.e. tree specific communities or isolates. The endophytic community analyses were conducted using culture dependent analysis and culture independent analysis that involved the use of the developed pre-digestion method in combination with T-RFLP. A summary of the results obtained is shown in Table 8.1.

114 115

Culture Dependent Method Culture Independent Method Significant Plant Species Season Plant Species Season Difference in Total R2A R2A Not Applicable Not Applicable Bacteria None Autumn Abundance (Greater) TSA TSA None Autumn (Greater) Bacterial R2A R2A Forward T-RFs Forward T-RFs species or None None None None phylotype richness TSA TSA Reverse T-RFs Reverse T-RFs None None Acer negundo None (Greater) Bacterial R2A R2A Forward T-RFs Forward T-RFs diversity None None None Winter (Greater) TSA TSA Reverse T-RFs Reverse T-RFs None None None None Community R2A R2A Forward T-RFs Forward T-RFs Structure Ulmus pumila All seasons Acer negundo Between all vs. Ulmus spp. seasons TSA TSA Reverse T-RFs Reverse T-RFs None Winter Acer negundo Between all vs. Ulmus spp. seasons

The following conclusions were made from the analyses.

1. The plant genera had a significant effect on the endophytic community present within the plant. This was clearly seen through the culture independent analysis but not through the culture dependent analysis.

116

2. In both culture dependent and culture independent analyses, the season from which the samples were collected had an influence in the endophytic community. This resulted in statistically significant differences between the endophytic communities of samples collected in Winter, Summer and Autumn.

3. Even with different endophytic community profiles between the samples based on the season and the plant species, there were bacterial genera that were found in all the plants examined, throughout all the seasons. The culture independent analysis revealed that there was also a set of unidentified T-RFs that were found in all the samples. These might represent a core number of bacterial species or genera that are shared in all the trees analyzed. There were also T-RFs that were associated with solely Acer negundo, Ulmus parvifolia or Ulmus pumila, potentially indicating plant species specific bacterial phylotypes.

The conclusions obtained from this research have a potential impact on future research associated with the study of endophytic communities of plants in contaminated sites and phytoremediation, the use of plants to remediate contaminated sites. The changes in the endophytic community composition with the changing seasons are important variations to take into consideration when the endophytic community of a plant is being studied. If the composition changes, the role different of endophytes might potentially change with the changing seasons. It is possible that bacteria with certain capabilities (i.e. ability to degrade hydrocarbons or other organic contaminants) could potentially only be present or found in higher abundance in certain seasons. Therefore sometimes it might be necessary to study the endophytic communities of the plants through different seasons and this might improve the understanding of plant-endophyte interaction.

At the same time if different plants are able to actively select the bacteria that will colonize their tissues and certain bacteria are associated with certain plant genotypes, this will affect the study of endophytes of plants in a contaminated site. In future studies the search for catabolic endophytes cannot solely rely on looking at only one plant species but instead it might be necessary to study a variety of plant species from the contaminated site might each possess different bacteria with different catabolic capabilities. With regards to future phytoremediation endeavors, the endophytic community of plants used for phytoremediation purposes needs to be studied carefully. In the cases where genetically engineered bacteria with certain degradative

117 genes or the ability to confer advantage in growth promotion or pathogen protection were to be introduced into a plant, a bacteria that is naturally found in the plant would be the best choice. This would ensure maximum success for establishment and integration of the bacteria in the plant since the bacteria would already be present in the plant and minimize any antagonistic effects or disruption to the endophytic community of the plant.

Although a lot of data was obtained from this project there is still considerable amount of research that could be done to enhance the information obtained. In the method development section of this research a lot of focus was placed on minimizing or eliminating plastid contamination, while maximizing the detection of bacterial DNA. Plastid interference was successfully minimized however mitochondrial contamination was still present. Ideally the next part of this research would be to determine if the post extraction methods could be used to minimize or eliminate the mitochondrial contamination. This could involve either using the subtractive hybridization protocol using the designed mitochondrial probe or finding a restriction enzyme that would be able to digest mitochondrial DNA while leaving bacterial DNA intact. Ideally the latter option would be less time consuming and more economically feasible over the former option.

With the minimization of the plant organelle interference, pyrosequencing would be the next step in analyzing the endophytic diversity of the plants used. This would also help generate a database of sequences and their corresponding T-RFs that could be matched to those obtained through T-RFLP and help with the identification of the core fragments that were found in all the plant samples and the plant species specific T-RFs.

The isolated endophytes could also be used for future research that would involve further delving in the following topics:

• What are the metabolic capabilities of the isolated endophytes, as well do any of the endophytes produce secondary metabolites?

• Given that the endophytes were isolated from hydrocarbon-contaminated site, what is the gene content of these bacteria? Do they possess specific catabolic genes or plasmids and are these important functional genes the same between seasons and strains?

118

• Would the bacterial species found from one season to the next be the same strain?

• Are the isolates of the same genus found in all three trees also exactly the same strains?

There is still a lot more research that needs to be conducted on the endophytes that are beyond the scope of this study. Ultimately, studying the endophytic community serves as a gateway to developing ways to utilize the capabilities of endophytes such as in phytoremediation and further understanding the roles endophytes play within various plants.

Bibliography

Abdo, Z., Schüette, U. M. E., Bent, S. J., Williams, C. J., Forney, L. J., & Joyce, P. (2006). Statistical methods for characterizing diversity of microbial communities by analysis of terminal restriction fragment length polymorphisms of 16S rRNA genes. Environmental microbiology, 8(5), 929–38. doi:10.1111/j.1462-2920.2005.00959.x Adams, P., & Kloepper, J. (2002). Effect of host genotype on indigenous bacterial endophytes of cotton (Gossypium hirsutum L.). Plant and Soil, 240, 181–189. Retrieved from http://www.springerlink.com/index/404GBXJVAXEAG3C0.pdf Adhikari, T. B., Joseph, C. M., Yang, G., Phillips, D. a., & Nelson, L. M. (2001). Evaluation of bacteria isolated from rice for plant growth promotion and biological control of seedling disease of rice. Canadian Journal of Microbiology, 47(10), 916–924. doi:10.1139/cjm-47- 10-916 Amann, R. I., Ludwig, W., & Schleifer, K. H. (1995). Phylogenetic identification and in situ detection of individual microbial cells without cultivation. Microbiological reviews, 59(1), 143–69. Retrieved from http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=239358&tool=pmcentrez&rend ertype=abstract Araújo, W., Marcon, J., Maccheroni, W., Van Elsas, J. D., Van Vuurde, J. W., & Azevedo, J. L. (2002). Diversity of endophytic bacterial populations and their interaction with Xylella fastidiosa in citrus plants. Applied and environmental microbiology, 68(10), 4906–4914. doi:10.1128/AEM.68.10.4906 Archer, M. J., & Lin, B. (2011). Development of a single-step subtraction method for eukaryotic 18S and 28S ribonucleic acids. Journal of biomedicine & biotechnology, 2011, 910369. doi:10.1155/2011/910369 Archer, M. J., Lin, B., Wang, Z., & Stenger, D. a. (2006). Magnetic bead-based solid phase for selective extraction of genomic DNA. Analytical biochemistry, 355(2), 285–97. doi:10.1016/j.ab.2006.05.005 Aronsson, H. Ã., & Jarvis, P. (2002). A simple method for isolating import-competent Arabidopsis chloroplasts. FEBS letters, 529, 215–220. Retrieved from http://www.sciencedirect.com/science/article/pii/S0014579302033422

119 120

Bacon, C., & Hinton, D. (2006). Bacterial endophytes: the endophytic niche, its occupants, and its utility. In S. S. G. Gnanamanickam (Ed.), Plant-Associated Bacteria (pp. 155–194). Springer Netherlands. Retrieved from http://www.springerlink.com/index/Q652V7H070G51M63.pdf Bakken, L. R., & Olsen, R. a. (1983). Buoyant densities and dry-matter contents of microorganisms: conversion of a measured biovolume into biomass. Applied and environmental microbiology, 45(4), 1188–95. Retrieved from http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=242437&tool=pmcentrez&rend ertype=abstract Bankston, J. L., Sola, D. L., Komor, A. T., & Dwyer, D. F. (2002). Degradation of trichloroethylene in wetland microcosms containing broad-leaved cattail and eastern cottonwood. Water research, 36(6), 1539–46. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11996342 Barkan, A. (2011). Expression of plastid genes: organelle-specific elaborations on a prokaryotic scaffold. Plant physiology, 155(4), 1520–32. doi:10.1104/pp.110.171231 Berensmeier, S. (2006). Magnetic particles for the separation and purification of nucleic acids. Applied microbiology and biotechnology, 73(3), 495–504. doi:10.1007/s00253-006-0675-0 Bloom, A., Chapin III, F., & Mooney, H. (1985). Resource limitation in plants - an economic analogy. Annual review of Ecology and Systematics, 16(1985), 363–392. Retrieved from http://www.jstor.org/stable/10.2307/2097053 Cambours, M. a., Nejad, P., Granhall, U., & Ramstedt, M. (2005). Frost-related dieback of willows. Comparison of epiphytically and endophytically isolated bacteria from different Salix clones, with emphasis on ice nucleation activity, pathogenic properties and seasonal variation. Biomass and Bioenergy, 28(1), 15–27. doi:10.1016/j.biombioe.2004.06.003 Chelius, M. K., & Triplett, E. W. (2001). The Diversity of Archaea and Bacteria in Association with the Roots of Zea mays L. Microbial ecology, 41(3), 252–263. doi:10.1007/s002480000087 Cole, J. R., Wang, Q., Cardenas, E., Fish, J., Chai, B., Farris, R. J., Kulam-Syed-Mohideen, a S., et al. (2009). The Ribosomal Database Project: improved alignments and new tools for rRNA analysis. Nucleic acids research, 37(Database issue), D141–5. doi:10.1093/nar/gkn879

121

Compant, S., Duffy, B., Nowak, J., Clement, C., & Barka, E. A. (2005). Use of Plant Growth- Promoting Bacteria for Biocontrol of Plant Diseases: Principles , Mechanisms of Action , and Future Prospects. Applied and Environmental Microbiology, 71(9), 4951–4959. doi:10.1128/AEM.71.9.4951 Conn, V., & Franco, C. (2004). Analysis of the Endophytic Actinobacterial Population in the Roots of Wheat ( Triticum aestivum L .) by Terminal Restriction Fragment Length Polymorphism and Sequencing of 16S rRNA Clones Analysis of the Endophytic Actinobacterial Population in the Roots . Applied and environmental microbiology, 70(3), 1787–1794. doi:10.1128/AEM.70.3.1787 Cook, R. J., Thomashow, L. S., Weller, D. M., Fujimoto, D., Mazzola, M., Bangera, G., & Kim, D. (1995). Molecular mechanisms of defense root disease by rhizobacteria against. Proceedings of the National Academy of Sciences, 92(May), 4197–4201. Coombs, J. T., Michelsen, P. P., & Franco, C. M. M. (2004). Evaluation of endophytic actinobacteria as antagonists of Gaeumannomyces graminis var. tritici in wheat. Biological Control, 29(3), 359–366. doi:10.1016/j.biocontrol.2003.08.001 Cox, S. E., & Stushnoff, C. (2001). Temperature-related shifts in soluble carbohydrate content during dormancy and cold acclimation in Populus tremuloides. Canadian Journal of Forest Research, 31(4), 730–737. doi:10.1139/cjfr-31-4-730 De Mendiburu, F. (2012). agricolae: Statistical Procedures for Agricultural Research. Retrieved from http://cran.r-project.org/package=agricolae De Weert, S., Vermeiren, H., Mulders, I. H. M., Kuiper, I., Hendrickx, N., Bloemberg, G. V, Vanderleyden, J., et al. (2002). Flagella-driven chemotaxis towards exudate components is an important trait for tomato root colonization by Pseudomonas fluorescens. Molecular plant-microbe interactions : MPMI, 15(11), 1173–80. doi:10.1094/MPMI.2002.15.11.1173 Diekmann, K., Hodkinson, T. R., Fricke, E., & Barth, S. (2008). An optimized chloroplast DNA extraction protocol for grasses (Poaceae) proves suitable for whole plastid genome sequencing and SNP detection. PloS one, 3(7), e2813. doi:10.1371/journal.pone.0002813 Doty, S. L., Oakley, B., Xin, G., Kang, J. W., Singleton, G., Khan, Z., Vajzovic, A., et al. (2009). Diazotrophic endophytes of native black cottonwood and willow. Current Microbiology, 23–33.

122

Dowd, S. E., Sun, Y., Wolcott, R. D., Domingo, A., & Carroll, J. a. (2008). Bacterial tag- encoded FLX amplicon pyrosequencing (bTEFAP) for microbiome studies: bacterial diversity in the ileum of newly weaned Salmonella-infected pigs. Foodborne pathogens and disease, 5(4), 459–72. doi:10.1089/fpd.2008.0107 Drummond, A., Ashton, B., Buxton, S., Cheung, M., Cooper, A., Duran, C., Heled, J., et al. (2012). Geneious v5.6. Elvira-Recuenco, M., & Van Vuurde, J. W. (2000). Natural incidence of endophytic bacteria in pea cultivars under field conditions. Canadian journal of microbiology, 46(11), 1036–41. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11109492 Figueiredo, J., Gomes, E., Guimaraes, C. T., Lana, U. G. de P., Teixeira, M. A., Lima, G. V. C., & Bressan, W. (2009). MOLECULAR ANALYSIS OF ENDOPHYTIC BACTERIA FROM THE GENUS BACILLUS ISOLATED FROM TROPICAL MAIZE (Zea mays L.). Brazilian Journal of Microbiology, 40, 522–534. Retrieved from http://www.scielo.br/scielo.php?pid=S1517-83822009000300014&script=sci_arttext Franzreb, M., Siemann-Herzberg, M., Hobley, T. J., & Thomas, O. R. T. (2006). Protein purification using magnetic adsorbent particles. Applied microbiology and biotechnology, 70(5), 505–16. doi:10.1007/s00253-006-0344-3 Galluzzi, L., Bertozzini, E., Del Campo, a, Penna, a, Bruce, I. J., & Magnani, M. (2006). Capture probe conjugated to paramagnetic nanoparticles for purification of Alexandrium species (Dinophyceae) DNA from environmental samples. Journal of applied microbiology, 101(1), 36–43. doi:10.1111/j.1365-2672.2006.02952.x Garbeva, P., Overbeek, L. S., Vuurde, J. W. L., & Elsas, J. D. (2001). Analysis of Endophytic Bacterial Communities of Potato by Plating and Denaturing Gradient Gel Electrophoresis (DGGE) of 16S rDNA Based PCR Fragments. Microbial ecology, 41(4), 369–383. doi:10.1007/s002480000096 Gardner, J. (1982). Identity and behavior of xylem-residing bacteria in rough lemon roots of Florida citrus trees. Appl. Environ. Microbiol., 43(6), 1335–1342. Retrieved from http://aem.asm.org/content/43/6/1335.short Germaine, K. J., Liu, X., Cabellos, G. G., Hogan, J. P., Ryan, D., & Dowling, D. N. (2006). Bacterial endophyte-enhanced phytoremediation of the organochlorine herbicide 2,4- dichlorophenoxyacetic acid. FEMS microbiology ecology, 57(2), 302–10. doi:10.1111/j.1574-6941.2006.00121.x

123

Gottel, N. R., Castro, H. F., Kerley, M., Yang, Z., Pelletier, D. a, Podar, M., Karpinets, T., et al. (2011). Distinct microbial communities within the endosphere and rhizosphere of Populus deltoides roots across contrasting soil types. Applied and environmental microbiology, 77(17), 5934–44. doi:10.1128/AEM.05255-11 Gray, M. W., Burger, G., & Lang, B. (2001). Minireview The origin and early evolution of mitochondria. Genome Biology, 1–5. Gualberto, J. M., Handa, H., & Grienenberger, J. M. (1995). Chapter 12 Isolation and Fractionation of Plant Mitochondria and Chloroplasts: Specific Examples. In D. P. B. and H. J. B. B. T.-M. in C. B. David W. Galbraith (Ed.), Methods in Plant Cell Biology (Vol. Volume 50, pp. 161–175). Academic Press. doi:http://dx.doi.org/10.1016/S0091- 679X(08)61029-8 Hallmann, J., Quadt-Hallmann, A., Mahaffee, W. F., & Kloepper, J. W. (1997). Bacterial endophytes in agricultural crops. Canadian Journal of Microbiology, 43(10), 895–914. doi:10.1139/m97-131 Hinesley, L. E., Pharr, D. M., Snelling, L. K., & Funderburk, S. R. (1992). Foliar raffinose and sucrose in four conifer species: relationship to seasonal temperature. Journal of the American Society for Horticultural Science, 117(5), 852–855. Retrieved from http://journal.ashspublications.org/content/117/5/852.short Hollis, J. P. (1951). Bacteria in healthy potato tissue. Phytophathology, 41, 350–366. Izumi, H., Anderson, I., Killham, K., & Moore, E. (2008). Diversity of predominant endophytic bacteria in European deciduous and coniferous trees. Canadian journal of …, 54, 173–179. doi:10.1139/W07-134 Jansson, S., & Douglas, C. J. (2007). Populus: a model system for plant biology. Annual review of plant biology, 58, 435–58. doi:10.1146/annurev.arplant.58.032806.103956 Jiao, J.-Y., Wang, H.-X., Zeng, Y., & Shen, Y.-M. (2005). Enrichment for microbes living in association with plant tissues. Journal of applied microbiology, 100(4), 830–7. doi:10.1111/j.1365-2672.2006.02830.x Keeling, P. J. (2004). DIVERSITY AND EVOLUTIONARY HISTORY OF PLASTIDS AND THEIR HOSTS. American journal of botany, 91(10), 1481–1493. Kent, A. D., & Triplett, E. W. (2002). Microbial communities and their interactions in soil and rhizosphere ecosystems. Annual review of microbiology, 56, 211–36. doi:10.1146/annurev.micro.56.012302.161120

124

Kitts, C. L. (2001). Terminal restriction fragment patterns: a tool for comparing microbial communities and assessing community dynamics. Current issues in intestinal microbiology, 2(1), 17–25. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11709853 Kley, J., Heil, M., Muck, A., Svatos, A., & Boland, W. (2010). Isolating intact chloroplasts from small Arabidopsis samples for proteomic studies. Analytical biochemistry, 398(2), 198– 202. doi:10.1016/j.ab.2009.11.016 Kuklinsky-Sobral, J., Araújo, W. L., Mendes, R., Geraldi, I. O., Pizzirani-Kleiner, A. A., & Azevedo, J. L. (2004). Isolation and characterization of soybean-associated bacteria and their potential for plant growth promotion. Environmental microbiology, 6(12), 1244–51. doi:10.1111/j.1462-2920.2004.00658.x Landhausser, S. M., & Lieffers, V. J. (2003). Seasonal changes in carbohydrate reserves in mature northern Populus tremuloides clones. Trees, 17(6), 471–476. doi:10.1007/s00468- 003-0263-1 Lemanceau, P., Corberand, T., Gardan, L., Latour, X., Laguerre, G., Boeufgras, J., & Alabouvette, C. (1995). of two plant species, flax (Linum usitatissinum L.) and tomato (Lycopersicon esculentum Mill.), on the diversity of soilborne populations of fluorescent pseudomonads. Applied and environmental microbiology, 61(3), 1004–1012. Retrieved from http://aem.asm.org/content/61/3/1004.short Li, C., Junttila, O., & Palva, E. (2004). Environmental regulation and physiological basis of freezing tolerance in woody plants. Acta Physiologiae Plantarum, 26(2), 213–222. Retrieved from http://www.springerlink.com/index/F07618KW066VM718.pdf Lodewyckx, C., & Vangronsveld, J. (2002). Endophytic bacteria and their potential applications. Critical Reviews in …, 21(6), 583–606. doi:10.1080/0735-260291044377 Lucero, M. E., Unc, A., Cooke, P., Dowd, S., & Sun, S. (2011). Endophyte microbiome diversity in micropropagated Atriplex canescens and Atriplex torreyi var griffithsii. PloS one, 6(3), e17693. doi:10.1371/journal.pone.0017693 Manter, D. K., Delgado, J. a, Holm, D. G., & Stong, R. a. (2010). Pyrosequencing reveals a highly diverse and cultivar-specific bacterial endophyte community in potato roots. Microbial ecology, 60(1), 157–66. doi:10.1007/s00248-010-9658-x Mengoni, A., Pini, F., Huang, L.-N., Shu, W.-S., & Bazzicalupo, M. (2009). Plant-by-plant variations of bacterial communities associated with leaves of the nickel hyperaccumulator

125

Alyssum bertolonii Desv. Microbial ecology, 58(3), 660–7. doi:10.1007/s00248-009-9537- 5 Miché, L., Battistoni, F., Gemmer, S., Belghazi, M., & Reinhold-Hurek, B. (2006). Upregulation of jasmonate-inducible defense proteins and differential colonization of roots of Oryza sativa cultivars with the endophyte Azoarcus sp. Molecular plant-microbe interactions : MPMI, 19(5), 502–11. doi:10.1094/MPMI-19-0502 Miflin, B. J., & Beevers, H. (1974). Isolation of intact plastids from a range of plant tissues. Plant physiology, 53(6), 870–4. Retrieved from http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=541465&tool=pmcentrez&rend ertype=abstract Mocali, S., Bertelli, E., Di Cello, F., Mengoni, A., Sfalanga, A., Viliani, F., Caciotti, A., et al. (2003). Fluctuation of bacteria isolated from elm tissues during different seasons and from different plant organs. Research in microbiology, 154(2), 105–14. doi:10.1016/S0923- 2508(03)00031-7 Moore, F. P., Barac, T., Borremans, B., Oeyen, L., Vangronsveld, J., Van der Lelie, D., Campbell, C. D., et al. (2006). Endophytic bacterial diversity in poplar trees growing on a BTEX-contaminated site: the characterisation of isolates with potential to enhance phytoremediation. Systematic and applied microbiology, 29(7), 539–56. doi:10.1016/j.syapm.2005.11.012 Muyzer, G., & Smalla, K. (1998). Application of denaturing gradient gel electrophoresis (DGGE) and temperature gradient gel electrophoresis (TGGE) in microbial ecology. Antonie van Leeuwenhoek, 73(1), 127–41. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/9602286 Newman, L. a., Wang, X., Muiznieks, I. a., Ekuan, G., Ruszaj, M., Cortellucci, R., Domroes, D., et al. (1999). Remediation of Trichloroethylene in an Artificial Aquifer with Trees: A Controlled Field Study. Environmental Science & Technology, 33(13), 2257–2265. doi:10.1021/es981217k Niklas, K. J., & Kutschera, U. (2010). The evolution of the land plant life cycle. The New phytologist, 185(1), 27–41. doi:10.1111/j.1469-8137.2009.03054.x Noguchi, K., Hui, W. L. W., Gel, Y. R., Gastwirth, J. L., & Miao, W. (2009). lawstat: An R package for biostatistics, public policy, and law. Retrieved from http://cran.r- project.org/package=lawstat

126

Oksanen, J., Blanchet, F. G., Kindt, R., Legendre, P., Minchin, P. R., O’Hara, R. B., Simpson, G. L., et al. (2012). vegan: Community Ecology Package. Retrieved from http://CRAN.R- project.org/package=vegan Okubo, T., Ikeda, S., Kaneko, T., Eda, S., Mitsui, H., Sato, S., Tabata, S., et al. (2009). Nodulation-Dependent Communities of Culturable Bacterial Endophytes from Stems of Field-Grown Soybeans. Microbes and Environments, 24(3), 253–258. doi:10.1264/jsme2.ME09125 Olsen, G. J., Lane, D. J., Giovannoni, S. J., Stahl, D. A., & Pace, N. R. (1986). Microbial ecology and evolution: a ribosomal RNA approach. Ann. Rev. Microbiol., 40, 1078–1082. Olsvik, O., Popovic, T., & Skjerve, E. (1994). Magnetic separation techniques in diagnostic microbiology. Clinical microbiology …, 7(1). doi:10.1128/CMR.7.1.43.Updated Pace, N. R., Stahl, D. A., Lane, D. J., & Olsen, G. J. (1986). The analysis of natural microbial populations by ribosomal RNA sequences. Adv. Microbial Ecol., 9, 1–55. Patriquin, D. G., & Döbereiner, J. (1978). Light microscopy observations of tetrazolium- reducing bacteria in the endorhizosphere of maize and other grasses in Brazil. Canadian journal of microbiology, 24(6), 734–42. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/667740 Pereira, P., Ibáñez, F., Rosenblueth, M., Etcheverry, M., & Martínez-Romero, E. (2011). Analysis of the Bacterial Diversity Associated with the Roots of Maize (Zea mays L.) through Culture-Dependent and Culture-Independent Methods. ISRN Ecology, 2011, 1–10. doi:10.5402/2011/938546 Phillips, L., Greer, C. W., Farrell, R. E., & Germida, J. J. (2009). Field-scale assessment of weathered hydrocarbon degradation by mixed and single plant treatments. Applied Soil Ecology, 42(1), 9–17. doi:10.1016/j.apsoil.2009.01.002 PIllay, V. ., & Nowak, J. (1997). lnoculum density, temperature, and genotype effects on in vitro growth promotion and epiphytic and endophytic colonization of tomato. Canadian journal of microbiology, 43(4), 354–361. Possingham, J. (1980). Plastid Replication and Development in the LIfe Cycle of Higher Plants. Annual Review of Plant Physiology, 31, 113–129. Retrieved from http://www.annualreviews.org/doi/pdf/10.1146/annurev.pp.31.060180.000553 Qin, S., Chen, H.-H., Zhao, G.-Z., Li, J., Zhu, W.-Y., Xu, L.-H., Jiang, J.-H., et al. (2012). Abundant and diverse endophytic actinobacteria associated with medicinal plant Maytenus

127

austroyunnanensis in Xishuangbanna tropical rainforest revealed by culture-dependent and culture-independent methods. Environmental Microbiology Reports, 4(5), 522–531. doi:10.1111/j.1758-2229.2012.00357.x Quadt-Hallmann, A., & Kloepper, J. W. (1996). Immunological detection and localization of the cotton endophyte Enterobacter asburiae JM22 in different plant species. Canadian Journal of …, 1154, 1144–1154. Retrieved from http://www.nrcresearchpress.com/doi/abs/10.1139/m96-146 R Core Team. (2012). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. Retrieved from http://www.r-project.org/ Rasche, F., & Trondl, R. (2006). Chilling and cultivar type affect the diversity of bacterial endophytes colonizing sweet pepper (Capsicum anuum L.). Canadian journal of …, 1045, 1036–1045. doi:10.1139/W06-059 Rees, G. N., Baldwin, D. S., Watson, G. O., Perryman, S., & Nielsen, D. L. (2004). Ordination and significance testing of microbial community composition derived from terminal restriction fragment length polymorphisms: application of multivariate statistics. Antonie van Leeuwenhoek, 86(4), 339–47. doi:10.1007/s10482-004-0498-x Reiter, B., & Sessitsch, A. (2006). Bacterial endophytes of the wildflower Crocus albiflorus analyzed by characterization of isolates and by a cultivation-independent approach. Canadian journal of microbiology, 149, 140–149. doi:10.1139/W05-109 Renaut, J., Hoffmann, L., & Hausman, J.-F. (2005). Biochemical and physiological mechanisms related to cold acclimation and enhanced freezing tolerance in poplar plantlets. Physiologia Plantarum, 125(1), 82–94. doi:10.1111/j.1399-3054.2005.00554.x Rosenblueth, M., & Martínez-Romero, E. (2006). Bacterial endophytes and their interactions with hosts. Molecular plant-microbe interactions : MPMI, 19(8), 827–37. doi:10.1094/MPMI-19-0827 Ryan, R. P., Germaine, K., Franks, A., Ryan, D. J., & Dowling, D. N. (2008). Bacterial endophytes: recent developments and applications. FEMS microbiology letters, 278(1), 1– 9. doi:10.1111/j.1574-6968.2007.00918.x Saito, A., Ikeda, S., Ezura, H., & Minamisawa, K. (2007). Microbial community analysis of the phytosphere using culture-independent methodologies. Microbes and Environments, 22(2), 93–105. Retrieved from http://joi.jlc.jst.go.jp/JST.JSTAGE/jsme2/22.93?from=Google

128

Sakai, M., Matsuka, A., Komura, T., & Kanazawa, S. (2004). Application of a new PCR primer for terminal restriction fragment length polymorphism analysis of the bacterial communities in plant roots. Journal of microbiological methods, 59(1), 81–9. doi:10.1016/j.mimet.2004.06.005 Santarius, K. (1973). The Protective Effect of Sugars on Chloroplast Membranes during Temperature and Water Stress and Its Relationship to Frost, Desiccation and Heat Resistance. Planta, 114, 105–114. Retrieved from http://www.springerlink.com/index/L77M416543827271.pdf Schütte, U. M. E., Abdo, Z., Bent, S. J., Shyu, C., Williams, C. J., Pierson, J. D., & Forney, L. J. (2008). Advances in the use of terminal restriction fragment length polymorphism (T- RFLP) analysis of 16S rRNA genes to characterize microbial communities. Applied microbiology and biotechnology, 80(3), 365–80. doi:10.1007/s00253-008-1565-4 Sessitsch, A., Reiter, B., Pfeifer, U., & Wilhelm, E. (2002). Cultivation-independent population analysis of bacterial endophytes in three potato varieties based on eubacterial and Actinomycetes-specific PCR of 16S rRNA genes. FEMS microbiology ecology, 39(1), 23– 32. doi:10.1111/j.1574-6941.2002.tb00903.x Shah, J. (2009). Plants under attack: systemic signals in defence. Current opinion in plant biology, 12(4), 459–64. doi:10.1016/j.pbi.2009.05.011 Stewart, E. J. (2012). Growing unculturable bacteria. Journal of bacteriology, 194(16), 4151– 60. doi:10.1128/JB.00345-12 Stoltzfus, J., So, R., & Malarvithi, P. (1997). Isolation of endophytic bacteria from rice and assessment of their potential for supplying rice with biologically fixed nitrogen. Plant and Soil, 194, 25–36. Retrieved from http://www.springerlink.com/index/W955HR02K216138H.pdf Strobel, G., Daisy, B., Castillo, U., & Harper, J. (2004). Natural products from endophytic microorganisms. Journal of natural products, 67(2), 257–68. doi:10.1021/np030397v Sun, L., Qiu, F., Zhang, X., Dai, X., Dong, X., & Song, W. (2008). Endophytic bacterial diversity in rice (Oryza sativa L.) roots estimated by 16S rDNA sequence analysis. Microbial ecology, 55(3), 415–24. doi:10.1007/s00248-007-9287-1 Suzuki, R., & Shimodaira, H. (2011). pvclust: Hierarchical Clustering with P-Values via Multiscale Bootstrap Resampling.

129

Taghavi, S., Barac, T., Greenberg, B., Borremans, B., Vangronsveld, J., & Lelie, D. Van Der. (2005). Horizontal Gene Transfer to Endogenous Endophytic Bacteria from Poplar Improves Phytoremediation of Toluene. Society, 71(12), 8500–8505. doi:10.1128/AEM.71.12.8500 Takahashi, Y., Matsumoto, A., Morisaki, K., & Omura, S. (2006). Patulibacter minatonensis gen. nov., sp. nov., a novel actinobacterium isolated using an agar medium supplemented with superoxide dismutase, and proposal of Patulibacteraceae fam. nov. International journal of systematic and evolutionary microbiology, 56(Pt 2), 401–6. doi:10.1099/ijs.0.63796-0 Tervet, J. W., & Hollis, J. P. (1948). Bacteria in the storage organs of healthy plants. Phytophathology, 38, 960–962. Tesar, M., Reichenauer, T., & Sessitsch, A. (2002). Bacterial rhizosphere populations of black poplar and herbal plants to be used for phytoremediation of diesel fuel. Soil biology and biochemistry, 34, 1883–1892. Retrieved from http://www.sciencedirect.com/science/article/pii/S003807170200202X Thomas, P., & Soly, T. a. (2009). Endophytic bacteria associated with growing shoot tips of banana (Musa sp.) cv. Grand Naine and the affinity of endophytes to the host. Microbial ecology, 58(4), 952–64. doi:10.1007/s00248-009-9559-z Ulrich, K., Ulrich, A., & Ewald, D. (2008). Diversity of endophytic bacterial communities in poplar grown under field conditions. FEMS microbiology ecology, 63(2), 169–80. doi:10.1111/j.1574-6941.2007.00419.x Vargas, C., Muniz de Padua, V. L., De Matos Nogueira, E., Vinagre, F., Masuda, H. P., Rodrigues da Silva, F., Baldani, J. I., et al. (2003). Signaling pathways mediating the association between sugarcane and endophytic diazotrophic bacteria: A genomic approach. Symbiosis, 35, 159–180. Retrieved from http://cat.inist.fr/?aModele=afficheN&cpsidt=14707953 Wang, H.-X., Geng, Z.-L., Zeng, Y., & Shen, Y.-M. (2008). Enriching plant microbiota for a metagenomic library construction. Environmental microbiology, 10(10), 2684–91. doi:10.1111/j.1462-2920.2008.01689.x Wicke, S., Schneeweiss, G. M., Depamphilis, C. W., Müller, K. F., & Quandt, D. (2011). The evolution of the plastid chromosome in land plants: gene content, gene order, gene function. Plant molecular biology, 76(3-5), 273–97. doi:10.1007/s11103-011-9762-4

130

Wilson, K., & Walker, J. (Eds.). (2010). Biology, Principles and Techniques of Biochemistry and Molecular (7th ed., p. 760). New York, NY: Cambridge University Press. Yousaf, S., Ripka, K., Reichenauer, T. G., Andria, V., Afzal, M., & Sessitsch, a. (2010). Hydrocarbon degradation and plant colonization by selected bacterial strains isolated from Italian ryegrass and birdsfoot trefoil. Journal of applied microbiology, 109(4), 1389–401. doi:10.1111/j.1365-2672.2010.04768.x Yrjala, K., & Mancano, G. (2010). The incidence of Burkholderia in epiphytic and endophytic bacterial cenoses in hybrid aspen grown on sandy peat. Boreal environment …, 6095(February), 81–96. Retrieved from http://cat.inist.fr/?aModele=afficheN&cpsidt=22477170 Zammatteo, N., Alexandre, I., Ernest, I., Le, L., Brancart, F., & ́ R. (1997). Comparison between microwell and bead supports for the detection of human cytomegalovirus amplicons by sandwich hybridization. Analytical biochemistry, 253(2), 180–9. doi:10.1006/abio.1997.2352 Zimbro, M. J., Power, D. A., Miller, S. M., Wilson, G. E., & Johnson, J. A. (Eds.). (2009). Difco & BBL Manual - Manual of Microbiological Culture Media. Manual of Microbiological Culture Media. BD (Second.). Maryland.

Appendix A – RDP Database Information

Table A.1: Bacterial hits with restriction sites for enzymes PvuII and MscI in the RDP Database (*hits as of August 3, 2012), using sequences of good quality and >1200bp.

Enzyme PvuII (number of hits/total) MscI (number of hits/total) PvuII and MscI combined (number of hits/total)

Bacteria hits domain Bacteria domain Bacteria domain Bacteria (based on (62244/1173481) (84757/1173481) (8346/1173481) phylum) phylum "Actinobacteria" phylum "Actinobacteria" phylum "Actinobacteria" (2558/175006) (4488/175006) (43/175006) phylum "Aquificae" (33/928) phylum "Aquificae" (152/928) phylum "Aquificae" (10/928) phylum "Bacteroidetes" phylum "Bacteroidetes" phylum "Bacteroidetes" (16732/134129) (16043/134129) (2970/134129) phylum "Caldiserica" phylum "Caldiserica" (1/219) phylum "Caldiserica" (1/219) (215/219) phylum "Chlamydiae" phylum "Chlamydiae" phylum "Chlamydiae" (176/403) (176/403) (398/403) phylum "Chlorobi" (117/1005) phylum "Chlorobi" (9/1005) phylum "Chlorobi" (30/1005) phylum "Chloroflexi" phylum "Chloroflexi" phylum "Chloroflexi" (3003/19933) (89/19933) (1213/19933) phylum "Chrysiogenetes" phylum "Chrysiogenetes" phylum "Chrysiogenetes" (0/11) (0/11) (0/11) phylum "Deferribacteres" phylum "Deferribacteres" phylum "Deferribacteres" (11/366) (0/366) (15/366) phylum "Deinococcus- phylum "Deinococcus- phylum "Deinococcus- Thermus" (106/1709) Thermus" (13/1709) Thermus" (388/1709) phylum "Dictyoglomi" (0/22) phylum "Dictyoglomi" (0/22) phylum "Dictyoglomi" (1/22) phylum "Elusimicrobia" phylum "Elusimicrobia" phylum "Elusimicrobia" (13/140) (0/140) (13/140) phylum "Fibrobacteres" phylum "Fibrobacteres" phylum "Fibrobacteres" (25/279) (13/279) (66/279) phylum "Fusobacteria" phylum "Fusobacteria" phylum "Fusobacteria" (21/9348) (0/9348) (98/9348) phylum "Gemmatimonadetes" phylum "Gemmatimonadetes" phylum "Gemmatimonadetes" (875/1133) (32/1133) (92/1133) phylum "Lentisphaerae" phylum "Lentisphaerae" phylum "Lentisphaerae" (49/1656) (1/1656)

131 132

(202/1656) phylum "Nitrospira" (79/1165) phylum "Nitrospira" (0/1165) phylum "Nitrospira" phylum "Planctomycetes" phylum "Planctomycetes" (65/1165) (2065/10868) (303/10868) phylum "Planctomycetes" phylum "Proteobacteria" phylum "Proteobacteria" (861/10868) (16634/317416) (1074/317416) phylum "Proteobacteria" phylum "Spirochaetes" phylum "Spirochaetes" (15908/317416) (3112/9165) (16/9165) phylum "Spirochaetes" phylum "Synergistetes" phylum "Synergistetes" (698/9165) (611/1090) (423/1090) phylum "Synergistetes" phylum "Tenericutes" phylum "Tenericutes" (664/1090) (322/2918) (20/2918) phylum "Tenericutes" phylum phylum (105/2918) "Thermodesulfobacteria" "Thermodesulfobacteria" phylum (1/104) (1/104) "Thermodesulfobacteria" phylum "Thermotogae" phylum "Thermotogae" (94/104) (11/575) (1/575) phylum "Thermotogae" phylum BRC1 (46/395) phylum BRC1 (2/395) (60/575) phylum OD1 (13/137) phylum OD1 (1/137) phylum BRC1 (21/395) phylum OP11 (1/93) phylum OP11 (0/93) phylum OD1 (15/137) phylum SR1 (1/227) phylum SR1 (0/227) phylum OP11 (51/93) phylum TM7 (100/2052) phylum TM7 (0/2052) phylum SR1 (4/227) phylum WS3 (13/521) phylum TM7 (23/2052) phylum "Armatimonadetes" phylum WS3 (7/521) phylum WS3 (16/521) (430/986) phylum "Armatimonadetes" phylum "Verrucomicrobia" phylum "Armatimonadetes" (47/986) (2329/9083) (18/986) phylum "Verrucomicrobia" phylum "Acidobacteria" phylum "Verrucomicrobia" (578/9083) (1223/13260) (60/9083) phylum "Acidobacteria" phylum Firmicutes phylum "Acidobacteria" (1607/13260) (20876/403727) (202/13260) phylum Firmicutes phylum phylum Firmicutes (14252/403727) Cyanobacteria/Chloroplast (1467/403727) phylum (8123/18799) phylum Cyanobacteria/Chloroplast Cyanobacteria/Chloroplast (2740/18799) (1144/18799)

Chloroplast genus Streptophyta (1501/1801)genus Streptophyta (469/1801) genus Streptophyta (375/1801) Hits

133

Table A.2: List of some of the bacterial species or genera that have restriction sites corresponding to enzymes PvuII and MscI in the RDP database. Restriction PvuII MscI PvuII and MscI Enzyme combined Bacterial Actinomyces Corynebacterium Rhodococcus genus or Corynebacterium Mycobacterium Microbacterium species Dietzia Rhodococcus Kocuria targeted Rhodococcus Bacillus Streptomyces Glycomyces Kineococcus Bacteroides Brachybacterium Cellulomonas Deinococcus Tetrasphaera Agrococcus Methylobacterium lxb-3 Agrococcus Frigoribacterium Roseobacter Curtobacterium Microbacterium Paracoccus Leucobacter Plantibacter Rhodobacter Microbacterium Leifsonai sp. CJ-G-R2A8 Rhodovulum Micrococcus Okibacterium Sphigomonas Arthrobacter Arthrobacter Citrobacter freundii Kocuria Kocuria Enterobacter hormaechei Dactylosporangium Micromonospora Nocardioides Amycolatopsis Friedmanniella Streptomyces Microlunatus Micropolyspora Propinicicella Desulfurobacterium Streptomyces Bacteroides Bacteroides Porphyromonas Prevotella Prevotella Flavobacterium Flavobacterium Flavisolibacter Sphingobacterium Pedobacter Dehalococcoides Sphingobacterium Denitrovibrio

134

Nitrospira Deinococcus Rhodopirellula Planctomyces Bradyrhizobium Brevundimonas diminuta Rhodopseudomonas Methylobacterium Methylobacterium Mesorhizobium Mesorhizobium Rhizobium Rhizobium Paracoccus Nitratireductor Rhodobacter Rhodobacter Rhodovulum Rhodovulum Roseobacter Roseobacter Roseomonas Wolbachia Rhodospirillium Sphingomonas Burkholderia sp. Novosphingobium Ralstonia Burkholderia Variovorax Duganella Methylobacillus Acidovorax Methylophilus Thiobacillus Methylovorus Thiobacillus Salinimonas Denitratisoma Nitrosococcus Desulfurivibrio Seratia marcescens Aeromonas Pantoea sp JS-2 Ferrimonas Buchnera Pseudoalteromonas Methylococcus Shewanella Mannheimia Thiorhodococcus Pasteurella Citrobacter Acinetobacter Pantoea olea Pseudomonas sp. Serratia Pseudoxanthomonas Methylobacter Xantomonas campestris Pseudomonas sp. Paenibacillus

135

Vibrio Lysinibacillus Stenotrophomonas Staphylococcus Xanthomonas sp. Lactobacillus Alcanivoraceae Streptococcus Bacillus sp. Clostridium Paenibacillus Vibrio spp. Staphylococcus Lactobacillus Streptococcus Acetobacterium Clostridium

Appendix B – Bacterial MspI Restriction Fragment Sizes

Table B.1: List of endophytic bacterial species downloaded from RDP database and the corresponding forward terminal MspI restriction digested fragment from the amplicons amplified with primers 27F-1492R. Size of Forward Terminal MspI restriction digested Bacterial Species from RDP database fragment Bacillus safensis 80 Geodermatophilus sp. 124 Microbacterium sp. 11.2 125 Kineococcus gynurae 140 Bacillus firmus 145 Sanguibacter inulins 145 Arthrobacter sp. 147 Methylobacterium adhaesivum 150 Methylobacterium sp. PB282 150 Paenibacillus lautus 150 Sphingomonas aerolata 150 Paenibacillus sp. HDDMM03 150 Bacillus simplex 152 Sphingomonas sp. SPC-1 152 Staphylococcus sp. AMF3826 154 Kineococcus sp. YIM 158 Frigoribacterium sp. PDD 162 Frigoribacterium sp. LPPA 163 Plantibacter sp. S51 163 Frigoribacterium sp. ULA1 164 Clavibacter michiganensis 165 Bacillus sp.E3 166 Micrococcineae bacterium 167

136 137

Brevibacterium sp. 167 Curtobacterium flaccumfaciens 175 Curtobacterium sp. U17 175 Xanthomonas sp. B05 192 Wiliamsia sp B-1-24 202 Williamsia sp. NRRL B 202 Friedmanniella sp. EL-17a 203 Deinococcus sp. 267 Microbacterium phyllosphaerae 280 Kocuria sp. 280 Sinorhizobium medicae 401 Ralstonia sp. 435 Pseudomonas rhizosphaerae 489 Pseudomonas sp. NCCP 489 Pseudomonas fulva 490 Variovorax paradoxus 491 Xanthomonas arboricola 497

Table B.2: Database of cultured endophytic bacterial species from Acer negundo and Ulmus spp. and their corresponding forward terminal MspI restriction digested fragment from the amplicons amplified with primers 27F-1492R. Size of Forward Terminal MspI restriction digested Cultured Bacterial Endophyte fragment Amnibacterium sp. 80 Bacillus spp. 80 Rahnella spp. 87 Microbacterium sp. 124 Burkholderia spp. 140 Bacillus spp. 142 Sanguibacter sp 147

138

Paenibacillus spp. 149 Sphingomonas sp. 149 Methylobacterium sp. 150 Paenibacillus sp. 150 Bradyrhizobium sp. 151 Bacillus spp. 152 Microbacterium spp. 152 Staphylococcus sp. 155 Kineococcus sp. 158 Friedmanniella sp. 160 Friedmanniella sp. 161 Frigoribacterium spp. 161 Plantibacter sp. 163 Clavibacter sp. 165 Bacillus spp. 166 Brevibacterium sp. 169 Curtobacterium spp. 174 Bradyrhizobium sp. 198 Williamsia sp. 201 Rhizomonas sp. 210 Microbacterium spp. 279 Leucobacter sp. 282 Kocuria spp. 285 Devosia sp. 436 Paracoccus spp. 437 Dyella sp. 462 Rhizobium sp. 463 Chryseobacterium sp. 484 Massilia spp. 487 Pseudomonas spp. 489 Variovorax sp. 490

139

Pseudomonas spp. 491 Erwinia sp. 493 Xanthomonas sp. 497

Appendix C – Cultured Bacterial Phyla Abundances

Table C.1: Percentage of bacterial phyla cultured in each season in R2A and TSA media.

Winter Summer Autumn

Bacterial Phyla R2A TSA R2A TSA R2A TSA

Actinobacteria 42.00 50.08 66.62 70.88 71.14 64.85

Firmicutes 23.88 23.39 6.67 5.53 7.21 4.67

Bacteroidetes 0.27 0 0.18 0 4.38 2.72

Deinococcus-Thermus 1.59 0 0 0 0 0

Proteobacteria

Alphaproteobacteria 13.08 6.52 12.32 2.27 13.99 3.87

Betaproteobacteria 0.03 0 0 0 0 0

Gammaproteobacteria 17.88 15.16 0.41 1.92 0.09 10.32

Unknown 1.28 4.84 13.79 19.40 3.19 14.19

140 141

Table C.2: Percentage of bacterial phyla cultured in each plant species in R2A and TSA media.

Acer negundo Ulmus parvifolia Ulmus pumila

Bacterial Phyla R2A TSA R2A TSA R2A TSA

Actinobacteria 61.79 62.47 58.02 58.30 66.05 70.54

Firmicutes 7.21 7.87 18.66 14.76 9.89 5.17

Bacteroidetes 3.17 0.95 1.96 2.07 0 0

Deinococcus-Thermus 0 0 1.3 0 0 0

Proteobacteria

Alphaproteobacteria 16.85 4.32 10.73 3.57 10.67 5.04

Betaproteobacteria 0.02 0 0 0 0 0

Gammaproteobacteria 7.55 4.50 3.05 14.56 3.63 9.57

Unknown 3.37 19.89 6.29 6.71 9.76 9.66

Appendix D – Hierarchical Clustering D.1 Culture Dependent Analyses – Hierarchical Clustering

Figure D.1a: Hierarchical clustering of bacterial community profiles collected from R2A media separated based on the season: Winter. Dendrograms include approximately unbiased (AU) p- values in red and bootstrap probability (BP) values in blue, with samples labeled with M represent Acer negundo, EC represent Ulmus parvifolia and ES represent Ulmus pumila. Branches outlined in red represent clusters with ≥95% AU values.

142 143

144

145

146

147

148

149

150

151

152

D.2 Culture Independent Analyses – Hierarchical Clustering

153

Figure D.5b: Hierarchical clustering of bacterial community profiles collected from forward T- RFs data separated based on the season: Summer. Dendrograms include approximately unbiased (AU) p-values in red and bootstrap probability (BP) values in blue, with samples labeled with M represent Acer negundo, EC represent Ulmus parvifolia and ES represent Ulmus pumila. Branches outlined in red represent clusters with ≥95% AU values.

154

155

156

157

158

159

160

161