Research Collection

Doctoral Thesis

Post-Translational and Transcriptional Regulation of Storage Glucan Metabolism in Model and Non-Model Plants

Author(s): Umhang, Martin

Publication Date: 2011

Permanent Link: https://doi.org/10.3929/ethz-a-6618292

Rights / License: In Copyright - Non-Commercial Use Permitted

This page was generated automatically upon download from the ETH Zurich Research Collection. For more information please consult the Terms of use.

ETH Library DISS. ETH Nr. 19562

Post-Translational and Transcriptional Regulation of Storage Glucan Metabolism in Model and Non-Model Plants

A dissertation submitted to the

ETH ZÜRICH

For the Degree of Doctor of Science

Presented by

Martin Umhang

MSc in Biology, University of Bern, Switzerland

Born the 20 th July, 1980 in Thun, Switzerland

Accepted on the recommendation of

Prof. Samuel C. Zeeman, examiner Prof. Julia Vorholt, co-examiner Dr Lee Sweetlove, co ‐‐‐examiner

2011

Acknowledgements

I would like to thank Prof. S. C. Zeeman for giving me the opportunity to do my PhD in his lab. Sam has provided me with inputs for my scientific work and gave advice when I needed it. He gave me the possibility to explore exciting new areas in research and helped me to tackle the challenges. I am very grateful for the time I could work with him. I have certainly learned a lot; not only in scientific terms – thank you!

I am thankful to Sebastian Streb; besides having helped a lot in the analyses of the C. peltata glucan structures and sugars, he always provided new ideas and helped to critically review results.

My thanks go to Oliver Kötting; he has provided the SEX4 protein as well as the P-oligos I have been able to use. I am glad to see that he is following up some of my work. I am also very grateful to Sang- Kyu Lee, Michaela Stettler, Heike Reinhold and Sylvain Bischof; they helped in some of my experiments, did initial experiments or provided materials I could profit from.

I have very much appreciated the work done by Simona Eicke, who did all the TEM work and helped in many other aspects of daily lab business. Our gardeners, Sabine Klarer and André Imboden, have been invaluable, taking care of my Arabdopsis or C. peltata plants.

Weihong Qi, Marzanna Küenzli and Paolo Nanni from the FGCZ have helped a lot with either the C. peltata transcriptome sequencing or the mass spectrometry analyses. I am very thankful for all the work they have done to support my projects.

Furthermore, I would like to thank all members of the ‘Plant Biochemistry’ group. I have spent a great time in the D36 and B33 labs. It was a pleasure to work with you, Carmen, Lilly or Ernst, just to name a few among many.

I am grateful to Alex Graf and Jychian Chen; I have been able to use some of their results and materials for my thesis.

My family and my friends have always supported me in my plans and projects – I am very thankful for the many relaxing hours; past or yet to come

Finally, I feel very grateful to my girlfriend, Sonja Pfammatter. She encouraged me when things did not work well and she has shared my joy about my achievements. She had to share my attention with a time consuming PhD project but has always supported me in my dreams. Thank you! SUMMARY ...... 1

ZUSAMMENFASSUNG ...... 3

ABBREVIATIONS ...... 5

1. INTRODUCTION ...... 7 1.1 Starch - A Vital Plant for Mankind ...... 7 1.2 Starch Is the Major Carbon Storage Compound in Plants ...... 8 1.3 The Structure and Composition of Starch Granules ...... 9 1.3.1 Starch Granule Morphology ...... 9 1.3.2 Starch Composition ...... 10 1.3.3 Starch Granule Architecture ...... 11 1.4 Starch Synthesis in Autotrophic and Heterotrophic Tissue ...... 13 1.5 The Synthesis of Amylopectin and Amylose ...... 17 1.5.1 Starch Synthases ...... 17 1.5.2 Starch Branching ...... 18 1.5.3 Debranching Enzymes in Starch Synthesis ...... 19 1.5.4 Protein Complex Formation among Starch Synthesizing Proteins...... 20 1.5.5 The Architecture and Structure of ...... 21 1.6 Factors Affecting the Crystallinity of the Starch Granule ...... 22 1.7 Starch Degradation in Leaves ...... 23 1.7.1 Starch Phosphorylation ...... 23 1.7.2 Degradation of Linear Chains in Starch Degradation ...... 26 1.7.3 Debranching Enzymes in Starch Degradation ...... 27 1.7.4 Starch De-Phosphorylation ...... 28 1.7.5 Metabolism of Starch Breakdown Products ...... 28 1.8 Cecropia peltata , Glycogen Synthesis in a Vascular Plant ...... 29 1.8.1 Cecropia peltata – A Many-Sided Subtropical Tree ...... 29 1.8.2 Müllerian Bodies – Myrmecophytic Food Structures of Cecropia peltata ...... 30 1.8.3 Glycogen Deposition in Müllerian Bodies ...... Fehler! Textmarke nicht definiert. 1.9 Scope of the Work Presented ...... 33

2. MATERIAL AND METHODS ...... 34 2.1 Plant material ...... 34 2.2 Standard SDS-PAGE and Western Blotting ...... 35 2.3 Native PAGE and native PAGE blotting ...... 35 2.4 Co-Immunoprecipitation of BAM1 and LSF1 ...... 36 2.5 Gel Filtration Chromatography ...... 37 2.6 Chloroplast Isolation and Measurement of GAPDH and PEP-Carboxylase Activities...... 37 2.7 Carbohydrate Extraction and Measurements ...... 38 2.7.1 Quantification of Starch and WSP ...... 39 2.7.2 Chain Length Distributions of Starch and Water Soluble Polyglucans ...... 39 2.8 High pH Anion Exchange Chromatography Coupled to Pulsed Amperometric Detection ...... 40 2.9 Plant Transformation and Complementation ...... 41 2.10 Tandem Affinity Purification and Myc-CoIP...... 42 2.11 Mass-Spectrometric Analyses of Tandem-Affinity-Purified Protein Samples ...... 43 2.12 Purification and Digestion of Phospho-Oligosaccharides ...... 44 2.13 Digestion of Amylopectin Using BAM1-TAP( bam1 ) and BAM1-TAP( lsf1 ) ...... 45 2.14 Measurement of β-Amylase Activity ...... 45 2.15 Transmission Electron Microscopy of C. peltata leaves and Müllerian Bodies ...... 46 2.16 mRNAseq of C. peltata Leaves and Müllerian Bodies ...... 46 2.16.1 C. peltata RNA Extraction, cDNA Synthesis and Normalization ...... 46 2.16.2 Library preparation for Illumina and 454 Pyrosequencing ...... 48 2.16.3 Assembly of 454 Contigs and Illumina Reads ...... 48 2.16.4 Short Read Mapping and Transcript Annotation ...... 49 2.16.5 Read Count Normalization, Identification of Differentially Expressed Transcripts and GO Enrichment Analysis ...... 50

3. PROTEIN-PROTEIN INTERACTIONS BETWEEN PROTEINS INVOLVED IN STARCH DEGRADATION ...... 51 3.1 Introduction ...... 51 3.2 Results ...... 52 3.2.1 Gel Filtration Chromatography – Column Calibration ...... 52 3.2.2 Gel Filtration Chromatography of Proteins Involved in Starch Degradation ...... 54 3.3 Discussion...... 55 3.3.1 Enzymes Involved in Starch Degradation Are Present as HMW Forms ...... 55 3.3.2 14-3-3 Proteins - Potential Interaction Partners of Proteins Involved in Starch Degradation ...... 58 3.3.3 Limitations of a GFC Based Approach ...... 59 3.3.4 A Knowledge-Based Decision to Focus on BAM1 and LSF1 to Confirm and Characterize Complex Formation ...... 59

4. THE INTERACTION NETWORK OF BAM1 AND LSF1 ...... 60 4.1 Introduction ...... 60 4.2 Results ...... 63 4.2.1 bam1 and lsf1 Mutants Lack One Common β-Amylase Activity ...... 63 4.2.2 The Electrophoretic Mobility and Native Molecular Weight of BAM1 and LSF1 Depend on LSF1 and BAM1 Respectively...... 66 4.2.2.1 BAM1 and LSF1 are Present in High Molecular Weight Complexes ...... 67 4.2.2.2 BAM1 and LSF1 are Present in the Same Complex ...... 69 4.2.3 Reciprocal TAP-tagging Confirms the BAM1-LSF1 Interaction ...... 70 4.2.3.1 Complementation of Activities on Native Gels ...... 70 4.2.3.2 Complementation of sex Mutant Phenotypes ...... 73 4.2.3.3 Confirmation of Interaction of BAM1 and LSF1 ...... 75 4.2.3.4 BAM1 and LSF1 Interact with a Plastid Localized NAD-Dependent Malate-Dehydrogenase ...... 79 4.2.4 LSF1 lacks an NAD-dependent MDH activity...... 82 4.2.5 Functional Characterization of the BAM1-LSF1 Complex ...... 83 4.2.5.1 BAM1 and LSF1 Interact in the Chloroplast ...... 83 4.2.5.2 Redox Dependence of the BAM1-LSF1 Complex ...... 86 4.2.5.3 BAM1-LSF1 Cannot Dephosphorylate Phosphorylated Oligosaccharides...... 88 4.2.5.4 Identification of BAM1 Phospho-Peptides ...... 89 4.2.5.5 LSF1 Increases the β-Amylolytic Activity of BAM1-TAP ...... 90 4.2.6 Characterization of bam1 /lsf1 , bam3 /lsf1 and bam1 /bam3 /lsf1 Multiple Mutants ...... 92 4.3 Discussion...... 95 4.3.1 BAM1 and LSF1 Are Present in a Common Complex ...... 95 4.3.2 LSF1 Interacts With BAM3 ...... 97 4.3.3 A Plastid Localized NAD-dependent Malate Dehydrogenase is Present in the BAM1-LSF1 Complex ...... 97 4.3.4 Plastid localized NAD-dependent Malate Dehydrogenase – The Malate-Valve ...... 98 4.3.5 Models to Explain the Starch Excess Phenotype of lsf1 Mutants – BAM Stimulation versus the Malate Valve in the Dark ...... 101 4.3.6 Investigation of Multiple Mutants to Characterize the Function of LSF1 in Starch Degradation ...... 102 4.4 Conclusion and Outlook ...... 103

5. CECROPIA PELTATA – GLYCOGEN SYNTHESIS IN A VASCULAR PLANT ...... 105 5.1 Introduction ...... 105 5.1.1 C. peltata as a Model System for Water Soluble Synthesis in Plants ...... 105 5.1.2 Approaches to Characterize Müllerian Body Metabolism ...... 105 5.1.3 Transcriptome Sequencing in Non-Model Species ...... 106 5.1.3.1 Next Generation Sequencing Technologies ...... 106 5.1.3.2 Considerations for Next Generation Sequencing Transcriptome Analysis ...... 108 5.2 Results Part I – Biochemical and Enzymatic Characterization of Müllerian Bodies ...... 111 5.2.1 C. peltata ; Plant Growth, Propagation and Development of Myrmecophytic Traits ...... 111 5.2.2 Cellular and Subcellular Structures of Leaves and Müllerian Bodies ...... 112 5.2.3 Soluble and Insoluble Polyglucan Levels ...... 114 5.2.4 Structural Characterization of Polyglucans ...... 115 5.2.5 Soluble Sugar Levels ...... 120 5.2.6 Starch Modifying Activities in C. peltata Leaves and Müllerian Bodies ...... 120 5.2.7 Conclusions ...... 125 5.2.7.1 Müllerian Bodies Accumulate Large Amounts of WSP ...... 125 5.2.7.2 CLD Profiles are Insufficient to Explain Müllerian Body WSP Accumulation ...... 125 5.2.7.3 Branch Points are Located Close to Each Other in Müllerian Body WSP ...... 126 5.2.7.4 Starch Branching Enzyme and Starch Synthase Activities are higher in Müllerian Bodies as Compared to Leaves...... 126 5.3 Results Part II – Characterization of the C. peltata Transcriptome ...... 128 5.3.1 454 Sequencing of a Normalized C. peltata cDNA Library and Pre-Assembly ...... 128 5.3.2 Sequencing of Non-Normalized Leaf and Müllerian Body cDNA Libraries ...... 129 5.3.3 Mixed Sequence Assembly ...... 130 5.3.4 Mapping Reads to the Reference Transcripts and Identification of Differentially Expressed Transcripts ...... 131 5.3.4.1 Read Mapping and Quality Control ...... 131 5.3.4.2 Normalization of Read Counts ...... 134 5.3.4.3 Identification of Differentially Expressed Transcripts using ‘edgeR’ ...... 134 5.3.5 Annotation of Transcripts and Gene Ontology Enrichment Analysis ...... 136 5.3.5.1 Transcript Annotation by BLAST Searches ...... 136 5.3.5.2 GO Analysis of Transcripts Uniquely Expressed or Over-Expressed in Leaves ...... 137 5.3.5.3 GO Analysis of Transcripts Uniquely Expressed or Over-Expressed in Müllerian Bodies ...... 139 5.3.6 A Close-Up View on Starch Metabolic Genes in Müllerian Bodies ...... 140 5.3.6.1 Expression of Transcripts with Homology to Arabidopsis Genes Involved in Starch Degradation ...... 141 5.3.6.2 Expression of Transcripts with Homology to Arabidopsis Genes Involved in Starch Synthesis ...... 142 5.3.6.3 Expression of Transcripts with Homology to Arabidopsis Genes Involved in Sucrose Metabolism and Hexose-Phosphate Import into Plastids ...... 144 5.3.7 Conclusions ...... 146 5.3.7.1 NGS Provides an Overview of the C. peltata Transcriptome ...... 146 5.3.7.2 Transcriptomes of C. peltata Leaves and Müllerian Bodies are Highly Different ...... 146 5.3.7.3 Müllerian Bodies: A Rapidly Growing, Heterotrophic Tissue ...... 147 5.3.7.4 Genes Encoding Proteins Involved in Sucrose Catabolism and Glc-6P Import into Plastids are Over-Expressed in Müllerian Bodies ...... 148 5.3.7.5 WSP Accumulation in Müllerian Bodies; Triggered by Altered Starch Synthase Isoform Expression? ...... 149 5.4 Synthesis of Part I and Part II ...... 150

6. GENERAL CONCLUSIONS ...... 153 6.1 Proteins Involved in Starch Degradation are Present in High Molecular Weight Complexes ...... 153 6.2 BAM1, LSF1 and p-NAD-MDH: The ‘When’ ‘Where’ and ‘How Much’ of Protein-Protein Interactions ...... 154 6.3 Local Redox Modulation by p-NAD-MDH ...... 156 6.4 Transcript Profiling of Non-Model Species Using RNA-Seq ...... 157 6.5 The Limits of Transcriptomics – The Next Steps ...... 158 6.6 Outlook...... 159

7. REFERENCES ...... 160

8. APPENDIX ...... 175

9. CURRICULUM VITAE ...... 181

Summary

Starch is the major storage carbohydrate in plants and a key renewable resource for mankind. During photosynthesis, plants store their carbon in the form of starch to break it down again at times when no photosynthesis is possible. Starch consists almost entirely of glucan chains, where single units are α(1,4) linked to form linear chains. These chains are branched via α(1,6) linkages to form tree-like glucose polymers. In its composition as well as in its function starch is similar to glycogen, the major storage carbohydrate of bacteria, fungi and animals. The key difference between starch and glycogen is that starch accumulates as large, insoluble, semi-crystalline granules, while glycogen accumulates as small, soluble particles. Current structural models for starch and glycogen suggest that the frequency and local distribution of α(1,6) branch points as well as the glucan chain length distribution determine the crystallization competence of a polyglucan and are key factors which are different between starch and glycogen. The synthesis of starch and glycogen requires the activities of two different types of enzymes: synthases, which elongate glucan chains and branching enzymes, which introduce branch points. The synthesis of starch is more complex and a third enzyme activity is required to synthesize semi-crystalline granules: debranching enzymes, which remove excess branch points. Not only the synthesis of starch requires the coordinated activities of different enzyme activities, also its degradation is complex. The major enzyme activities required for starch degradation in plant leaves have only been identified in the past years. Among those are β-amylases: glucan , releasing maltose from starch granules, as well as enzymes mediating transient phosphorylation and dephosphorylation of starch. A process, thought to disrupt the tight packing of glucan chains at the granular surface, which is crucial for starch degradation. Both, degradation and synthesis of starch are complex processes. We do not yet fully understand the details of regulation and coordination of enzymes involved in starch degradation or the relative importance of the different enzyme activities for starch synthesis. In my thesis, I have used two different approaches to address these problems. In a first part of my thesis, I have investigated the coordination of starch breakdown by protein-protein interactions. In a screening approach I have looked for proteins, involved in starch degradation, which are present in high-molecular-weight forms. By this approach, I have identified a novel protein-protein interaction between a β-amylase (BAM1) and a putative phospho-glucan phosphatase (LSF1). I confirmed the interaction of BAM1 and LSF1 by different affinity purifications of each protein and subsequent identification of their interaction partners by mass-spectrometry. In addition to BAM1, LSF1 interacts with another β-amylase, BAM3, and a plastid localized malate dehydrogenase. I propose that the interaction of β-amylases with the starch-binding LSF1 protein helps to provide efficient and controlled starch degradation. I hypothesize that LSF1 targets BAM1 and BAM3 to the granule surface, enhancing β-amylolysis. Alternatively, LSF1 could also be involved in redox control in the chloroplast through its interaction with the malate dehydrogenase and thus influence starch metabolism indirectly. Protein-protein interactions between starch degrading enzymes have not previously been described and potentially provide a hitherto unknown mechanism for the regulation and coordination of starch breakdown. In the second part of my thesis, I have addressed the question of how plants manage to synthesize semi-crystalline starch granules as opposed to soluble glycogen. Several isoforms of all starch synthetic enzymes exist in plants. To disentangle their relative contributions to the synthesis of the crystallization-competent structure of starch, I investigated polyglucan metabolism in a subtropical tree, Cecropia peltata . This tree accumulates starch granules in the chloroplasts of its leaves and a glycogen-like polyglucan in myrmecophytic food bodies, so-called Müllerian bodies. The synthesis of two structurally different glucan polymers in separate tissues of the same plant offers a unique possibility to study the relative contributions of starch synthetic enzymes to insoluble (i.e.: starch) or soluble (i.e.: glycogen) polyglucan synthesis. Comparing activities of enzymes involved in starch synthesis between leaves and Müllerian bodies using activity gels, I show that Müllerian bodies have increased branching enzyme activity. As the scope of an activity gel-based approach is limited, I investigated the expression levels of genes coding for starch synthetic enzymes using next generation sequencing technologies. I partially sequenced the C. peltata transcriptome and quantitatively compared gene expression levels between leaves and Müllerian bodies. I conclude that the accumulation of large amounts of a soluble polyglucan in Müllerian bodies is due to a selective over- expression of a gene coding for a synthase preferentially synthesizing short glucan chains in combination with the down-regulation of genes coding for synthases synthesizing long chains as well as an increase in glucan chain branching activity. My use of next generation sequencing shows that biological questions may be addressed in non-model species. In addition, I provide the sequence information for further investigations of the Müllerian body transcriptome to understand the metabolism of this myrmecophytic food body. Zusammenfassung

Stärke ist die in Pflanzen am weitesten verbreitete Kohlenstoff-Speicherform. Stärke ist auch ein essentieller Bestandteil unserer täglichen Mahlzeiten sowie wichtiger Rohstoff für verschiedenste industrielle Anwendungen. Sowohl in Zusammensetzung als auch Funktion ist Stärke ähnlich zu Glykogen, der häufigsten Kohlenstoff-Speicherform in Pilzen, Bakterien und Tieren. Der wichtigste Unterschied zwischen Stärke und Glykogen liegt darin, dass Stärke in unlöslichen, semi-kristallinen Körnern vorliegt, während Glykogen in Form von kleinen, löslichen Partikeln akkumuliert. Wenn Photosynthese stattfindet wird in Pflanzen Stärke synthetisiert und wieder abgebaut, wenn Photosynthese nicht möglich ist. Stärke besteht fast ausschliesslich aus Glukose-Einheiten, die durch α(1,4)-Verbindungen zu linearen Ketten verknüpft sind. Die linearen Ketten wiederum sind durch α(1,6)-Verbindungen zu einem verzweigten Glukose-Polymer zusammengefügt. Die α(1,6)- Verzweigungen in Stärke sind nicht homogen verteilt. Regionen mit einer hohen Dichte an Verzweigungen alternieren mit schwach verzweigten Regionen. Aktuelle Modelle der Stärke-Struktur zeigen, dass benachbarte Glukose-Ketten Doppel-Helices bilden können, sofern sie lange genug sind und sich in schwach verzweigten Regionen befinden. Dies wird als Bedingung für die Bildung von semi-kristallinen Stärkekörnern angesehen. Für die Synthese von Stärke sind drei Enzymaktivitäten nötig: 1) Enzyme, welche Glukose-Ketten verlängern, 2) Enzyme, welche Verzweigungen einfügen und 3) Enzyme, welche überschüssige Verzweigungen wieder entfernen. Mehrere Isoformen all dieser Enzyme wurden in Pflanzen gefunden. Deren relativer Beitrag zur Synthese von Polyglukanen, welche semi-kristalline Strukturen bilden können, ist jedoch noch unklar. Im Jahr 1971 wurde ein subtropischer Baum, Cecropia peltata , beschrieben, welcher in myrmekophytischen Fruchtkörperchen, sogenannten Müllerschen Körperchen, ein glykogen-ähnliches Polyglukan akkumuliert (Rickson, 1971). In den Chloroplasten der Blätter jedoch finden sich Stärkekörner. Die Synthese von zwei strukturell unterschiedlichen Polyglukanen in verschiedenen Geweben derselben Pflanze bietet eine einzigartige Gelegenheit, die relativen Beteiligungen der stärke-synthetisierenden Enzyme, respektive deren Isoformen, zu studieren. In meinen Analysen zeige ich, dass in Müllerschen Körperchen grosse Mengen löslicher Polyglukane akkumulieren und dass diese Polyglukane einen hohen Anteil kurzer Glukose-Ketten haben. Ich vermute, dass die grosse Anzahl kurzer Ketten verhindert, dass genügend Doppel-Helices zwischen Glukose-Ketten gebildet werden und deshalb das Polyglukan in Müllerschen Körperchen glykogen-ähnlich und löslich ist. Ich zeige, dass die Aktivität von Enzymen welche Verzweigungen zwischen Glukose-Ketten bilden in Müllerschen Körperchen, im Vergleich zu Blättern, erhöht ist. Eine erhöhte Menge an α(1,6)- Verzweigungen kann die Löslichkeit eines Polyglukans erhöhen. Dank neuen Sequenziertechnologien konnte ich die Transkriptome von Müllerschen Körperchen und Blättern von C. peltata vergleichen. Im Detail habe ich die Expression von verschiedenen Isoformen von Genen untersucht, welche für stärke-synthetisierende Enzyme kodieren. Dabei konnte ich zeigen, dass von allen Enzymen, welche Glukose-Ketten verlängern, die Expression jener Isoform in Müllerschen Körperchen stark erhöht ist, welche kurze Ketten synthetisiert. Im Gegensatz dazu ist die Expression von Isoformen welche längere Ketten synthetisieren in Müllerschen Körperchen tiefer als in Blättern. Aus meinen Untersuchungen schliesse ich, dass die Akkumulation von glykogen-ähnlichen Polyglukanen in Müllerschen Körperchen auf der Überexpression eines kurze Glukose-Ketten synthetisierenden Enzyms, sowie der erhöhten Aktivität von α(1,6)-Verzweigungen bildenden Enzymen basiert. Im Weiteren zeige ich, dass mit neuen Sequenziertechnologien biologische Fragestellungen auch in Nicht-Modell-Organismen untersucht werden können und stelle mit meiner Transkriptom-Analyse Daten zur weiteren Erforschung der Müllerschen Körperchen zur Verfügung. In einem zweiten Projekt habe ich den Einfluss von Protein-Protein Interaktionen auf die Regulation des Stärkeabbaus in Arabidopsis thaliana untersucht. Die wichtigsten Enzyme, die im Stärkeabbau in Blättern involviert sind, wurden in den letzen Jahren beschrieben. Darunter sind β-Amylasen: Glukan- Hydrolasen, welche Maltose aus Stärkekörnern freisetzen. Weiter auch Glukan Kinasen und Phospho- Glukan Phosphatasen, welche Stärke transient phosphorylieren und dephosphorylieren. Obwohl viele der Enzyme bekannt sind, weiss man wenig über deren Interaktionen und Regulationen. In meinen Untersuchungen zu Protein-Protein Interaktionen zwischen stärke-abbauenden Proteinen habe ich eine bis anhin unbekannte Interaktion zwischen einer β-Amylase (BAM1) und einer putativen, an Stärkekörner bindende Phospho-Glukan Phosphatase (LSF1) identifiziert. Ich habe diese Interaktion mittels Co-Immunopräzipitation und Tandem-Affinitäts-Aufreinigung von BAM1 und LSF1 bestätigt. Dabei hat sich gezeigt, dass LSF1 zusätzlich zu BAM1 noch mit einer anderen β-Amylase (BAM3) sowie einer chloroplast-lokalisierten Malat-Dehydrogenase interagiert. Die Interaktion von Proteinen kann einen effizienten und kontrollierten Abbau von Stärke garantieren. Ich schlage vor, dass die Funktion von LSF1 darin liegen könnte, BAM1 und BAM3 an den Stärkekörnern zu lokalisieren und dadurch den Stärkeabbau effizienter zu machen. Alternativ dazu wäre es möglich, dass LSF1 die Redox-Balance des Chloroplasten, über die Funktion der interagierenden Malat-Dehydrogenase, beeinflusst und damit indirekt auch den Abbau von Stärke. Ich zeige in meiner Arbeit, dass Protein- Protein Interaktionen zwischen stärke-abbauenden Proteinen existieren und bis anhin unbekannte regulatorische Mechanismen im Stärkeabbau übernehmen könnten.

Abbreviations 3-PGA 3-phosphoglycerate ACN acetonitrile APS1 small subunit of AGPase in Arabidopsis APL large subunit of AGPase in Arabidopsis ADP adenosin-5-diphosphat ADP-Glc adenosine-5-diphosphoglucose AGPase ADP-Glc pyrophosphorylase AmBic ammonium bicarbonate AMP adenosin-5-monophosphate AMY α-amylase ATP adenosine-5-triphosphate AU arbitrary units BAM β-amylase CLD chain length distribution DBE debranching enzyme DE differential expression DHAP dihydroxy acetone phosphate DP degree of polymerization DPE1 disproportionating enzyme 1 DPE2 disproportionating enzyme 2, cytosolic DTT dithiothreitol EDTA ethylenediaminetetraacetic acid EOD end of day (end of the photoperiod) EON end of night (end of the dark phase) EtBr ethidium bromide FA formic acid FBA fructose-1,6 bisphosphate aldolase FBP fructose-1,6 bisphosphate FBPase fructose-1,6 bisphosphatase Fru fructose Fru-6P fructose-6-phosphate GAPDH glyceraldehyde 3-phosphate dehydrogenase GBSS granule bound starch synthase GFC gel filtration chromatography Glc glucose Glc-1P glucose-1-phosphate Glc-6P glucose-6-phosphate GO gene onthology GPT glucose-6-phosphate/phosphate translocator GWD α-glucan water, dikinase HEPES 2-(4-(2-hydroxyethyl)- 1-piperazinyl) ethanesulfonic acid HMW high molecular weight HPAEC - PAD high-performance anion exchange chromatography with pulsed amperometric detection

ISA isoamylase LDA limit-dextrinase LMW low molecular weight LSF1 like-sex4-one MDH malate dehydrogenase MES 2-(N-morpholino) ethanesulfonic acid MEX1 maltose excess 1, maltose exporter at the chloroplast membrane MOPS 3-(N-morpholino)propanesulfonic acid MS/MS tandem mass spectrometry NADH nicotinamide adenine dinucleotide NADPH nicotinamide adenine dinucleotide phosphate NBT p-nitro-blue tetrazolium chloride NMW native molecular weight OAA oxaloacetate PAGE polyacrylamide gel electrophoresis PCR polymerase chain reaction PEP-C phosphoenolpyruvate carboxylase PGI phosphoglucose GLT plastidic glucose transporter PGM in

Pi inorganic phosphate PMS phenazine methosulfate

PP i pyrophosphate PWD phospho-glucan, water dikinase RuBisCO ribulose-1,5-bisphosphate carboxylase oxygenase SBE starch branching enzyme SDS sodium dodecylsulfate SE standard error sex starch-excess SEX4 starch excess 4, phospho-glucan phosphatase SSS soluble starch synthase Suc sucrose SuSy sucrose synthase TAP tandem affinity purification TEM transmission electron microscopy TFA trifluoroacetic acid TPT triose-phosphate/phosphate translocator T6P trehalose-6-phosphate Tris tris(hydroxymethyl)aminomethane UDP-Glc uridine-5-diphosphoglucose WSP water soluble polysaccharide β-CLD chain length distribution of β-limit dextrins

1. Introduction 7 ______

1. Introduction

1.1 Starch - A Vital Plant Product for Mankind

The annual production of plant biomass has been estimated to be around 170 x 10 9 tons, 75% of which consist of carbohydrates, 20% are made up of lignin and the remaining 5% comprise other natural products such as oils, fats and proteins. Approximately 3.5% of the total biomass produced is used by mankind. The biomass which is used splits up into approximately one third wood, one third cereals and into one third of mixed material like oil seeds, fruits, vegetables but also sugar beet and sugar cane (Roper, 2002). Apart from wood, which is used for building and heating purposes, the majority of the biomass is used as food (62%) and only 5% are used in the chemical industry. In the years 1999 to 2001, in the developed world, cereals provided 30.9% of the daily caloric input while 4.1% of the daily calories came from roots and tubers (www.fao.org). As the main compound used from cereals, roots and tubers for food applications is starch, it is not surprising that the total share of starch of the daily input of calories in developed countries sums up to about 35%. Thus starch is a plant product, vital for mankind. Of a total of 80 x 10 6 tons of agricultural commodities produced per year in Europe, 19.1 x 10 6 tons of raw materials are used to produce approximately 7.7 x 10 6 tons of starch. The major sources for the starch are not surprisingly potatoes (8.8 x 10 6 tons), maize (5.9 x 10 6 tons) and wheat (4.4 x 10 6 tons, Roper, 2002).

55.8% of the total production of starch are used in food related applications and the remaining 44.2% of the starch (3.4 x 10 6 tons) are being used in industrial applications. In the industrial sector, the use of starch in paper and corrugating applications is the major drain for starch.

Starch is to a great extent used in the production of binders, which help to compact powders and dusts to pellets, briquettes and other formed pieces. Starch based products are used in the detergent industry, where as hydrophilic end group in surfactants, as starting material for polycarboxylates and as the basis for bleach activators they replace petrochemical-based components. In powder detergents, starch based products make up to 50-60% of the components. To solve environmental and waste problems, more and more starch is being used in the production of biodegradable thermoplastic materials. In the past, non-degradable polymers like polyethylene and polypropylene have been mixed with 6-20% starch, to give a mixed, partially degradable polymer. Today, composite materials with up to 50% starch are being produced, which have good to very good biodegradability. In the pharmaceutical industry, starch and starch derivatives are widely used as excipients for tablets, as binders for coatings and even as matrices for the controlled release of active ingredients. Biodegradable starch microspheres given by intra-arterial injection can provide a way to deliver cytotoxic drugs to metastases and simultaneously reduce the systemic toxicity (Ball, 1991). In the cosmetic industry hydrophobic starches are a widely used raw material for the production of 1. Introduction 8 ______emulsions. A more complete overview of industrial applications of starch and starch based products can be found in the reviews by Jobling (2004) and Roper (2002).

With the current attempts to reduce CO 2 emissions in the context of greenhouse-gas driven global warming and with the increased awareness of limited non-renewable resources, the production of liquid bio-fuels has gained attraction. Starch and sugars are the major sources for the production of bio-ethanol (Smith, 2008) and an increasing amount of starch is used each year. While sugarcane already represents the major source for bio-fuel production in Brazil, more than 30% of the US maize crop is predicted to be used for bio-ethanol production. A mandate established in the EU requires that renewable sources account for 10% of the transportation sector’s energy use by 2020. Even though only 60% of the renewable resources are expected to be achieved from annual crop-feedstock and even though increased oilseed production will alleviate the demand of starch containing crops, a world-wide increase of starch use for bio-ethanol production can be expected (http://www.ers.usda.gov/Briefing/Baseline/trade.htm).

The use of starch in many different applications is supported by the wide range of physical and chemical properties of the different plant starches (e.g.: gelatinization temperature, retrogradation tendency or swelling capacity). For instance, the naturally high levels of phosphate in potato starch are at least in part responsible for its high swelling power and the starch paste properties (Blennow et al., 2002; Jobling, 2004). In some cases the natural botanical variety is sufficient to meet the different needs but in most cases, the native starches are modified to meet the desired physico-chemical properties. Even though some of the modifications are easily achieved through chemical and enzyme treatment and/or physical treatment (Jobling, 2004), some modifications require high energetic input and potentially harmful chemicals. Altering structure and composition of starch produced in plants, to provide an improved raw material, offers a great possibility to save costly and environmentally unfriendly conversion steps in industry. To be able to make changes in the starches which are produced in the plants, it seems obvious that the synthesis of the granule, its semi-crystalline structure as well as its degradation process in the plant organ should be understood in detail.

1.2 Starch Is the Major Carbon Storage Compound in Plants

Having talked in detail about the relevance of starch to mankind, it should be emphasized that plants are not synthesizing starch for man’s sake. Most living organisms store carbon in order to deal with fluctuating availability of nutrients and energy. Fluctuations may be long-term (seasonal changes throughout a year) or short-term (changes of irradiance for plants in the diurnal cycle). Different glucose polymers are used for storage and structural purposes. In the structural component cellulose for instance, glucose units are connected to each other by β(1,4) linkages. In starch and glycogen, the most widespread forms of storage in living cells, glucose units are α(1,4) linked 1. Introduction 9 ______linear and α(1,6) branched glucose polymers (Ball and Morell, 2003b; Zeeman et al.). Glycogen is found in the majority of animal, bacterial, fungal and archaebacterial species as well as some thermophilic red algae and interestingly also in some higher green plants (see below). Starch on the other hand is found in either photosynthetic eukaryotes or nonphotosynthetic derivatives (apicomplexa parasites, nonphotosynthetic dinoflagellates and a cyanobacterium (Coppin et al., 2005; Deschamps et al., 2008).

In higher plants, the accumulation of starch serves two functions, both related to carbon storage on different time-scales. Storage starch accumulated in roots, tubers, stems and seeds, serves as a long- term carbon store to support periods of heterotrophic growth or increased carbon demand. This starch, mankind is most interested in, as it accumulates in our staple crops (e.g.: cereal seeds, potato tubers or the storage roots of cassava.). In contrast to the storage starch, transient starch accumulates in the photosynthetic leaves and is crucial for normal plant growth and metabolism. During photosynthesis, carbon from atmospheric CO 2 is fixed into hexose sugars and used to synthesize either sucrose or starch. Sucrose is exported to support metabolism in non-photosynthetic parts of the plant, while starch accumulates as semi-crystalline starch grains inside chloroplasts. Transient starch is degraded at night, when photosynthesis is not possible and provides the substrates for leaf respiration and continued sucrose synthesis. The importance of transitory starch is reflected by the retarded growth phenotypes of some mutants unable to synthesize starch (Gibon et al., 2004a; Smith and Stitt, 2007). Sulpice et al. (2009) have shown that over a range of 94 Arabidopsis accessions, starch levels are negatively correlated with biomass. Though this may appear counterintuitive, it reflects the fact that synthesis of starch is an investment, locking carbon which could be directly used for growth during the day. Interestingly, the negative correlation is lost in very short day periods, pointing out the importance of starch as carbon storage and a large flexibility in carbon allocation. In Arabidopsis , the amount of carbon diverted to starch synthesis is up to 50% during the day (Zeeman and Ap Rees, 1999; Zeeman et al., 2006). The amount of carbon partitioned into starch and sucrose is not only dependent on day length, with an increasing amount of carbon stored in starch with decreasing day length, but also differing between plant species. Pisum sativum and Spinacia oleracea for instance accumulate significant amounts of sucrose next to starch and in these species starch synthesis might more work like an overflow mechanism for photo-assimilates when the need of sucrose is satisfied.

1.3 The Structure and Composition of Starch Granules

1.3.1 Starch Granule Morphology

Granule size and shape are related to the biological source from which the starch is isolated. Granule sizes can be in a range from smaller than 1 µm up to more than 100 µm (Lindeboom et al., 2004). The largest difference can be seen between granules from leaves, where the starch is synthesized and 1. Introduction 10 ______degraded in one day/ night cycle and the granules from storage organs, where the starch is synthesized over a longer period of time. Leaf starch granules are typically small, <2 µm and discoid (Zeeman et al., 2002). Starch granules from most storage organs are larger, mostly 15-100 µm, and roughly spherical or oval in shape (Jane et al., 1994). In some species like barley, rye and wheat, a bimodal population of granules exist, with small granules in the range of 1-5 µm and large granules with a size range of 12-36µm (Lindeboom et al., 2004). Among starch granules, the morphology of the maize granules is special, as they have a clear polygonal structure, a phenotype shared with rice granules, where the polygonal structure is even more pronounced. Due to their clear polygonal appearance, rice granules have been named composite or fused starch granules, as the single granules tend to cluster together in their polygonal forms (Jane et al., 1994). In Arabidopsis, mutants have been identified with granule morphologies largely different from the wild type (e.g.: Zeeman et al., 2002). In potato, mutants with altered numbers of starch granules have been described (Bustos et al., 2004). Even though some information is available on factors which control granule morphology and number either directly or indirectly, a comprehensive picture of how granule morphology and number are determined is still missing.

1.3.2 Starch Composition

Starch is primarily composed of two different polymers of glucose. The mass of naturally occurring starches is made up to 70-85% by amylopectin, a branched glucose polymer, in which the single glucose units are connected to each other by α(1,4) linkages forming chains which are then connected to each other by α(1,6) linkages (branch points, Fig1-1A). This branched molecule is comparably large with an estimated molecular weight of 10 7-10 9 Daltons (Buleon et al., 1998; Yoo and Jane, 2002). The remaining 15-30% are made up by a second, smaller glucose polymer, called amylose, with an estimated molecular weight of 10 5-10 6 Daltons (Perez and Bertoft, 2010; Zeeman et al., 2010). In contrast to amylopectin, which has an average branching frequency of 5%, amylose is smaller and almost linear (Buleon et al., 1998). A single amylose molecule contains 1000 to 10000 glucose units and the branching frequency varies from species to species but is very low, ranging from about one in 800 glucose residues in maize to one in 4500 in potato (Davis et al., 2003).The exact ratios of amylopectin and amylose contents of starch granules can vary between starch from storage tissue and starch from leaf tissue and between species (Zeeman et al., 2002). For the overall starch granule crystallinity and morphology, however, amylose is not required (Ball et al., 1996). Even a total lack of amylose does not impair the capacity of the plant to produce starch granules (Noda et al., 2002). Mutants which contain more than 95% amylopectin have been described in maize, rice, barley, wheat and potato (Davis et al., 2003). High amylose starches with an amylose content of between 50% and 85% have been reported in maize, rice and pea (Bhattacharyya et al., 1990). 1. Introduction 11 ______

Minor components of starch include proteins, lipids and phosphate. In cereals, starch protein levels range from 0.25-0.5%, while in potato and cassava, protein levels are generally <0.1% (Davis et al., 2003). The most abundant proteins present in starch granules are enzymes of the starch biosynthetic machinery. These enzymes either have functions inside the starch granule (granule bound starch synthase, GBSS, see Section 1.5.1) or may be trapped inside the growing granule during synthesis (Boren et al., 2004; Grimaud et al., 2008). Already in the early 20th century, the presence of small amounts of monoesterified phosphate groups were detected in potato starch (Fernbach, 1904; Blennow et al., 2002). Almost all starches extracted from different plant species contain some phosphate esters (Blennow et al., 2000a). Nevertheless the level of phosphorylation varies considerably with the botanical origin of the starch. Phosphate bound to starch is present in minute amounts or even below the level of detection in cereal starches, whereas approximately 1 in 200 glucose units can be phosphorylated in potato starch. In Arabidopsis , the frequency of phosphorylation is 1 in 2000 residues (Yu et al., 2001). In tuberous starch it has been shown, that the phosphate is covalently bound to amylopectin and not to amylose. In potato tubers, starch bound phosphate can account for 30 to 50% of the total phosphate. Between 60% to 70% of the phosphate are bound to the C6 position of the glucose, the remainder is bound to the C3 position (Blennow et al., 2000b; Haebel et al., 2008). The phosphorylation level of starch has a great impact on its physical properties. Phosphate groups in starch increase the hydration capacity of starch pastes and thus influences the gel-forming capacity (Blennow et al., 2002). Even though the steady-state levels of phosphate are comparably low, the phosphorylation of glucan chains on the granule surface is of crucial importance for the degradation of starch and the rate of phosphorylation increases during degradation (Ritte et al., 2004). Not only due to its industrial relevance but also due to its recognition as a crucial process controlling starch degradation, reversible phosphorylation of starch has been investigated intensely in the past years (Blennow et al., 2002; Zeeman et al., 2010). The enzymes and the specific impact of the starch phosphorylation dynamics on starch degradation will be discussed in Section 1.7.4

1.3.3 Starch Granule Architecture

Despite their apparent chemical simplicity, starch granules display a complex higher order organization on different levels. Viewed under polarized light, starch granules are birefringent, which indicates a high degree of radial molecular orientation of the molecules within the granules. The complex architecture of granules is further supported by concentric zones of alternating crystalline material, resistant to acid hydrolysis and amorphous material, susceptible to acid hydrolysis, which have been revealed in electron microscopic investigations of storage starch granules. These alternating zones are reminiscent of the growth rings in trees and they are often referred to in this way (Zeeman et al., 2002). The high architectural complexity and crystallinity of granules is often explained by the 1. Introduction 12 ______distribution and frequency of chain lengths as well as of branch points in amylopectin. The branch points in amylopectin are thought to be non-homogenously distributed, where regions with high branch point frequency alternate with regions of low branch point frequency. In regions with comparably lower branching frequency, neighbouring linear chains are able to form double helices (Fig1-1A). These double helices are thought to cluster together forming so called crystalline lamellae. These crystalline lamellae are alternating with zones of lower crystallinity, which are thought to have a higher amount of branching which interferes with the formation of double helices between adjacent chains. The zones of high branching frequency are often called amorphous lamellae (Fig1-1A). The repetition of one crystalline lamella and one amorphous lamella every 9-10nm is the lowest order structure of the starch granule and is thought to allow the formation of higher order structures. This 9- 10nm repeat is strikingly constant, well conserved in the plant kingdom and can be found in leaf starches as well as in storage starches (Jenkins et al., 1993; Zeeman et al., 2002).

The inner glucan chains in amylopectin have been characterized as A, B or C chains. A chains are external chains that carry no other chains; they are linked via their reducing end to the C6 of a glucose residue of a B chain. B chains carry one or more chains via such a branch point. They can span one crystalline lamella (having a typical degree of polymerization (DP) of 12 to 15 glucose units), two lamellae (DP35-DP40) or even three lamellae (DP70-DP80). One C chain exists per amylopectin molecule; it is the only B chain still carrying the reducing terminal glucose residue (Perez and Bertoft, 2010).

Atomic force microscopy and scanning electron microscopy experiments have suggested another level of organization between the large growth rings and the amylopectin lamellae. This level of granular organization has been termed the ‘blocklet structure’ (Gallant et al., 1997; Perez and Bertoft, 2010). Crystalline and amorphous lamellae have been proposed to organize into blocklets (Gallant et al., 1997), where repeating units of crystalline and amorphous lamellae are grouping into discrete elongated structures (Fig1-1A). Blocklets are arranged into concentric layers (crystalline shells) where two to three layers of blocklets make up a shell. The crystalline shells are interspaced by amorphous material. The alternation between these crystalline shells and the soft amorphous regions make up the so called growth rings described in starch granules (Fig1-1B).

Starch granule architecture at least as far as the concept of the alternating crystalline and amorphous lamellae is concerned, seems to be the same in granules from green, photosynthesizing tissue and from non-green, storage tissue. This makes it possible to use the transitory starch as a model for the synthesis and the degradation of the starch granule in general (Zeeman et al., 2002). 1. Introduction 13 ______

A Blocklet

Amorphous Lamellae 9nm Repeat Crystalline Lamellae

B

Crystalline Shell

Amorphous Shell

Blocklet

Figure 1-1. A Model of the Starch Granule Architecture (A) Starch mainly consists of amylopectin, a branched glucose polymer. Single glucose units are linked by α(1,4) linkages to form linear chains. Single chains are connected to each other by branch points, α(1,6) linkages. Amylopectin can be visualized as a tree-like molecule with alternating regions of high and low branch-point density. In regions of low branch-point density, neighbouring chains form double helices with each other. Helices cluster together in crystalline lamellae, alternating with amorphous lamellae, where amylose is thought to be localized. The space interval between one crystalline lamella and the next one is highly conserved in the plant kingdom and usually is 9-10 nm. Amorphous and crystalline lamellae group together to form blocklets, a higher-order structure of starch. (B) Starch granules from storage organs display ‘growth-rings’, zones of relatively high and low crystallinity. Partial acid hydrolysis of starch granules and subsequent scanning electron microscopy reveals the zones of high and low susceptibility to acid hydrolysis (left). Blocklets arranged into crystalline shells, interspaced by amorphous shells are the building structure of the starch granule ‘growth-rings’.Starch Synthesis in Auto- and Heterotrophic Tissues

1.4 Starch Synthesis in Autotrophic and Heterotrophic Tissue

In vascular plants, the activated sugar ADP-Glucose (ADP-Glc) is the committed precursor for the synthesis of starch. In contrast, in red algae and in glaucophytes, UDP-glucose serves the same function. Only starch synthesis of green plants will be reviewed here. The synthesis of ADP-Glc is discussed in the following (Fig1-2A). ATP and NADPH are produced in the photosynthetic light reactions and used in the Calvin-cycle for the fixation of CO 2 derived carbon. The triosephosphate 3- phosphoglycerate (3-PGA) is one of the intermediates of the Calvin-cycle. By activity of the triosephosphate isomerase, the 3-PGA is transformed into dihydroxyacetone-3-phosphate (DHAP), 1. Introduction 14 ______which then undergoes an aldol condensation with another molecule of 3-PGA to yield fructose-1,6- bisphosphate (Fru-1,6BP). Fru-1,6BP is dephosphorylated by Fru-1,6BP phosphatase yielding fructose-6-phosphate (Fru-6P) and inorganic phosphate. The Fru-6P feeds the hexose phosphate pool, where Fru-6P, glucose-6-phosphate (Glc-6P) and glucose-1-phosphate (Glc-1P) are interconverted. More precisely, phosphoglucose-isomerase (PGI) interconverts Fru-6P into Glc-6P and phosphoglucomutase (PGM) interconverts the Glc-6P into Glc-1P. It is the Glc-1P which is consecutively used by the enzyme ADP-Glc pyrophosphorylase (AGPase), the first committed step to starch synthesis. This enzyme uses Glc-1P and ATP and liberates ADP-Glc and pyrophosphate. The latter is readily broken down to inorganic phosphate by the plastidial pyrophosphatase, driving the AGPase catalyzed reaction into the direction of ADP-Glc synthesis (Weiner et al., 1987). Being the first step of starch synthesis, it is not surprising that AGPase is one of the most important regulatory points of the pathway. The control of its activity contributes to the overall regulation of the partitioning of photosynthates between the synthesis of sucrose in the cytosol and the synthesis of starch in the chloroplast. AGPase is subject to many regulatory effects.

The AGPase of higher plants is a tetrameric enzyme, consisting of two small and two large subunits. The formation of an intramolecular disulfide bridge between two cysteines of the small subunits is an important regulatory mechanism for AGPase. With the small subunits in their reduced state, AGPase has high affinity for its substrates (Fu et al., 1998; Ballicora et al., 2000; Tiessen et al., 2002; Geigenberger et al., 2005), high sensitivity to the allosteric activator 3-PGA and low sensitivity to the inhibitory P i. Consequently, the high 3-PGA/P i ratio in the chloroplast stroma, when the supply of photosynthates exceeds demand, results in the activation of AGPase and increased starch synthesis. Thus redox regulation leads to allosteric activation of AGPase in the light, when starch is synthesized, an effect which has been proposed to channel carbon into starch (Stitt and Quick, 1989; Tiessen et al., 2002) Sugars have been shown to influence the redox activation of AGPase. More precisely, high sucrose levels have been suggested to increase starch synthesis by an SnRK1 (sucrose-non- fermenting-1-related protein kinase) dependent signaling pathway and glucose levels have been suggested to influence AGPase redox state in a hexokinase dependent way (Tiessen et al., 2003). Trehalose-6-phosphate (T6P) has been identified as potential signalling compound involved in the SnRK1 dependent regulation of AGPase activity (Kolbe et al., 2005). It has been suggested that carbohydrate depletion and thus starvation at night might lead to an increased synthesis of T6P, resulting in an increased activation state of AGPase in the next light period and by this an increased partitioning of carbon into starch. This regulatory network, which is not completely understood, reflects the function of starch not only as a simple overflow of photosynthates but as an important buffer and carbon store with adjustable capacity.

In a ‘classical’ view of the starch synthesis pathway, ADP-Glc synthesis occurs in the chloroplast. However, in most non-photosynthetic tissues (e.g.: storage organs) starch is synthesized in non-green 1. Introduction 15 ______plastids and ADP-Glc is synthesized from sucrose imported via the phloem from the photosynthetic parts of the plant (Fig1-2B). In the cytosol of amyloplast containing cells, sucrose is metabolized by sucrose synthase to give UDP-Glucose and fructose, or by invertase to give glucose and fructose. Both pathways feed into the hexose-phosphate pool, from which Glc-6P is imported into the plastids and thus used for ADP-Glc synthesis (Hill and Smith, 1991, 1995). In the cereal endosperm, the situation differs. A significant amount of ADP-Glc is synthesized by a cytosolic form of AGPase and then imported into plastids to fuel starch synthesis (Fig1-2B). While the importance of ADP-Glc import into plastids is generally accepted, in the cereal endosperm, its import into the chloroplasts of green leaves is a question of debate. The fact that Arabidopsis mutant lines lacking plastidic AGPase, PGI or PGM are virtually starchless, suggests that ADP-Glc synthesis in the chloroplast is the major pathway to provide starch precursors. Tiny levels of starch have, however, been detected in pgm mutants of Arabidopsis, lacking plastidic PGM activity. It cannot be excluded that ADP-Glc from the cytosol leaks into plastids (Streb et al., 2008). Alternatively, the import of Glc-1P has recently been shown into Arabidopsis chloroplasts as well as potato amyloplasts (Fettke et al., 2010b; Fettke et al., 2010a). Glc-1P imported into the plastid can be used for starch synthesis without the activity of PGM.

1. Introduction 16 ______

CO 2 Light A Chloroplast Cytosol

Calvin- Cycle 6 Triose-Phosphate Starch ADP-Glc 4 5 Glc-6P Glc-1P PP i ATP

Triose-Phosphate PP 1 UTP i Sucrose 2 3 4 7 Fru-6P Glc-6P Glc-1P UDP-Glc

B Amyloplast Cytosol

6 8 UDP-Glc Fructose Starch 10 9 ATP ADP-Glc 6 PP Glc-1P Fru-6P i 5 PP i ATP 5 4 4 3 ADP-Glc Glc-1P Glc-6P ADP-Glc Glc-6P

Figure 1-2. Sugar Conversions for Starch Synthesis in Autotrophic and Heterotrophic Tissue during the Day (A) Starch synthesis in autotrophic tissue: Atmospheric CO 2 is fixed in the Calvin-Cycle consuming ATP and reducing equivalents provided by the light reactions of photosynthesis. Triose-phosphates are exported from the chloroplast and converted to sucrose by conversion to fructose-6-phosphate (Fru-6P) by fructose-1,6-bisphosphate aldolase (1) and fructose-1,6-bisphosphate phosphatase (2). Fru-6P is interconverted to glucose-6-phosphate (Glc-6P) by cytosolic phosphoglucose-isomerase (PGI, 3). Glc-6P is interconverted to glucose-1-phosphate (Glc-1P) by cytosolic phosphoglucomutase (PGM, 4). Finally, UDP-Glc is synthesized from Glc-1P by UDP-Glucose pyrophosphorylase ( 7) and sucrose synthesized from UDP-Glc in autotrophic source tissue can be exported to heterotrophic sink tissue. Inside chloroplasts, Glc-6P from the Calvin-Cycle is interconverted to Glc-1P by plastid PGM (4) and ADP-Glucose pyrophosphorylase (5) synthesizes ADP-glucose (ADP-Glc) from Glc-1P and ATP. In a series of steps (6, see Section 1.5 for details) starch is synthesized from ADP-Glc. (B) Starch synthesis in heterotrophic tissue: Sucrose is imported into sink tissue and converted by cytosolic sucrose synthases ( 8) into UDP-Glc and fructose. Fructose is phosphorylated by hexokinase (9) to give Fru-6P. Cytosolic PGI and PGM convert Fru-6P to Glc-1P. UDP-Glc is converted to Glc- 1P by UDP-Glc pyrophosphorylase ( 10 ) and metabolised to Glc-6P. Glc-6P is imported into amyloplasts by the glucose-phosphate/phosphate translocator; Glc-6P is converted to ADP-Glc in the amyloplasts by (4) and (5). In cereal amyloplasts, ADP-Glc can be synthesized from Glc-1P in the cytosol and imported by the ADP-Glc transporter (pathway in grey).

1. Introduction 17 ______

1.5 The Synthesis of Amylopectin and Amylose

Starting from ADP-Glc, the semi-crystalline starch granule is synthesized. Three enzyme activities have been identified to contribute to starch synthesis. The concerted activities of starch synthases (SS), starch branching enzymes (SBE) and debranching enzymes (DBE) are all required for the synthesis of starch granules. The large number of isoforms in each enzyme class (e.g. five starch synthase isozymes) makes the investigation of starch synthetic enzymes a challenging task.

1.5.1 Starch Synthases

Starch synthases (ADP-glucose: (1 →4)-α-d-glucan 4-α-d-glucosyltransferase; EC: 2.4.1.21) transfer the glucosyl moiety of ADP-Glc to the non-reducing end of an existing glucan chain, thereby creating a new α(1,4) linkage and elongating the chain by one glucose unit. Based on phylogenetic comparisons, five major classes of SSs have been identified (Ball and Morell, 2003b). Four are termed soluble starch synthases (SSS) SSSI, SSSII, SSSIII and SSSIV and one is termed granule bound SS (GBSS). Different species have different numbers of genes per class of SS. In monocots for instance, two forms of SSSII have been identified which supposedly have evolved through gene duplication (Harn et al., 1998).

GBSSs bind tightly to starch granules and are responsible for amylose synthesis, as shown by the lack of amylose in the waxy mutants of maize, which are defective in GBSS (Shure et al., 1983). Similar amylose free mutants exist in a range of species and they have all been shown to be mutated in a gene coding for GBSS (Ball et al., 1998). The synthesis of the branched molecule, amylopectin, is more complex, involving four different isoforms of the soluble starch synthases, with the individual isoforms preferentially elongating different chain lengths. By the analysis of glucan chain lengths in the amylopectin of mutant plants, lacking one or several of the SSS, the chain length preferences of the different SSS have been disentangled (Tomlinson and Denyer, 2003; James et al., 2003). The SSSI isoform is thought to be involved in the synthesis of short glucan chains, with a DP of 10 and less glucosyl units. In Arabidopsis , SSSI is thought to be involved in the synthesis of small outer chains during amylopectin synthesis (Delvalle et al., 2005).SSII and SSIII preferentially act on and synthesize longer chains than SSI. Mutations of SSII in pea result in the synthesis of an amylopectin with more short chains (DP<10) and long chains (DP>25) but less intermediary chains (Craig et al., 1998). The amylopectin of this mutant has an altered crystallinity and organization of starch. In Arabidopsis , null mutations of the SSII gene lead to an amylopectin with decreased numbers of chains of DP12 to DP28 (Zhang et al., 2008). Elimination of SSIII activity in potato tuber has a large impact on the synthesis of amylopectin, resulting in decreased starch synthesis and an altered chain length profile of the amylopectin (Edwards et al., 1999). In Arabidopsis , however, lack of SSIII alone did not cause major changes in the amylopectin structure (Zhang et al., 2008). The sssII/sssIII double mutant, 1. Introduction 18 ______however, displayed a strong reduction in chains of DP12 to DP28, more pronounced than in the sssII single mutant. This finding suggests that in Arabidopsis , there is some redundancy between the activities of SSII and SSIII. The SSIV class of enzymes has been suggested to be involved in the priming of granule synthesis. While total starch levels and chain length distribution of amylopectin are similar to wild type, the number of granules is drastically decreased and granule size increased in the sssIV mutant of Arabidopsis . The sssIII/sssIV double mutant has been reported to be virtually starch free, thus suggesting that these two activities are essential for starch synthesis (Roldan et al., 2007; Szydlowski et al., 2009).

1.5.2 Starch Branching Enzymes

Branching of glucan chains in amylopectin synthesis is thought to proceed concurrently with chain elongation. Starch branching enzymes ( α-1,4-glucan: α-1,4-glucan-6-; EC: 2.4.1.18) catalyze the cleavage of the α(1,4) bond and transfer the released glucan chain with its reducing end to a C6 hydroxyl, thereby creating a new linkage, an α(1,6) branch point, on the same chain or on a chain close by.

Two to three genetically independent isoforms of SBEs have been found in higher plants and depending on their peptide sequence have been categorized in two classes: SBEI and SBEII. The fact that in some species the SBEII class is represented by more than one isoform complicates the investigation of specificities. In maize, SBEIIa and SBEIIb are found, in Arabidopsis , SBE 2 and SBE 3 both belong to the SBEII class. The SBEI and SBEII classes differ in the length of chains they transfer, with the former showing a bias towards the transfer of longer chains compared to SBEII, which seems to transfer shorter chains, although not shorter than DP6 (Takeda et al., 1993; Rydberg et al., 2001). SBEII class enzymes generally show higher affinity for amylopectin when expressed in vitro. The down regulation of SBEI in monocots and dicots seems to have a minimal effect on starch synthesis and composition (Blauth et al., 2002; Tetlow et al., 2004a). In contrast, Jobling et al. (1999) found that when SBEII in potatoes is anti-sensed, the average chain length of amylopectin was significantly increased. The simultaneous lack of SBE 2 and SBE 3 in Arabidopsis abolishes starch synthesis (Dumez et al., 2006).

Further light has been shed on the roles of the branching enzymes by heterologous expression of the enzymes in yeast, where it was shown that SBEI is unable to act in absence of SBEIIa or SBEIIb (Seo et al., 2002). All of the SBEs have been shown to be phosphorylated in wheat and the formation of complexes between SBEIIb, SBEI and starch (see Section 1.5.4) as well as the activation of SBEIIa are phosphorylation dependent (Tetlow et al., 2004b).

1. Introduction 19 ______

1.5.3 Debranching Enzymes in Starch Synthesis

The third enzyme activity required for efficient synthesis of starch granules is catalyzed by debranching enzymes (DBEs). DBEs ( α-1,6-glucanohydrolase) hydrolyze the α(1,6) bond between two glucose chains. In all higher plants which have been described so far, two different sub-classes of debranching enzymes exist, which differ in their ability to hydrolyze different substrates. The isoamylase-type DBE (ISA, EC: 3.2.1.68) readily cleaves the α(1,6) bonds in amylopectin and its counterpart glycogen. The second sub-class of DBE (EC:3.2.1.142) is able to hydrolyze the α(1,6) linkages in the yeast polymer pullulan, (maltotriosyl sub-units linked to each other by α(1,6) bonds) and therefore has often been termed pullulanase (Wu et al., 2002). Owing to this enzyme’s activity on limit-dextrins (dextrins digested with an excess of α- or β-amylase), the term limit-dextrinase (LDA) has been proposed and is used in this work.

Four genes coding DBE- like proteins are encoded by the Arabidopsis genome. Three encode ISA- type DBEs ( AtISA1 , AtISA2 and AtISA3 ) and one gene encodes an LDA-type of DBE ( AtLDA ). The distribution of three ISA-type of DBE to one LDA-type of DBE is conserved in all higher plants investigated so far. Phylogenetic clustering confirms the classification as sequence comparisons have shown that the ISA- and the LDA-type of DBEs have been conserved separately in evolution. Each plant enzyme is more similar to the bacterial enzyme from the same class than to the plant enzymes belonging to the other class. Thus suggesting that ISA- and LDA-type of DBEs have diverged before the establishment of the plant kingdom and that the specific function of each class of enzymes has independently been conserved in the evolution of plants (Beatty et al., 1999).

The role of DBEs in the synthesis of starch has already been discussed by Pan and Nelson (1984) in the context of sugary1 mutants in maize. In maize sugary1 mutants, one of the commercial sweet corn varieties, the principal polysaccharide storage product is not starch but a more highly branched water- soluble polysaccharide (WSP), with a high molecular weight. The accumulation of WSP has been shown to positively correlate with the numbers of sugary1 alleles and it has later been shown that the sugary1 locus encodes a structural gene for an ISA-type DBE (Pan and Nelson, 1984; James et al., 1995). Mutants lacking ISA-type DBEs have been described in maize, rice and barley (James et al., 1995; Kubo et al., 1999; Burton et al., 2002), as well as in Chlamydomonas (Dauvillee et al., 2001a) and Arabidopsis (Zeeman et al., 1998c).

In mutants lacking either the ISA1 or ISA2 DBE a certain level of WSP accumulated, implicating these proteins in starch synthesis. In contrast, the ISA3 DBE (and to some extent also the LDA DBE) has most frequently been associated with starch degradation rather than synthesis. In Arabidopsis, ISA1 and ISA2 form a heteromultimeric complex, which is roughly 500kDa in size (Delatte et al., 2005). It has been shown that in the endosperm of rice and maize, ISA1 exists both, as a heteromultimer with ISA2 and as a homomultimer (Utsumi and Nakamura, 2006; Kubo et al., 2010). 1. Introduction 20 ______

ISA2 has been suggested to be catalytically inactive and potentially modulate the activity, specificity and stability of ISA1 (Hussain et al., 2003). In the endosperm of maize mutants lacking ISA2, near- normal levels of starch accumulate, suggesting that ISA1 might be sufficient for starch synthesis (Kubo et al., 2010). The specific functions of ISA1 and ISA2 in the ISA1/ISA2 complex have not been revealed yet and remain to be investigated.

The accumulation of WSP caused by the lack of ISA1/ISA2 debranching activity has been explained by the presence of ‘wrongly positioned’ glucan chains, introduced by SBEs. Such excess glucan chains have been suggested to inhibit the self-organization of amylopectin into crystallization- competent starch granules (Zeeman et al., 2002; Zeeman et al., 2010). The removal of excess or ‘wrongly positioned’ chains by the ISA1/ISA2 enzyme is thought to facilitate the formation of semi- crystalline starch granules. The mechanism by which the ISA1/2 enzyme identifies the glucan chains to be removed from the nascent amylopectin molecule, however, is currently unclear.

The accumulation of WSP in isoamylase mutants is not only dependent on the DBEs. In all cases (except for Chlamydomonas ), there is some residual insoluble glucan polymer, reminiscent of starch, left in the plants lacking the ISA1/ISA2 activity (Dauvillee et al., 1999; Delatte et al., 2005). In the Arabidopsis mutants Atisa1 and Atisa2 WSP accumulate in the plastids of the palisade and the spongy mesophyll, whereas in other cell types such as epidermal cells, bundle sheath cells or companion cells, starch granules accumulate. This residual starch is in its composition of glucan chains (its chain length profile), significantly different from wild-type starch and remarkably similar to the WSP, despite being insoluble. The distribution of the branch points, however, is significantly different between the WSP and the residual starch; the branch points are located closer to each other in the soluble polymer (Delatte et al., 2005).

Although WSP accumulation is mostly due to the lack of the ISA-type of DBE, the severity of the phenotype seems to be modulated by the residual LDA activity in maize and rice kernels. A direct spatial link between WSP accumulation in the inner regions of rice kernels and the lack of LDA activity has been shown by Kubo et al. (1999). This led to the conclusion that LDA can partially complement for the lack of ISA activity and thereby is involved in the synthesis of starch. In agreement with this, it has recently been suggested for Arabidopsis that LDA might be involved in the synthesis of starch (Wattebled et al., 2005), providing an overlapping function to the isoamylases involved in the synthesis of starch. A more detailed introduction into the different factors influencing the accumulation of semi-crystalline starch in contrast to WSP can be found in Section 1.6.

1.5.4 Protein Complex Formation among Starch Synthesizing Proteins

The enzymes of starch synthesis all act on the same substrate; elongating, modifying and refining amylopectin structure. The local confinement and the high degree of interdependency between these 1. Introduction 21 ______enzymes require tight coordination of their activities. Protein-protein interactions are a possibility to achieve such coordination. In wheat amyloplasts, SBEII homodimers have been shown to exist both, alone and in complex with SSI or SSII isoforms. In addition, SSSI, SSSII and SBEII heterotrimeric complexes have been identified (Tetlow et al., 2004a; Tetlow et al., 2008). The phosphorylation status of the proteins in the complexes has been shown to be crucial for the stability and formation of the complexes. In maize, the formation of a large complex comprising SSSIII, SSSIIa, SBEIIa and SBEIIb has been shown as well as a smaller complex comprising SSSIIa, SBEIIa and SBEIIb (Hennen-Bierwagen et al., 2008b; Hennen-Bierwagen et al., 2009). Interestingly, DBEs have not been identified in any of the SSS and SBE complexes; they might be only very loosely attached or not be involved at all in complex formation with SSS and SBE. The regulation of the complex formation as well as the protein phosphatases and kinases controlling the phosphorylation status of the different proteins are still unknown. The specific functions of the different interactions are not fully unravelled at the moment. It has been shown, though, that the high molecular weight forms of SBEII in wheat have higher affinity for amylopectin in vitro than monomeric SBEII (Tetlow et al., 2008). In addition, pyruvate orthophosphate dikinase and the large and small subunits of the AGPase enzyme have been found in putative multisubunit complexes with SSS and SBEs in maize. This has led to speculation about regulation of carbon allocation and substrate channeling by means of complex formation. It has been suggested that on the one hand, ADP-Glc could be synthesized in close physical proximity of the enzymes using it to elongate glucan chains. On the other hand, local alterations of pyrophosphate levels by pyruvate orthophosphate dikinase might affect the activation state of AGPase (Hennen- Bierwagen et al., 2009).

1.5.5 The Architecture and Structure of Glycogen

Glycogen is a key energy reserve for animals, bacteria and fungi. Like starch, it is an α(1,4) polyglucan with α(1,6) branch points. In yeast and animals, its synthesis is primed by a protein, glycogenin, which adds glucose from the activated glucose donor to a tyrosine residue followed by the addition of approximately seven or more glucose residues (Alonso et al., 1995; Cheng et al., 1995; Sullivan et al., 2010). In eukaryotes, the activated sugar for glycogen synthesis is UDP-Glc, while bacteria use ADP-Glc for glycogen synthesis. Chain elongation is catalyzed by and branch points are homogenously introduced by glycogen branching enzymes (Manners, 1991). In contrast to starch, the glycogen branching pattern is thought to be homogenous with an average chain length of DP11 and an approximate branching level of 9% (Sullivan et al., 2010). In brief, glycogen accumulates as small particles of 10-30nm in size and has been suggested to be limited in size by the density of the glucan chains on its surface. The short average chain length and homogenous branching prevent the formation of regular structures like those seen in starch, hence glycogen is essentially 1. Introduction 22 ______water soluble (Meléndez et al., 1998). Thus, even though starch and glycogen are made out of the same components, their physical properties are completely different.

1.6 Factors Affecting the Crystallinity of the Starch Granule

The capacity to synthesize starch is thought to have arisen from the ancestral capacity to synthesize glycogen (Ball and Morell, 2003a). The uptake of a cyanobacterial cell into a nonphotosynthetic eukaryotic host (endosymbiosis) might have been one of the earliest steps in this process. It had long been reported that cyanobacteria accumulate glycogen (Ball and Morell, 2003b) but recent reports provide evidence that amylopectin-like polyglucans with average chain lengths comparable to higher plant amylopectin can be synthesized in some species (Nakamura et al., 2005; Deschamps et al., 2008). GBSS has been suggested to be among the earliest factors contributing to such an evolution (Ral et al., 2006; Zeeman et al., 2010). GBSS is not required for starch granule synthesis in higher plants, however, it has been reported to synthesize crystalline structures within starch granules or to produce small amounts of insoluble polyglucan, even in the absence of starch granules, in Chlamydomonas (Dauvillee et al., 1999; Wattebled et al., 2002). Its capability to synthesize the long chains required for crystallization competent polyglucans might have been taken over by other starch synthases.

DBEs certainly represent another enzymatic activity which is important for the synthesis of insoluble starch granules as has been described in Section 1.5.3. However, the accumulation of WSP in plants lacking ISA1 or ISA2 is often not exclusive and starch granules are still observed in the same plastid (Zeeman et al., 1998c), in the same cell or in different cell types (Delatte et al., 2005). Other DBEs, like LDA and ISA3, might complement for the loss of ISA1 or ISA2 as is suggested by the fact that quadruple DBE mutants of Arabidopsis are devoid of starch granules in all tissue types (Streb et al., 2008). The accumulation of starch granules can as well be influenced by SBE activities. The ae mutation (amylose extender, missing SBEIIb (Fisher et al., 1996; Gao et al., 1997) in maize has been shown to be at least partially epistatic over the sugary1 mutation, reverting the sugary1 phenotype back to starch accumulation (Black et al., 1966; Ayers and Creech, 1969). Likewise, a lack in the GBSS or SSIII activities causes a decrease of the phytoglycogen accumulation phenotype of the sugary1 mutation in maize (Black et al., 1966). Inversely, the over-expression of rize BEIIb has led to the accumulation of a water-soluble polysaccharide (Tanaka et al., 2004). It has also been shown recently that synthesis of insoluble polysaccharides is possible without any DBEs. A quadruple mutant, lacking all DBEs in Arabidopsis accumulates only WSP and no more starch granules. However, if an endoamylase (AMY3) is missing at the same time, small but insoluble polyglucan aggregates, resembling starch granules, are observed again (Streb et al., 2008). This suggests that enzymes thought to be involved in starch degradation may influence the structure and nature of the 1. Introduction 23 ______accumulated polysaccharide. It appears likely that nascent glucose chains are susceptible to degradation even during synthetic phases, when crystallization of the glucan polymer has not yet occurred. Degradation of glucan chains will create short chains and thereby render it less likely that the polyglucan as a whole can form semi-crystalline structures. It therefore appears that a balance between chain elongation, chain branching, debranching and also shortening of chains controls the crystallization state of a polyglucan. Thus, alterations in relative activities of enzymes will define the structure of a polyglucan. Any analysis trying to explain the accumulation of a changed polyglucan has to consider the network of starch synthetic (and even degrading) activities. Furthermore, comparisons of expression or activity levels have to be made on an isoform basis, as different isoforms of enzymes have different specificities for different chain lengths and branch point distributions.

1.7 Starch Degradation in Leaves

Being an important polymer for mankind, the interest has been large to determine the pathways of starch degradation. Most of the earlier work on starch degradation however, has been done in cereals, where starch is made and stored in amyloplasts in the endosperm and where starch breakdown is thought to proceed via α-amylolytic degradation; triggered by gibberellic acid released by the embryo. However, the degradation of starch in the endosperm is quite different from the breakdown of starch in photosynthesizing tissue, where degradation is not developmentally separated from synthesis. For example, leaf starch degradation in Arabidopsis can be triggered in the middle of the light period, by simply shading the plants or lowering the light intensity (Zeeman et al., 2002). Such a rapid switch from synthesis to degradation requires tight regulation and fast induction of the degradation process. It is therefore not surprising that the process of starch degradation is different inside leaf chloroplasts as compared to the cereal endosperm (Zeeman et al., 2004b; Yu et al., 2005).

1.7.1 Starch Phosphorylation

First hints at the importance of transient starch phosphorylation in the context of starch degradation were obtained by identifying proteins with starch binding capacity. Such a protein (termed R1) was identified in a starch binding screen by (Lorberth et al., 1998). Potato plants with decreased R1 protein levels had reduced phosphate levels in their starch (reduced to about 10-50% as compared to wild type) and lost the ability to efficiently remobilize the transitory leaf starch during the night. By screening mutagenized Arabidopsis populations for such starch excess mutants ( sex ) and applying map-based cloning techniques, the R1 ortholog in Arabidopsis was identified (Yu, Kofler et al. 2001). The Arabidopsis R1 gene encodes a glucan, water dikinase (GWD, α-glucan, water dikinase; EC 1. Introduction 24 ______

2.7.9.4), an enzyme which transfers the β-phosphate of ATP to glucose residues in starch and the γ- phosphate to water (Yu et al., 2001; Ritte et al., 2002). A drastic increase in the amount of leaf starch present at the end of the night and strongly reduced phosphate levels in the starch granules were the consequences of diminished GWD activity. A second glucan, water dikinase has been identified, able to phosphorylate starch, pre-phosphorylated by GWD. Consecutively, the enzyme was named phosphor-glucan, water dikinase (PWD; EC 2.7.9.5) (Kötting et al., 2005). Characterization of the gwd and pwd mutant phenotypes and the enzyme’s specificities established that GWD specifically phosphorylates the C6 positions of glucose residues, while PWD phosphorylates C3 positions (Ritte et al., 2006).

Transient phosphorylation of the starch granule surface is increased during breakdown, in accordance with the importance of granule surface phosphorylation as a prerequisite for the initiation of the degradation (Ritte et al., 2004). Using suspensions of insoluble crystalline maolto-oligosaccharides as a model for the starch granule, it has been shown that GWD has high affinity for crystalline maltodextrins and that GWD mediated phosphorylation results in increased solubilisation of the glucan substrate (Hejazi et al., 2008). The structural specificity rather than the chemical composition of the C6-phosphorylated glucan chains seem to be the requirement for subsequent C3 phosphorylation by PWD (Hejazi et al., 2009). It has been speculated that the C6 phosphorylation might induce hydration and cause voids between helices, resulting in helix-helix separation and thereby stimulating amylolytic degradation (Fig1-3). The C3 phosphorylation has been suggested to have ‘helix breaker’ action, thus permitting exo-amylases to access glucan chains and branch points (Blennow and Engelsen, 2010). The stimulatory effect of granule phosphorylation on exo-amylases has been studied by investigating the activities of recombinant β-amylases and DBEs. β-Amylases are exo-acting amylases, releasing maltose from glucan chains (see Section 0 for details). GWD catalyzed phosphorylation stimulated the activity of β-amylases by 2-3 fold. Likewise, the debranching activity of ISA3 was stimulated by phosphorylation and the simultaneous presence of β-amylases. Interestingly, the activity of GWD was also stimulated by the β-amylase activities (Edner et al., 2007). These findings suggest interdependence between the enzymes of starch degradation (Fig1-3).

1. Introduction 25 ______

1.7.2 Degradation of Linear Chains in Starch Degradation

α-Amylases are among the first enzymes which have been studied for their importance in starch degradation. α-Amylases are endo-amylases (1,4-α-D-Glucan glucanohydrolase; EC 3.2.1.1.) which can release branched malto-oligosaccharides from starch granules (Delatte et al., 2006). Three genes encoding α-amylases are found in the Arabidopsis genome, only one of which has a predicted chloroplast transit peptide (Stanley et al., 2002).

7 Branched malto- Chloroplast Starch oligosaccharides Stroma AMP + Pi 5,6 Pi 4 5 Linear malto- ATP oligosaccharides 3 8 3 Glucose 10 Maltose ATP

GWD

PWD 1,2 BAM1 9 Chloroplast BAM3 3 AMP + P i inner envelope SEX4 4 ISA3 5

Figure 1-3. Starch Degradation in Arabidopsis Chloroplasts Starch degradation in Arabidopsis leaves is initiated by disruption of the tight clustering of glucan chains by phosphorylation of glucose units through glucan, water dikinase (GWD; 1) and subsequently phosphoglucan, water dikinase (PWD; 2), a process requiring ATP (see inset, which represents a magnification of the granule surface). Glucan chains are degraded to maltose by β- amylases (BAM1 & BAM3; 3). BAM3 represents the major activity. Another β-amylase, BAM4, whose function in starch degradation is unclear, has been omitted from the model (see text for details). Linear malto-oligosaccharides are released from the granule surface by hydrolysis of the α(1,6) branch points by the debranching enzyme isoamylase (ISA3; 5). Branched malto- oligosaccharides are released from the granule by the endo-acting α-amylase (7). Small, soluble branched malto-oligosaccharides are debranched by LDA and ISA3 (5,6) and further degraded to maltose by β-amylases (3). Maltotriose units occur when β-amylases act on chains with an odd number of glucose units. Maltotriose cannot be degraded by β-amylases and thus is recycled by disproportionating enzyme (DPE1; 8). DPE1 releases glucose from maltotriose and elongates existing glucan chains with the maltosyl unit. Phosphorylated glucose residues limit β-amylase degradation of glucan chains. C3 and C6 bound phosphate is removed by the glucan-phosphatase, SEX4 (4). The major starch breakdown product, maltose, is exported from the chloroplast via the maltose transporter (MEX1; 9). Glucose can be exported from the chloroplast via the glucose transporter (10). Figure modified from Zeeman et al., (2010). 1. Introduction 26 ______

Arabidopsis mutants lacking plastidial α-amylase do not display a sex phenotype. In fact, all the α- amylase single knockouts as well as the triple mutant show a normal starch degradation pattern (Yu et al., 2005). Thus, α-amylases are not key to starch degradation in Arabidopsis leaves. However, they have been implicated in a minor pathway of starch degradation, in which α-amylases release small, branched malto-oligosaccharides from granules which can then be debranched and degraded in the stroma (Delatte et al., 2006). The existence of such a minor pathway is suggested by the observation that small, branched malto-oligosaccharides accumulate when the ISA3 and LDA DBE are missing and thus, the branched malto-oligosaccharides cannot be degraded. Branched malto-oligosaccharides are not detectable in an isa3/lda/ α-amylase triple mutant, suggesting that they were released by α- amylase (Dr S. Streb, personal communication, Fig1-3).

Another enzyme which has long been supposed to be responsible for the degradation of starch was α- glucan phosphorylase ( α-1,4-glucan phosphorylase; EC 2.4.1.1), an exo-acting enzyme which can release Glc-1P from glucan chains. Phosphorylase activity has been shown to be present in the chloroplasts of a range of species (Beck and Ziegler, 1989) and the production of Glc-1P has been suggested to represent an energetically favourable end product of starch degradation for transport to the site of sucrose synthesis in the cytoplasm. In Arabidopsis however, a knockout of the plastidial α- glucan phosphorylase does not cause a sex phenotype and potato plants with reduced α-glucan phosphorylase levels are not impaired in starch metabolism (Sonnewald et al., 1995; Zeeman et al., 2004b). The function(s) of the α-glucan-phosphorylase is (are) still largely unknown and the enzyme has recently been suggested to be involved in starch granule initiation, owing to its ability to elongate glucan chains using Glc-1P as substrate. In addition, a low starch phenotype in rice pho1 mutants (lacking the plastidial α-glucan phosphorylase) has been observed when mutant plants were grown at 20°C but not when grown at 30°C (Satoh et al., 2008).

Even though α-amylase as well as α-glucan phosphorylase might contribute to starch degradation in Arabidopsis leaves under certain conditions, hydrolysis of glucan chains is primarily catalyzed by β- amylases (BAM, α-1,4-glucan maltohydrolase; EC: 3.2.1.2). The identification of the β-anomer of maltose as the major starch degradation product and its increase during night-time starch degradation supports the importance of BAMs in starch degradation (Critchley et al., 2001; Niittylä et al., 2004; Weise et al., 2004; Weise et al., 2005). Nine genes encoding BAMs or BAM-like proteins have been identified from the Arabidopsis genome, four of which (BAM1-4) were shown to be chloroplast localized (Lao et al., 1999; Fulton et al., 2008). Among those, BAM1 and BAM3 are active BAMs; BAM2 is also active but has very low specific activity. No activity at all could be detected for BAM4 (Fulton et al., 2008; Li et al., 2009). Mutants lacking BAM3 have a sex phenotype, whereas the single bam1 mutant is not impaired in starch degradation. The double bam1/bam3 mutant, however, has a largely increased sex phenotype as compared to the bam3 single mutant. Therefore BAM3 and BAM1 both are important contributors to starch degradation with BAM3 being able to fully compensate for 1. Introduction 27 ______the loss of BAM1 whereas BAM1 only partly compensates for loss of BAM3 (Fig1-3). It is unclear so far, whether BAM1 and BAM3 have distinct functions in starch degradation. This lack of specific knowledge taken together with observations suggesting that some BAMs are present in high molecular weight (HMW) forms prompted us to look in more detail at BAM1 (and BAM3); work explained in Chapter 4. Surprisingly, mutants lacking BAM4, have a sex phenotype, even though biochemical as well as bioinformatical studies suggest the protein is non-catalytic. The triple bam1/bam3/bam4 mutant has an increased sex phenotype even compared to bam1/bam3 (Fulton et al., 2008). This suggests that BAM4 influences starch degradation in a mechanism superimposed to β- amylolytic glucan chain hydrolysis. Maltose sensing and transduction of maltose levels to adjust starch degradation might be possible functions of BAM4.

Of the remaining five BAMs (BAM5 to BAM9), BAM5 has been shown to be localized to the phloem and its function remains unclear at the moment (Wang et al., 1995; Laby et al., 2001). BAM7 and BAM8 have been shown to be localized to the nucleus (Dr H. Reinhold, personal communication) and are likely involved in regulatory functions. BAM6 and BAM9 have recently been shown to be chloroplast localized (Dr S.-K. Lee, personal communication) and thus could be involved in starch degradation. Their functions in this process are unclear at the moment and remain to be investigated. A comprehensive picture of the diverse functions of this large and interesting gene family is yet missing but would be highly desirable.

The phosphorylation of glucose residues is not only required for β-amylolytic starch degradation but also limits the progression of β-amylases, which are not able to work past phosphorylated glucose residues (Takeda and Hizukuri, 1981). Likewise, the presence of α(1,6) linkages limit β-amylolytic activity. Therefore, for complete and efficient hydrolysis of starch granules, further enzymes are required. These are on the one hand DBEs, which are discussed in the following section and phospho- glucan phosphatases which remove the phosphate from C6 and C3 phosphorylated glucose residues and which are discussed in Section 1.7.4.

1.7.3 Debranching Enzymes in Starch Degradation

ISA3 and LDA are the DBEs active in starch degradation. Both enzymes preferentially remove short branches (Wu et al., 2002; Hussain et al., 2003; Delatte et al., 2006; Takashima et al., 2007) and have maximum activity on β-limit dextrins (dextrins digested by an excess of β-amylase, which have external chains of DP2 or DP3). In Arabidopsis , loss of the ISA3 activity results in a sex phenotype (Wattebled et al., 2005; Delatte et al., 2006). The amylopectin in the Atisa3 mutant displays a higher portion of very short chains (DP3-DP5) as compared to the wild type, reflecting the decreased capacity to remove short branches during starch degradation. Atlda mutants do not display a sex phenotype but the lack of LDA in an Atisa3 mutant background causes an increase of the sex 1. Introduction 28 ______phenotype, suggesting functional overlap between ISA3 and LDA. In addition to the increased sex phenotype, Atisa3/Atlda double mutants accumulate small, branched malto-oligosaccharides during the night, which are released by the plastidial α-amylase AMY3 (see Section 0 and Fig1-3). The relative contributions of ISA3 and LDA to starch degradation may be different between species, which is shown by the mild sex phenotype of maize mutants lacking LDA activity (Dinges et al., 2003).

1.7.4 Starch De-Phosphorylation

In Arabidopsis , the removal of phosphate from phosphorylated glucose residues is a requirement for complete β-amylolysis (Fig1-3). This is accomplished by the phospho-glucan phosphatase SEX4 (Kerk et al., 2006; Niittylä et al., 2006; Sokolov et al., 2006; Kötting et al., 2009). The protein contains a dual-specificity phosphatase (DSP) domain and was originally hypothesized to be a protein phosphatase (Kerk et al., 2006; Niittylä et al., 2006). Later it was shown to be able to dephosphorylate both C3 and C6 phosphorylated glucose residues in starch (Kötting et al., 2009; Hejazi et al., 2010). In addition, SEX4 binds to starch granules, possibly via an N-terminal carbohydrate-binding module (CBM). Mutants lacking SEX4 have a sex phenotype and when non-phosphorylated starch granules were incubated in vitro with BAM3, ISA3 and GWD as well as ATP, the addition of SEX4 increased the release of glucan from the granules by more than 2-fold (Kötting et al., 2009). A second phospho- glucan phosphatase-like protein can be found in Arabidopsis . This protein, LSF1 (Like-SEX-Four 1) is present in the chloroplast and mutants lacking LSF1 are also impaired in starch degradation. However, sequence comparisons as well as genetic analyses imply that the function of LSF1 may be fundamentally different from SEX4 (Comparot-Moss et al., 2010). In contrast to SEX4, LSF1 contains a sequence motif known to mediate protein-protein interactions. The lack of knowledge about LSF1 functions as well as its potential to be present in high molecular weight forms prompted us to investigate LSF1 in more detail, work which is presented in Chapter 4.

1.7.5 Metabolism of Starch Breakdown Products

Maltose is the major starch degradation product, but owing to the fact that β-amylases can only degrade glucan chains to within 2-3 glucose units from branch points, maltotriose is produced by DBEs, releasing these short chains from the granule. Furthermore, glucan chains with an uneven number of glucose units, released from the granule surface by ISA3, will also give rise to maltotriose, as β-amylases will shorten longer chains but are not able to metabolize maltotriose. Maltotriose is metabolized by a disproportionating enzyme (DPE1, α-1,4-glucan 4-α-glucanotransferase; EC: 2.4.1.25). This enzyme transfers the maltosyl group from maltotriose onto an existing glucan chain and thereby elongates this chain as well as it releases glucose (Critchley et al., 2001). The plastidial 1. Introduction 29 ______disproportionating activity is required for starch degradation as shown by the sex phenotype of the Arabidopsis dpe1 mutant. Maltose released by β-amylolysis is exported from the chloroplast by the maltose transporter MEX1 (Niittylä et al., 2004) and glucose released by DPE1 action is exported from the chloroplast by the glucose transporter of the inner envelope (Weber et al., 2000).

In the cytosol, another glucanotransferase, DPE2 (4-α-glucanotransferase; EC 2.4.1.25), processes most of the maltose. DPE2 is a transglucosidase which transfers one of the glucose units of maltose to heteroglycans in the cytosol, thereby releasing glucose (Steichen et al., 2008). The cytosolic heteroglycans are heterogeneous polysaccharides, consisting mainly of galactose and arabinose. Mannose, glucose, fucose and xylose are minor constituents. Glucose added to the heteroglycan by DPE2 has been suggested to be released by the cytosolic phosphorylase yielding Glc-1P (Fettke et al., 2004; Fettke et al., 2009). In mutants lacking DPE2, starch degradation is reduced and maltose accumulates to even higher levels than in mex1 mutants (Chia et al., 2004; Lloyd et al., 2004). The amount of glucose in cytosolic heteroglycans is low and the turnover of glucose must be relatively rapid. The function of the cytosolic heteroglycan might be to serve as a buffer between starch degradation and sucrose synthesis (Chia et al., 2004). Both, glucose exported from the chloroplast and Glc-1P released from heteroglycans can feed into the Glc-6P pool and thus feed sucrose synthesis for export to heterotrophic tissues (Fig1-2).

1.8 Cecropia peltata , Glycogen Synthesis in a Vascular Plant

1.8.1 Cecropia peltata – A Many-Sided Subtropical Tree

It has been noted in Section 1.2 that plants generally accumulate starch and not its water-soluble counterpart glycogen. Cecropia peltata represents a notable exception in the plant kingdom as it has been suggested to synthesize glycogen-like polysaccharides as well as starch simultaneously in different tissues.

Cecropia peltata , also called ‘trumpet tree’ or ‘yagrumo hembra’, is a fast growing neotropical tree. Mainly growing in disturbed habitats, it is native throughout the Greater and Lesser Antilles, in Central America from Yucatan, Mexico to Costa Rica (Fleming and Williams, 1990). The tree has been introduced to a number of African and South-East Asian Countries, where it has become an invasive species, competing with native pioneer species. As a matter of fact, C. peltata has been placed among the 100 most invasive species by the invasive species specialist group (http://www.issg.org/database/). The tree grows up to 20 meters high, has large leaves and a small crown. Its lifespan reaches 20-30 years (Fig1-4 A).

The light wood of C. peltata is often used for matchsticks, boxes and crates but also interior boarding and paper pulp. Leaf extracts of Cecropia species are widely used in folk medicine to cure a variety of 1. Introduction 30 ______diseases, among them arthritis and rheumatism (Perez-Guerrero et al., 2001). Even anti-malarial activity of C. peltata extracts has recently been reported (Uchoa et al., 2010). Best studied however, is the effect of C. peltata aqueous leaf extracts against type 2 diabetes (Andrade-Cetto and Heinrich, 2005). The hypoglycaemic effects of extracts have been shown in different model systems (Andrade- Cetto and Wiedenfeld, 2001; Cunha et al., 2004; Nicasio et al., 2005; Aragão et al.,2010). It has been suggested that the bioactive compounds in the extracts are chlorogenic acid and isoorientin (Andrade- Cetto and Wiedenfeld, 2001; Nicasio et al., 2005). While isoorientin potentially acts as an antioxidant, preventing the complications of diabetes, chlorogenic acid has been implicated to have hypoglycaemic effects by inhibiting the Glc-6P activity and thereby lowering hepatic glucose output (Andrade-Cetto and Heinrich, 2005; Andrade-Cetto and Vazquez, 2010).

Apart from interesting ethnopharmacological properties, Cecropia species are also well known for their potential myrmecophytic lifestyle. In most of their native range, a number of Cecropia species are inhabited by stinging ants, of the genus Azteca (Folgarait et al., 1994; Faveri and Vasconcelos, 2004) . The relationship between C. peltata and the ants has been described as symbiotic, with the ants living in the hollow stem of the tree and feeding on food structures produced by the tree. In return, ants have been suggested to protect the tree from herbivory and competition from vines (Janzen, 1969; Rickson, 1976a). However, Cecropia species exist which are not associated with ants and because it has been shown that Cecropia suffers a low incidence of attack by insect herbivores in the absence of ants (Faveri and Vasconcelos, 2004), the relationship should be termed facultative mutualism at best. The myrmecophytic capacity of Cecropia species has received quite some attention from a biochemical point of view, related to the food structures produced to feed the ants, the so- called ‘Müllerian bodies’.

1.8.2 Müllerian Bodies – Myrmecophytic Food Structures of Cecropia peltata

In 1876, Müller first described the myrmecophytic food structures (Müllerian bodies, MBs) produced by C. peltata trees (Müller, 1876). These are small (approximately 3mm x 1mm), egg-like bodies, produced by the tree on a pad of tissue densely covered in trichomes at the petiole-stem interface (Fig1-4 B). The pad of tissue from which the MBs emerge has been termed trichilium, owing to the high number of trichomes present on it, resulting in an almost felt-like, brown appearance of the structure. In mature trichilia, almost every epidermal cell has been converted into a simple trichome and the trichilium possesses a number of parallel vascular strands which are located approximately 20 cell layers beneath the epidermal surface as well as a network of bundles running throughout the internal trichilium tissue. The maturation of the trichilium of a given leaf is completed while the leaf is still tightly enveloped in the sheathing stipules of the subjacent leaf (Rickson, 1976a). Initially, trichilia are of a white colour but turn brown, as soon as they get exposed to sun and air. 1. Introduction 31 ______

MBs are produced from a trichilium just before the new leaf is exposed by stipule abscission. Initial events leading to the production of an MB consist of a series of cell divisions in the second or third cell layer beneath the epidermis. These cell layers contain chloroplasts with starch granules and well developed grana and thylakoid systems (Rickson, 1976b). Expansion and continued cell division push the epidermis upwards. While this happens, anticlinal cell divisions occur in the epidermis, so that the mature MB is covered by an epidermal cell layer. Cell divisions continue and finally raise the body above the trichilium surface. A lessening of cell division rate at the tip and continued cell expansion in the lower regions causes the egg-shaped structure of the food body (Fig1-4 B and Rickson (1976a)). No vascular tissue is associated with the MBs and no transfer cells were described between the body and the subjacent trichilium vascular tissue. Even though the production of MBs is stimulated by their removal (Folgarait et al., 1994), it is not dependent on any ant stimulus and can be seen on greenhouse specimens. In the absence of ants, which harvest the fully grown bodies, the MBs abscise at about the level of the trichilium.

1.8.3 Glycogen Deposition in Müllerian Bodies

Müller (1876) already observed that MBs stain dark yellow-brown with iodine solution and red with sulphuric acid. From these observations he concluded that the MBs mainly contained proteinaceous substances. Light microscopy has, however, shown that when an MB has reached three-fourths of its final size, large amounts of polysaccharide are deposited , starting from the tip of the body. Finally, each cell contains 20-30 plastids and ultrastructural investigations using transmission electron microscopy showed that chloroplasts, which had been present in the cells from which the MB differentiation started, had dedifferentiated to proplastids and were filled with small particles reminiscent of glycogen particles (Rickson, 1971, 1976b). Even though the plastids, 8-10µ in diameter and essentially spherical, had lost their thylakoid system, remnants of thylakoids were observed. When MBs were collected and the water-soluble polysaccharide accumulating in them was extracted and alcohol precipitated, a yield of 296mg total polysaccharide (dry weight) per gram MB

(fresh weight) could be obtained. The wavelength of maximum absorbance ( λmax ) of the polysaccharide-iodine complex was determined to be 500nm, which is considerably lower than the

λmax of amylopectin and slightly higher than that of the WSP, described at the time only from the maize sugary1 mutant (Archibald et al., 1961; Marshall and Rickson, 1973). From the degree by which isoamylases and limit-dextrinases degraded the MB polysaccharide it was further concluded that while C. peltata accumulated normal, water insoluble starch granules in leaves, a glycogen-like polysaccharide (WSP) accumulated in MBs. It was pointed out that since MBs do not contain starch granules at all, this may a better system to understand the biosynthesis of phytoglycogen than the sugary1 mutants of maize (see Section 1.5.3), where starch granules and phytoglycogen are present simultaneously (Marshall and Rickson, 1973). 1. Introduction 32 ______

A B

Figure 1-4. Cecropia peltata in Greenhouses of the Botanical Garden Bern and at the ETH Zürich (A) Trees of the species Cecropia peltata are slender, tall growing trees, native to subtropical regions of South and Central America. They mostly occur in disturbed habitats where they dominate in early succession stages. A full grown Cecropia peltata tree is found in the greenhouse of the Botanical Garden in Bern. (B) Head cuttings were made from newly emerging branches and transferred to pots in greenhouses at the ETH Zürich, as soon as small plants had established. A brownish pad of tissue (trichilium) develops at the abaxial side of the petiole-stem interface. Müllerian bodies are small, egg-like structures emerging from trichilia. These are made by the mature tree in Bern as well as by the small trees grown in greenhouses at the ETH Zürich (inset). Müllerian bodies are generally produced by the four youngest trichilia, even when the leaf is still covered by the stipule. Older trichilia stop producing Müllerian bodies, even when leaves are still healthy and green.

1. Introduction 33 ______

1.9 Scope of the Work Presented

In a first approach, this work aims at improving our understanding of the coordination of starch degradation. The fact that the enzymes involved in starch degradation depend on each other makes us believe that direct physical interactions might help such coordination. Investigating protein-protein interactions is a way to approach the regulation of starch degradation, where very little information is available at the moment.

The minimal set of enzymes required for the synthesis of semi-crystalline starch granules is a question which has not been resolved yet in investigations using neither model systems like Arabidopsis or Chlamydomonas nor crop species like maize or rice. By looking at how C. peltata synthesizes the WSP in MBs, in a second part of my work, I aim at adding knowledge about the impact of perturbations in the relative activities of starch synthetic enzymes on structure and crystallization competence of polyglucans. Knowledge which may be transferred to model or crop species in order to potentially alter crystallization state as well as the amount of the polysaccharide produced in leaves or storage organs. 2. Material and Methods 34 ______

2. Material and Methods

2.1 Plant material

Plants used for metabolite measurements were grown in a climate controlled Percival AR95 growth chamber (Perceival Scientific Inc., Perry USA). The diurnal cycle was set to a 12-h light and a 12-h dark period with a uniform light intensity of 150 µmol photons m -2 sec -1. Relative humidity was set to 70% and the temperature to constant 20°C. Plants for the extraction of soluble protein were grown in climate controlled chambers provided by Conviron (Type CMP3246, Winnipeg, Canada) with 12-h light/12-h dark conditions. Light intensity was 150 µmol photons m -2 sec -1, the relative humidity was 65% during the day and 60% during the night. The temperature was 22°C during the day and 18°C at night. Seeds were sown out onto specific fine-grade seed compost (ASB Greenworld Premium, Neuhausen, Switzerland) and transferred to potting compost (Substrat2, Klasmann-Deilsmann, Geeste, Germany), 2 weeks after germination. Seeds were covered with a clear propagating plastic top after sowing and stratified for 48h at 4°C. The propagating top was removed after cotyledons were fully emerged. All plants used in this work were Arabidosis thaliana of the Columbia (Col-2) ecotype. T-DNA insertion mutant lines were obtained from the Salk Institute (http://signal.salk.edu/) and the GABI-Kat project (Max Plank Institute for Plant Breeding Research, www.gabi-kat.de). The single mutant accessions used in this work are described in Table 2-1. The double mutants Atbam1 /Atlsf1 and Atbam3 /Atlsf1 were made by crossing the respective single mutants and by PCR based screening using primers described in Table 2-1. The triple Atbam1/Atbam3/Atlsf1 mutant was isolated from a cross between the double Atbam1/Atbam3 mutant and the Atlsf1 single mutant.

Table 2 -1: Mutant Lines Used in this Work and Primers For Genotyping

Gene- Mutant Name, AGI Type of Mutation Line Identifier Genotyping Primers (shown 5’ to 3’) Reference Alleles code BAM1 T-DNA insertion, Atbam1-1 SALK_039895 For: AGAACGTATAGAGAAGGAGGGATTG (Fulton et al., 2008) At3g23920 exon 1 Rev: CCGTCTCTGAACCTTGTGTTGTAGTA T-DNA: GCGTGGACCGCTTGCTGCAACT BAM2 T-DNA insertion, Atbam2-1 GK_132E06 For:GGCGATTAGGTTGAATCATAGTGTA (Fulton et al., 2008) At4g00490 exon 1 Rev: CAGATAGGACAGCACAGCAGA T-DNA: CCCATTTGGACGTGAATGTAGACAC BAM3 G to A substitution in Atbam3-1 CS92461 For: GAACAAGTGGACCTCATGATG (Fulton et al., 2008) At4g17090 exon 4 results in a Rev: TGAGAGTCTCCTCCCATGAC stop codon The wild-type amplicon (650bp) is cut by BsrI BAM4 T-DNA insertion, Atbam4-1 SALK_037355 59-GATGCTCGAGAGAAATCACGATCG (Fulton et al., 2008) At5g55700 inteon 6 59-TCTGCACTCATCTGTCTAATGAAAG T-DNA: GCGTGGACCGCTTGCTGCAACT LSF1 T-DNA insertion 104 Atlsf1-1 SALK_053285 For: AGTAAGAGGAGCTCGCCGAC (Comparot-Moss et At3g01510 bp upstream of ATG Rev: TTCGAGAGCTCCTAAACCGG al., 2010) T-DNA: TGGTTCACGTAGTGGGCCAT Mutant lines used in this work and the primers used for genotyping. Atbam2 and Atbam4 were only used for complementation analyses with the TAP-tagged full length cDNAs. Crosses between Atbam1 , Atbam3 and Atlsf1 were made. 2. Material and Methods 35 ______

2.2 Standard SDS-PAGE and Western Blotting

Separating gels used for the SDS-PAGE gels were 6, 7.5 or 10% (w/v) acrylamide gels depending on the protein size to be resolved. They contained 375mM Tris-HCl, pH 8.8 and 0.1% (w/v) SDS. The stacking gels used had a concentration of 3.8% (w/v) acrylamide, 0.125M Tris-HCl, pH 6.8, and 0.1% (w/v) SDS. Protein extracts were mixed with 1 volume of SDS sample buffer (0.1M Tris-HCl pH 6.8, 40% (v/v) glycerol, 3% (w/v) SDS, 0.015% (w/v) bromophenol blue). Gels were run at 25°C, at 200V constant, for 45min. Proteins were transferred onto PVDF membranes by electroblotting in 119mM Tris, 40mM glycine, pH 8.3, 10% (v/v) methanol (blotting buffer) at constant 100V for 1-2 h at 4°C.

Following transfer, membranes were washed in TBS (20mM Tris, pH7.5, 500mM NaCl) for 10 min, unspecific binding sites were blocked by incubating the membranes for 2h at 25°C in blocking solution (2% (w/v) non-fat dry milk in TBS). After blocking, membranes were washed in TTBS (TBS with 0.1% (v/v) Tween-20) for 10 min. Blots were probed with primary antibodies diluted in TTBS + 2% (w/v) non- fat dry milk. Antibody dilutions were: 1/1000 for SEX4, AMY3, ISA3, BAM3, GWD, PWD, DPE2; ESH and BAM2. For BAM1 and LSF1 antibody dilutions were 1/1000 if blots were made from SDS-PAGE and 1/500 if blots were made from native-PAGE (see Section 2.3). Blots were incubated at 4°C on a shaking table 16 h. The primary antibody solution was decanted and the blots washed twice for 10 min with TTBS. The membranes were incubated with the secondary antibody Goat-anti-Rabbit IgG (whole molecule) alkaline phosphatase conjugate (Sigma Aldrich, Buchs, Switzerland) diluted 1/6000 in TTBS + 2% (w/v) non-fat dry milk. Blots were developed using Sigma-Fast TM (Sigma Aldrich) alkaline phosphatase reagents, according to the manufacturers’ instructions

2.3 Native PAGE and native PAGE blotting

Soluble protein, were extracted from Arabidopsis plant leaf tissue in 100mM MOPS, pH 7,2; 1mM EDTA; 1mM DTT; 10% (v/v) ethanediol with a sample to medium ratio of 150mg/mL. For C. peltata , protein were extracted in 100mM MOPS pH 7.2, 5mM DTT, 10% Glycerol, 1mM EDTA, 50mg mL -1 PVPP, at 25mg mL -1 for leaves and 12.5mg mL -1 for MBs. Tissues were homogenized in all-glass homogenizers and cellular debris was removed by centrifugation for 10 min at 16000 g in a microfuge cooled to 4°C. Total soluble protein was determined using the Bradford-Kit from BioRad (Reinach, Switzerland). Unless otherwise indicated, 20µg of total protein were used to perform native PAGE gels which were performed as described previously (Zeeman et al., 1998a). Briefly, gels contained 6% (w/v) polyacrylamide and either 1% (w/v) oyster-glycogen (Sigma Aldrich), 0.2% (w/v) β-limit-dextrin (Sigma Aldrich), or 0.2% (w/v) potato amylopectin (Sigma Aldrich). After electrophoresis, gels were incubated for either 2 h at 37°C or 16 h at 25°C in a medium containing 100mM Tris, pH 7.2, 1mMMgCl2, 2. Material and Methods 36 ______

1mMCaCl, 2.5mM dithiothreitol. Gels were stained with Lugol solution (Iodine/Potassium iodide solution, Sigma Aldrich). Improved staining was achieved by incubating stained gels for 24 h at 4 °C and partial destaining by rinsing in water. For immunodetection of proteins in native gels, gels were incubated twice for 5 min in 20mM Tris, 150mM glycine, 1% (w/v) SDS, heated to 75°C. Gels were rinsed in blotting buffer and blotted as described for SDS-PAGE gels.

Gels assaying LDA activity contained 1% (w/v) Red-Pullulan (Sigma Aldrich) in the separating gel. Gels assaying SBEs were rinsed with 50mM HEPES-NaOH, pH7.5, 2mM DTT, 10% (v/v) glycerol and then incubated in 50mM HEPES-NaOH, pH7.5, 2mM DTT, 10% (v/v) glycerol, 50mM Glc-1-P, 2.5mM AMP, 50 units rabbit muscle phosphorylase a (Sigma Aldrich). Gels assaying SSSs contained 0.8% (w/v) oyster glycogen in the separating gel and were rinsed in a solution containing 100mM HEPES-NaOH, pH7.5, 2mM DTT, 10% (v/v) glycerol, 0.5mM EDTA, 0.5M Na-citrate and then incubated in the same solution supplied with 2mM ADP-glucose. Amylopectin (0.2% (w/v)) or glycogen (0.5% (w/v)) containing native gels for assaying MDH were incubated in 100mM Tris, pH8, containing 13.3mM + MgCl 2, 27µM phenazine methosulfate, 400µM nitrotetrazolium blue chloride, 502µM NAD and /or 448µM NADP + and 1.02 mM malate. In the presence of NADH or NADPH, nitrotetrazolium blue reacts with phenazine methosulfate to form an insoluble blue-purple formazan. Gels for LDA, SSS and SBE activity were incubated at 25°C for 16-18 h, gels for MDH activity were incubated 2-3 h at 25°C, until bands developed. Enzymatic activity could either be detected directly as a clear band on the red pullulan containing gels, as purple bands on MDH gels, or as bands of different colours after the SBE and SSS gels were stained with iodine solution.

2.4 Co-Immunoprecipitation of BAM1 and LSF1

Co-IP of BAM1 and LSF1 was conducted according to Serino and Deng (2007). In brief, 100 µl of the purified BAM1 and LSF1 antibodies were covalently coupled to 100µl Protein A-Sepharose 4B Fast Flow beads following manufacturer’s instructions (Sigma-Aldrich). Soluble proteins were extracted from wild-type, the Atbam1 and Atlsf1 leaves, freshly harvested at the end of the night period. Extractions were -1 performed at a 100 mg mL ratio in 50 mM Tris pH 7.5, 10 mM MgCl 2, 150 mM NaCl, 0.1 % (v/v) Triton X-100 and 1x Roche Life Sciences Protease Inhibitor (Roche Life Sciences, Basel, Switzerland) in glass homogenizers, at 4°C. Cellular debris was removed by centrifugation for 10 min, 20'000 g at 4 °C. Centrifugation was repeated with the supernatants to remove any insoluble material. Twenty µL antibodies coupled to beads were incubated with 400µL plant protein extract on a vertical rotary wheel for 4 h at 4°C. Beads were washed with 3 times 500µL 50 mM Tris pH 7.5, 10 mM MgCl 2, 150 mM 2. Material and Methods 37 ______

NaCl, 0.1 % (v/v) Triton X-100 and protein bound to the beads were eluted by boiling the beads in 50µL 1x SDS sample buffer. Eluates were analysed by standard SDS-PAGE and western blotting.

2.5 Gel Filtration Chromatography

Soluble protein for gel filtration chromatography (GFC) was extracted from leaves of 4-5-week-old plants harvested at the end of the night, in cold 100 mM tris pH 7.5, 10% (v/v) glycerol, 150 mM KCl 2, 5mM DTT, 1x Roche Life Sciences Protease Inhibitor, at a ratio of 800 mg mL -1, using cooled glass homogenizers. Cellular debris was removed by centrifugation at 18’000 g for 10 min at 4 °C. Supernatants were filtered through Minisart® Filter 20 µm (Sartorius Stedim Biotech, Aubagne Cedex, France) and concentrated using Amicon® Ultra-4 30 kDa Centrifugal Filter Devices (Millipore, Volketswil, Switzerland). Concentrated protein extracts were stored at -80°C until further use. GFC was performed with an ÄktaExplorer100 (GE Healthcare, Glattbrugg, Switzerland) and a custom-made Sephacryl 300HR™ (GE Healthcare) column with a total volume of 60mL in an XK16 casing (GE Healthcare). The column was prepared at 4°C, following the procedures suggested by the supplier (GE Healthcare). In brief, 1.5 times the column-volume of the Sephacryl 300HR™ resin was resuspended in 2 volumes 100 mM Tris pH 7.5, 10% (v/v) glycerol, 150 mM KCl 2. and poured into the column. The column was packed at a flow rate of 1mL min -1 for 2 h and 2.5mL min -1 for 1hr. In general, the column was operated on a vertical stand with a 100 mM Tris pH 7.5, 10% Glycerol, 150 mM KCl 2, 1mM DTT; 0.2 µm sterile filtered and degassed elution buffer. The flow rate was 0.4mL min -1for the separation of protein complexes and 1mL min -1for column wash and equilibration. Column calibration was performed using the Gel Filtration HMW Calibration Kit (GE Healthcare), according to the manufacturer’s instructions. For separation of Arabidopsis extracts, 200µl soluble protein extract with a maximum of

350µg total protein were applied to the column. Fractions of 0.5mL were collected, frozen in liquid N 2 and stored at -80°C. For SDS-PAGE and western blotting analysis, fractions were pooled as indicated in the results section.

2.6 Chloroplast Isolation and Measurement of GAPDH and PEP- Carboxylase Activities

Chloroplasts were isolated from intact protoplasts following a protocol described previosly (Ljerka Kunst, 1998). Protoplasts were prepared from rosette leaves of 5-6 week old wild-type and Atlsf1 mutant plants, grown under 12-h light / 12-h dark conditions. Leaves were cut into small pieces, and gently vacuum 2. Material and Methods 38 ______

infiltrated with 0.5M Sorbitol, 1mM CaCl 2, 0.5% (w/v) pectinase, 1% (w/v) cellulose (Onozuka R10), 10 mM MES pH 6.0. After 2h incubation in the dark, the digest was filtered through a 100 µm nylon mesh and spun for 5 min at 100 g, 4 °C in round-bottomed glass tubes. The pellet was resuspended in 5mL ice- cold 0.5 M sorbitol, 10 mM MES pH 6.0, 1mM CaCl 2 (PS-medium). Resuspended protoplasts were loaded on a 30mL percoll cussion (50% (v/v) percoll in 2xPS-medium), and spun for 10 min, 100 g at 4 °C, without deceleration. The dark green band containing intact protoplasts was removed and protoplasts pelleted for 5 min, 100 g at 4 °C. The pellet was resuspended in 5 mL PS-medium. Wild-type and Atlsf1 samples were adjusted to equal concentrations of intact protoplasts by counting protoplasts in a ‘Neubauer counting chamber’. Protoplasts were pelleted again, resuspended in 0.3 M sorbitol, 20 mM Tricine-KOH pH 8.4, 10 mM EDTA, 10 mM NaHCO 3. Protoplasts were broken by filtering through a 15 µm nylon mesh. Intact chloroplasts were collected by centrifugation for 5 min at 1000 g, 4 °C. The supernatant from this centrifugation contained mainly the cytosolic fraction of protoplasts, the pellet mainly the chloroplast fraction.

To determine cytosolic contamination of the chloroplast fraction and vice versa, NADP-GAPDH and PEP-Carboxylase were measured as chloroplast and cytosolic markers, respectively. For both enzymes, activity was determined by observing the decrease in absorbance at 340nm due to consumption of NADPH or NADH from NADP-GAPDH or PEP-Carboxylase activity, respectively. The NADP-GAPDH assay was done as described previously (Hancock et al., 2005). Different amounts of cytosolic and -1 chloroplast fractions were incubated in 100mM Tris, pH8, 10mM MgCl 2, 2.5mM 3-PGA, 8 units mL 3- PGA Kinase (EC 2.7.2.3), 2.5mM ATP, 0.2mM NADPH, 5mM DTT. For the PEP-Carboxylase assay, different amounts of cytosolic and chloroplast fractions were incubated in 100mM Tris, pH8, 10mM -1 MgCl 2, 10mM NaHCO 3, 0.2mM NADH, 2.5mM PEP, 10units mL malate dehydrogenase (EC 1.1.1.37). Linear decrease of absorbance at 340nm was monitored using a Discovery XS-R microtiter plate reader (BIO-TEK Instruments, Luzern, Switzerland). Linear rates of decrease in absorbance at 340nm were used to relatively compare the NADP-GAPDH and PEP-Carboxylase activities between the different samples based on equal volumetric inputs into the assay.

2.7 Carbohydrate Extraction and Measurements

Whole rosettes of 3-4-week-old plants were harvested at the times indicated in the results section and frozen directly into liquid N 2. Plant material for the complementation analysis was frozen in 96-format collection tubes and pulverized while still frozen using a Mixer Mill (Retsch, Haan, Germany). The frozen powder was extracted in ice-cold 0.7 M perchloric acid for 30 min with intermittent mixing. Plant 2. Material and Methods 39 ______material for starch quantification in the Atbam1 , Atbam3 and Atlsf1 single and multiple mutants was harvested into 1.5mL microcentrifuge tubes, homogenized frozen using the Mixer Mill and extracted as for the samples in the 96-format. C. peltata leaves and MBs were extracted using ice-cold 0.7M perchloric acid in cooled glass homogenizers. Starch was separated from soluble sugars (soluble fraction) by centrifugation for 5 min at 3000 g, 4°C

2.7.1 Quantification of Starch and WSP

Pellets, containing the starch fraction were washed three times with ethanol and then with water by repeated resuspension and centrifugation for 5 min at 3000 g, 4°C until pellets were white. Pellets were resuspended in a total volume of 1mL water. The soluble fraction was neutralized by adding cold 2M KOH 0.4M MES 0.4M KCl until a pH of between 5 and 7 was reached. Precipitated potassium perchlorate was removed by centrifugation for 10 min at 4°C, 16100 g.

Two hundred and fifty µL of the insoluble fraction were boiled for 15 min, cooled to room temperature and mixed with 250 µL 50mM sodium acetate pH4.8, 12.6 units of amyloglucosidase (EC 3.2.1.3) and 10 units of α-amylase (EC 3.2.1.98). For WSP quantification, 100 uL of the soluble fraction were mixed with 100 uL 50mM sodium acetate pH4.8 and 5 µL enzyme mixture containing 6.3 units of amyloglucosidase and 5 units of α-amylase. Starch and WSP were hydrolyzed to glucose at 37°C for 2h. For glucose quantification, 25 µL of the digest were mixed in a 200 µL assay containing 25 mM Hepes, pH 7.5, 1 mM + MgCl 2, 1 mM ATP, 1 mM NAD and 1.4 units hexokinase (EC2.7.1.1). The initial OD (340nm) was measured, and then 1 unit of glucose-6-phosphate dehydrogenase (EC1.1.1.49) was added to start the reaction. The conversion of NAD + to NADH was monitored using a Discovery XS-R microtiter plate reader (BIO-TEK Instruments). The amount of NADH produced is proportional to the amount of glucose in the assay. Starch or WSP content of the sample was calculated and expressed as mg glucose equivalents mg -1 fresh weight. Maltose, sucrose glucose and fructose were determined using high- performance-anion-exchange chromatography (HPAEC-PAD). Samples of the neutralized soluble fraction (100 µL) were applied to sequential 1.5-mL columns of Dowex 50 W and Dowex 1 (Sigma Aldrich), see Section 2.7.2. Neutral compounds were eluted with water, lyophilized and redissolved in 100 µL of water for HPAEC-PAD analysis (see Section 2.8).

2.7.2 Chain Length Distributions of Starch and Water Soluble Polyglucans

Insoluble starch or water soluble polyglucan (WSP) samples equivalent to 0.1mg glucose were diluted to a final volume of 470 µL with water. Samples were boiled for 15 min, cooled to room temperature and 25 µL 50mM sodium acetate pH3.5 with 10`000 units Pseudomonas amyloderanosa isoamylase (EC 2. Material and Methods 40 ______

3.2.1.41) and Klebsiella planticolla pullulanase (EC 3.2.1.68) (both from Sigma-Aldrich) were added. After incubation at 37°C for 2h to debranch starch and WSP, insoluble material was removed by centrifugation, (5 min, at 16100 g, 20°C). Reactions were stopped by boiling samples (10 min) and insoluble material was removed again by centrifugation (10 min at 1000 g). Charged compounds were removed from the uncharged glucan chains using DOWEX anion-/cation-exchange resins. DOWEX50 (cation-exchange) was charged using 2M HCl and DOWEX1 (anion-exchange) was charged using 1M sodium acetate. The resins were washed with water until the pH of the water was between pH 6 and 7. The resins were distributed into columns (barrels of 2-mL syringes plugged with Miracloth to retain the resin), such that each syringe contained 1.5 mL resin. Two hundred µL of the debranched samples or soluble sugars were diluted to 1mL with 800µL water, loaded onto the columns and allowed to flow through. Uncharged compounds were eluted with 3 washes of 1mL water, lyophilized, redissolved in 120 µL water and centrifuged for 10min 3000 g. One hundred µL were transferred to snap-cap vials for HPAEC-PAD (see Section2.8).

β-CLD is a variation of the CLD technique to investigate the length of internal chain segments. For the production of β-CLDs, 100 µg glucose equivalents of starch or WSP samples were diluted with water to a final sample volume of 100 µL to 500 µL, depending on the sample volume required for 100 µg glucose equivalents. Samples were boiled for 10 min and 2/5 of the sample volume of 200 mM MES; pH 6.5 buffer with 3 mM DTT and 100 units of barley β-amylase (Megazyme, Wicklow, Ireland) were added to the resuspended starch and WSP samples to shorten external chains. The mixture was incubated at 37°C for 3 h and then boiled for 5 min. The β-limit-dextrin was precipitated 16 h at -20°C in 75% (v/v) methanol and collected by centrifugation for 10 min, at 3`000 g, 4°C. The supernatant was discarded and the pellet washed twice with cold 75% (v/v) methanol without disturbing the pellet. The pellet was redissolved by boiling for 10 min in 110 µL of water. An additional centrifugation for 10 min, at 3`000 g, 4°C, was performed to remove most of the denaturated, aggregated proteins. The supernatant (100 µL) was transferred to a new microcentrifuge tube and the CLD digestion with isoamylase and pullulanase was performed as described above.

2.8 High pH Anion Exchange Chromatography Coupled to Pulsed Amperometric Detection

Neutral glucan chains were separated by HPAEC using a BioLC from Dionex (Olten Switzerland). Ten µL of the samples were injected into the system and separated using a CarboPac PA200, 3 × 250 mm column (Dionex, Olten, Switzerland). Long glucan chains from CLD and β-CLD preparations as well as 2. Material and Methods 41 ______soluble sugars were separated using the following eluants: eluent A, 100 mM NaOH; eluent B, 150 mM NaOH and 500 mM sodium acetate.

To separate glucan chains from CLD and β-CLD analyses, the following program with a flow rate of 0.5mL min -1 was used: starting conditions were 95% eluant A and 5% eluant B. The amount B was increased in a linear gradient up to minute 12.5, when the concentration of A was 60% and the concentration of B was 40%. From minute 12.5, the amount of B was further increased with a linear gradient until at minute 50, when 15% A and 85% B was reached. From minute 50 to 70, the column was re-equilibrated in starting conditions.

To separate malto-oligosaccharides including maltose and shorter glucan chains from the P- oligosaccharide digests (see Section 2.12) but also sucrose, glucose and fructose, the following conditions were used: 0 to 7 min, 100% A; 7 to 19.5 min, a linear gradient to 60% A, 40% B ; 19.5 to 35 min, gradient to 45% A and 55% B; 35 to 44 min, gradient to 15% A and 85% B; 45 to 60 min, re- equilibration of the column in 100% A.

For the CLD and β-CLD analyses as well as the P-oligosaccharide digests, peaks were identified by coelution with known linear glucan standards. Peak areas were determined using the Chromeleon software (Dionex, Olten, Switzerland) and areas for specific chain lengths were normalized to the total area of all peaks which have been identified (CLD and β-CLD) or shown as raw chromatograms (P- oligosaccharide digests). To account for unequal losses between samples during the Dowex steps and the lyophilisation, samples for the quantification of maltose and soluble sugars were spiked with cellobiose as internal standard and peak areas were corrected accordingly. Different concentrations of sugar standards were run in parallel so that the amount of maltose, sucrose, fructose and glucose could be quantified.

2.9 Plant Transformation and Complementation

Coding regions of AtBAM1 , AtBAM2 , AtBAM3 and AtLSF1 without their stop codons were cloned from the available pDONR plasmids into the pC-TAPa vector (Rubio et al., 2005) in a reaction mediated by Gateway LR Clonase Enzyme Mix (Invitrogen, Basel, Switzerland), according to the manufacturer’s instructions. The pC-TAPa vector contains a C-terminal IgG binding domain, a myc epitope as well as a 6x HIS-tag and a protease cleavage site between the IgG binding domain and the myc epitope. Ectopic expression of the transgene is driven by the cauliflower mosaic virus 35S promoter. The plasmids were transformed into Agrobacterium tumefaciens (strain GV3101) by electroporation and transformants selected by growth in selective medium with 100 µg/mL spectinomycin. Transformation of Arabidopsis was done by floral dip, according to Clough and Bent (1998). Transformed seedlings were selected on 2. Material and Methods 42 ______

MS-plates (1x Murashige and Skoog mineral salts (Serva, Heidelberg, Germany), 1% (w/v) sucrose, 0.05% (w/v) MES, pH5.7, 0.85% (w/v) agar) containing gentamycin at a concentration of 15-35 µg/mL. For complementation analysis and tandem affinity purification, several lines from independent transformation events were grown on selective plates, alongside with the respective wild type and mutant controls on non-selective MS plates in Perceival growth chambers. Gentamycin resistant seedlings and the relevant controls were transplanted to soil and grown for at least 2 more weeks before harvesting for metabolite measurements (see Section 2.7).

2.10 Tandem Affinity Purification and Myc-CoIP

Tandem affinity purification was performed according to Rubio et al., (2005) with minor modifications.

Up to 10g of plant material harvested at the end of the night were ground under liquid N 2 in mortars. Protein extracts were made from the leaf powder homogenized in cooled glass homogenizers in ice-cold extraction medium containing 100mM Tris-Cl, pH 7.5, 10% (v/v) glycerol, 150mM NaCl, 0.1% (v/v) Triton X-100, with 1x Roche Life Sciences Protease Inhibitor, at a ratio of 0.5:1 (w/v). Homogenates were filtered through 2 layers of miracloth and centrifuged at 13’000 g , 4°C for 15min. Supernatants were incubated with 300uL IgG beads (GE Healthcare) per 5g starting material. Extracts were incubated on a vertical rotary wheel for 4 h at 4°C. After incubation, beads were washed three times with 10mL extraction medium and once in 10mL extraction medium supplemented with 1mM DTT. Cleavage from the IgG beads was performed by incubation with 25units of 3C Prescission protease (GE Healthcare) in 5 mL extraction medium with 1mM DTT, at 4°C on a vertical rotary wheel for 1 h. After incubation, supernatants were recovered after centrifugation. Beads were washed with 5 mL extraction medium and the supernatant recovered again by centrifugation. The supernatant from the digest and the supernatant from the wash were pooled. Pooled supernatants were incubated for 2 h on a vertical rotary wheel at 4°C with 1 mL of Ni-Sepharose 6 Fast Flow beads (GE Healthcare). Ni-Sepahrose beads were washed three times with 10mL extraction medium and proteins were eluted in 5mL extraction medium containing 150mM imidazole. Eluates were concentrated approximately 40-fold using Amicon® Ultra-4 10 kDa Centrifugal Filter Devices (Millipore). This tandem affinity purification procedure was repeated at least twice for each tagged protein, each time using an independent batch of starting material. As negative control, wild-type extracts were processed through all stages of the tandem affinity purification procedure.

The tandem affinity purification procedure makes use of the IgG binding domain and the HIS-tag for the purification of tagged proteins. The pC-TAPa vector in addition offers a myc-tag which can be used for 2. Material and Methods 43 ______affinity purification of tagged proteins. A one-step purification of the tagged proteins using Anti-c-Myc Agarose (Sigma-Aldrich) was performed to corroborate results obtained from the tandem affinity purification by an independent affinity purification. I have called this one-step purification myc-CoIP in my results. The myc-CoIP was conducted by extracting leaf material at a 1:1 (w/v) ratio in ice-cold extraction medium using cooled glass homogenizers. Insoluble material was removed by centrifugation (20’000 g , 5min at 4°C). Four hundred µL of the supernatant were incubated with 50µL Anti-c-Myc Agarose (Sigma-Aldrich) for 3h on a vertical rotary wheel at 4°C. After incubation, agarose beads were washed four times with 500µL extraction medium. Proteins bound to the beads after the washes were eluted by boiling the beads for 5 min in SDS loading buffer (see Section 2.2).

2.11 Mass-Spectrometric Analyses of Tandem-Affinity-Purified Protein Samples

Protein samples from the tandem affinity purification and the myc-CoIP were separated on standard SDS- PAGE gels containing 10% (w/v) acrylamide. Gels were silver stained according to Shevchenko et al. (2002). In brief, gels were fixed in 50% (v/v) methanol, 12% (v/v) acetic acid, washed in 50% (v/v) ethanol and, pretreated for 1 min with 0.02% (w/v) Na 2S2O3. Gels were then incubated for 20 min in 0.2%

(w/v) AgNO 3 and washed three times during 20 sec in water before they were developed in 6% Na 2CO 3,

0.0005% (w/v) Na 2S2O3 and 0.05% (v/v) formaldehyde. Bands were excised and in-gel tryptic digests were performed. Gel pieces were destained in 1% (v/v) H 2O2 until the silver stain was gone, washed in 50% (v/v) acetonitrile (ACN) and dried in a speed vac. Disulphide bridges were reduced with 10mM DTT in 25mM ammonium bicarbonate, pH8 for 45min at 56°C. Cysteines were alkylated with 50mM iodoacetamide in 25mM ammonium bicarbonate, pH8, for 1h at room temperature in the dark. Gel pieces were washed and dehydrated in 50% (v/v) ACN and dried in a speed vac. For the tryptic digests, enough 25mM ammonium bicarbonate, pH8, with 2.5µg mL -1 Trypsin (Sequencing Grade, Promega AG, Wallisellen, Switzerland) was added to completely cover the gel pieces. Digestion was performed 16 h, at 37°C. Peptides were eluted from the gel pieces in 50% (v/v) ACN, 5% (v/v) trifluoroacetic acid (TFA). The eluted peptides were dried in a speed vac and resuspended in 3% (v/v) ACN, 0.1% (v/v) TFA. Peptides were purified using C18 ZipTips TM (Millipore) according to manufacturer’s protocols and dried in a speed vac. Purified peptides were resuspended in 5% ACN, 0.1% formic acid (FA). Peptide samples were analysed using an LTQ-Orbitrap mass spectrometer (ThermoFischer Scientific, Reinach, Switzerland) interfaced with a nanoelectrospray ion source. Peptides were separated using an Eksigent nano LC system (Eksigent Technologies, Dublin, Ireland), equipped with an 11-cm fused silica emitter 2. Material and Methods 44 ______

(75 mm i.d.; BGB Analytik, Böckten, Switzerland), packed in-house with a C18 resin (Michrom BioResources, Auburn, USA). Peptides were loaded from a Spark Holland autosampler and separated using an ACN/water solvent system containing 0.1% FA with a flow rate of 200 nL min -1. Peptide mixtures were separated by gradient elution from 3% to 35% ACN in 75min. Tandem mass spectra were processed by Xcalibur version 2.0.7 (Thermo Fisher Scientific). Measured spectra were analyzed using Mascot (Matrix Science, London, UK). Mascot was set up to search an Arabidopsis database containing known contaminants. Mascot searched with a fragment ion mass tolerance of 0.80 Da and a parent ion tolerance of 10.0 PPM. The iodoacetamide derivative of cysteine was specified in Mascot as a fixed modification. Oxidation of methionine was specified in Mascot as a variable modification. Scaffold (Proteome Software Inc., Portland, USA) was used to validate MS/MS based peptide and protein identifications. Peptide identifications were accepted if they could be established at greater than 80.0% probability as specified by the Peptide Prophet algorithm (Keller et al., 2002). Protein identifications were accepted if they could be established at greater than 95.0% probability and contained at least 2 identified peptides. Protein probabilities were assigned by the Protein Prophet algorithm (Nesvizhskii et al., 2003). With these settings, the rate of false positives was below 1% as determined by searching the Arabidopsis decoy database.

2.12 Purification and Digestion of Phospho-Oligosaccharides

Phosphorylated glucan chains (P-Oligos) were purified by Dr. O. Kötting and are available in the laboratory. In brief, 10g amylopectin from potato starch (Sigma-Aldrich) were dissolved in 200 mL 5mM sodium acetate, pH 4.8, and incubated with 25’000 U isoamylase (Sigma-Aldrich), 3.6 U pullulanase (Sigma Aldrich) and 1’000 U barley β-amylase (Megazyme) for 19h at 37°C. The reaction was boiled for 15 min to stop the reaction. Insoluble material was removed by centrifugation for 30min at 5’000 g, 4°C.

The supernatant was filtered through a 45µm filter (Sartorius Stedim Biotech), diluted to 1L with H 2O and the pH was adjusted to 7 with NaOH. P-Oligos were separated from neutral glucan chains by anion- exchange chromatography using a custom-made Q Sepharose FF (GE Healthcare) column, operated by an ÄKTAExplorer100 (GE Healthcare). The sample was applied to the column at a flow rate of 2mL min -1, the column was then washed with 2 column volumes of H 2O and P-oligos were eluted with 3 column volumes of 200mM NaCl, 10mM HCl. Purified P-oligos were precipitated in 75% (v/v) ethanol for 30min on ice. Precipitates were pelleted by centrifugation for 30min at 12’000 g, 4°C and supernatants were removed. The precipitated and purified P-oligos were redissolved in 2mM HEPES-KOH, pH7. Approximately 0.07µg mL -1 glucose equivalents of P-oligos were used for the digestion with BAM1-TAP and heterologously expressed BAM1 (BAM1-Het). BAM1-Het was expressed in E. coli and purified by 2. Material and Methods 45 ______

Dr. H. Reinhold as described previously (Fulton et al., 2008). SEX4 was expressed in E. coli and purified by Dr. O. Kötting as described previously (Kötting et al., 2009). BAM1-TAP refers to the tandem-affinity purified BAM1-TAP fusion protein which was purified according to the procedures described in Section 2.10. P-oligos and BAM1-TAP or BAM1-Het without or with SEX4 as dephosphorylation control, were incubated in 100mM sodium acetate, 50mM Bis-Tris, 50mM Tris pH 6, 1mM DTT, 1mM MgCl 2, 1mM

CaCl 2, 1mg/mL BSA for 3 h at 37°C. Reactions were stopped by boiling for 10 min and oligosaccharides which had not been dephosphorylated were removed by ion-exchange chromatography using the Dowex resins as described in Section 2.7.2. Maltose and dephosphorylated glucan chains were analyzed using HPAEC-PAD, as described in Section 2.8.

2.13 Digestion of Amylopectin Using BAM1-TAP( bam1 ) and BAM1- TAP( lsf1 )

BAM1-TAP fusion proteins expressed in the Atbam1 and Atlsf1 mutant backgrounds were purified as described in section 2.10. Relative BAM1-TAP levels in the purifications were compared by western blotting and immunodetection of the tagged BAM1, using a monoclonal Anti-c-Myc antibody according to manufacturer’s instructions (Sigma-Aldrich). BAM1-TAP purifications were incubated at 37°C in -1 100mM Tris-Cl, pH 7.5, 1mM MgCl 2, 1mM CaCl 2, 5mM DTT with 0.2µg µL solubilized amylopectin. After 30 min, 60 min and 120 min, aliquots were removed from the reactions and reactions stopped by adding 10 volumes of 20% (w/v) unbuffered Tris. For the 0 min time point, the enzyme-buffer solution was mixed with the amylopectin in the 20% (w/v) unbuffered Tris stopping solution. Samples were spiked with cellobiose, applied to sequential Dowex columns (Section 2.7.2) and maltose was quantified using HPAEC-PAD (Section 2.8) by comparison with the relevant maltose standards. Maltose release was corrected for the relative BAM1-TAP protein input into the assay and expressed as µM maltose based on equal BAM1-TAP levels (Arbitrary Protein Unit).

2.14 Measurement of β-Amylase Activity

For total soluble β-amylase activity measurements, soluble proteins were extracted from leaves as described in Section 2.3. Total protein concentrations were measured using the Bradford-Kit from BioRad, according to manufacturer’s instructions. Total β-amylase activity was measured using the Betamyl assay kit (Megazyme). Betamyl substrate contains PNPG5 ( p-nitrophenyl-α-D-maltopentaose) and α-glucosidase. After incubation for 10, 20 and 45 min at 30°C, 3 volumes of unbuffered 1% (w/v) 2. Material and Methods 46 ______

Tris were added to stop the reaction. The time zero control was made by adding the unbuffered Tris to the protein extracts before the betamyl substrate. The activity was defined as the difference in absorption at 410 nm between the control and the assay, corresponding to the amount of p-nitrophenol released from PNPG5 due to the sequential actions of β-amylase present in the plant extract and α-glucosidase (the latter is present in excess). The assay was linear over the time points and protein concentrations measured.

2.15 Transmission Electron Microscopy of C. peltata leaves and Müllerian Bodies

Leaf samples were harvested and cut into rectangles (2mm x 1mm). MBs were cut breadthways in half. Specimens were fixed in 2% (v/v) glutaraldehyde in 0.1 M sodium cacodylate buffer, pH 7.4, 16 h at 4 °C. Fixed samples were washed 5 times with ice-cold 0.1 M sodium cacodylate buffer, pH 7.4, postfixed for 16 h in 1% (w/v) aqueous osmium tetroxide, 0.1 M sodium cacodylate buffer, pH 7.4, at 4 °C, washed again 3 times in ice-cold 0.1 M sodium cacodylate buffer, pH 7.4 and once in water at 25°C. Samples were then dehydrated in increasing ethanol series (50%, 60%, 70%, 80%, 98% [v/v]), at 4°C for 2h and finally in 100% (v/v) ethanol at 4°C for 16 h. Dehydrated samples were embedded in epoxy resin (Spurr’s, Agar Scientific, Stansted, England). Increasing ratios of Spurr resin with ethanol were used for embedding (25%, 50%, 75% [v/v], each step 16 h at 25°C and 100% [v/v], 7 days at 25°C). Ultrathin sections were cut with a diamond knife, collected on copper grids and stained with uranyl acetate (2% (w/v) in 50% (v/v) acetone), washed three times with water and stained again in Reynold’s lead citrate. Stained sections were washed three times with water and examined using a Phillips BioTwin CM100 or a FEI Morgagni 268 transmission electron microscope.

2.16 mRNAseq of C. peltata Leaves and Müllerian Bodies

2.16.1 C. peltata RNA Extraction, cDNA Synthesis and Normalization

For the isolation of RNA used for the 454 sequencing, MBs of a maxiumum age of 1 day were harvested between 3 p.m. and 4 p.m. and pooled from different active trichilia. Leaf material was sampled at the same time, from a fully emerged leaf with an active trichilium. Plant material was frozen immediately in liquid N 2. For the isolation of RNA used for the Illumina RNA-seq, newly emerging MBs were harvested in a time period from 12 p.m. to 4 p.m., frozen in liquid N 2 and material from the same trichilium was pooled. Leaf material was harvested at 3p.m. from leaves with an active trichilium. Biological replicates from three active trichilia and the three corresponding leaves were harvested and processed further. Using 2. Material and Methods 47 ______glass beads and a Mixer Mill (Retsch, Haan, Germany), frozen leaf and MB material was ground to a fine powder . RNA was isolated from a maximum of 50 mg leaf and 30 mg MB powder using the Spectrum™ Plant Total RNA Kit (Sigma-Aldrich) according to manufacturer’s instructions. For the Illumina sequencing, RNA was shipped on dry ice to Fasteris SA (Geneva, Switzerland), where it was processed further. For the 454 pyrosequencing, cDNA was prepared using the Mint cDNA Synthesis Kit (Evrogen, Moscow, Russia). To modify polyA-tails (Beldade et al., 2006), modified primers were used for 1 st and 2nd strand cDNA synthesis (See Table 2-2). The 3’-Primer from the Mint cDNA Synthesis Kit was replaced by the polyTdeg primer to disrupt the polyA tail and the M1-primer was replaced by M1ACGG and polyTM1 to further modify what used to be the polyA-tail. All other conditions for double-stranded (ds) cDNA synthesis were according to manufacturer’s instructions, with 24 cycles being optimal for amplification of 1 st strand cDNA. The resulting amplified cDNA was purified using the QIAquick PCR Purification Kit (Qiagen, Basel, Switzerland). Purified cDNA was quantified using a NanoDrop ND-1000 Spectrophotometer (NanoDrop Technologies,Wilmington, USA). Nine hundred ng cDNA made from RNA from each tissue were precipitated by adding 0.1 volume 3M sodium acetate, pH4.8, and 2.5 volumes 98% (v/v) ethanol and centrifugation at 16’000 g. Pellets were washed in 80% (v/v) ethanol, dried and resuspended in water to a concentration of 150ng µL -1 for subsequent cDNA normalization.

Normalization was done separately for cDNA synthesized from MB RNA and from leaf RNA. The Trimmer cDNA Normalization Kit (Evrogen) was used according to manufacturer’s instructions. As the activity of the duplex specific nuclease (DSN) varies from batch to batch, optimal conditions for normalization had to be established. 1/6 of the standard DSN amount was sufficient for normalization of MBs and leaf ds cDNA as checked by standard EtBr agarose gels. Normalization reduces the abundance of highly prominent cDNAs. The high abundant cDNAs can be seen as discrete DNA bands on agarose gels. Successful normalization reduces the abundance of these bands. On agarose gels this results in a smear with similar average size as the discrete bands. For the final amplification of normalized cDNA, 13 cycles were optimal. Normalized cDNA was quantified by fluorometry using the Quant-iT DNA assay kit (Invitrogen) at the Functional Genomics Center Zürich (FGCZ, Zürich, Switzerland), where also library preparation and sequencing were done.

Table 2-2: Modified Primers for 1 st and 2 nd Strand cDNA Synthesis

Mint cDNA Synthesis Modified Primer Sequence Primer Name Name

3’-Primer polyTdeg 5‘-AAGCAGTGGTATCAACGCAGAGTAC (T)4G(T) 9C(T) 10 VN-3‘ M1ACGG 5‘-AAGCAGTGGTATCAACGCAGAGTACGG-3‘ M1 Primer polyTM1 5‘-AAGCAGTGGTATCAACGCAGAGTACTTTTGTCTTTTGTTCTGTTTCTTTTVN -3‘ Long homopolymeric stretches of nucleotides can be detrimental in 454 pyrosequencing. PolyA-tails in mRNA were modified using degenerate primers for cDNA synthesis with the Mint cDNA S ynthesis Kit.

2. Material and Methods 48 ______

2.16.2 Library preparation for Illumina and 454 Pyrosequencing

Library preparation for the Illumina sequencing was performed at Fasteris SA. In brief, the following steps were required for library preparation. PolyA transcript purification, poly-A transcript fractionation in the presence of Zn +, ds cDNA synthesis using random primers, RNase H treatment, end repair, and ligation of adapters were performed at FASTERIS SA but basically done as described in more detail previoously (Nagalakshmi et al., 2010). All library preparations were done separately for the three leaf and three MB RNA samples. Index codes (6bp) were incorporated into the ds cDNA libraries to attribute reads to specific samples (Leaf 1: ATCACG, Leaf 2: CGATGT, Leaf 3: TTAGGC, MB 1: TGACCA, MB 2: ACAGTG, MB 3: GCCAAT ). The ds cDNA fragments were gel purified to isolate specified insert sizes. Libraries were pooled and PCR amplification was used to generate the DNA Colonies Template Library. This, library was purified and diluted to 10nM. The Flow-cell was prepared on the Cluster Station as described by Illumina, Inc (San Diego, USA). High-throughput DNA sequencing was performed on the Illumina Genome Analyzer GAIIx. Seventy-six + 7(index) sequencing cycles were used. The Sequencing kit was the Chrysalis 36cycles v 4.0 (Illumina Inc).

For the 454 sequencing, library preparation was performed by Dr Marzanna Küenzli at the FGCZ, on the normalized cDNA. The sequencing libraries were prepared with the GS Titanium General Library Preparation Kit (Roche Life Sciences) using 5 µg of cDNA according to the manufacturer's protocol. The sequencing reactions were performed using a Roche Life Sciences 454 Genome Sequencer FLX with the GS Titanium Sequencing Kit XLR70 (Roche Life Sciences) on two big regions of the GS Titanium PicoTiterPlate Kit (70x75) (Roche Life Sciences), according to the manufacturer's instructions. Image and signal processing were done using the GS FLX SW v2.0.01.12, gs Run Processor (Roche Life Sciences) with full processing.

2.16.3 Assembly of 454 Contigs and Illumina Reads

Reads obtained from the Illumina sequencing are attributed to each sample according to their index code. Reads were first attributed to samples based on perfect matches of the index code and then with up to 2 mismatches allowed, unless a read could be attributed to 2 samples. Reads which could not be assigned to a sample were not used further. 454 reads were trimmed and adapter sequences were removed with a customized pipeline developed by Dr W. Qi at the FGCZ. The pipeline uses multiple software packages, the Newbler software (Roche Life Sciences), megablast (Zhang et al., 2000), seqclean (http://compbio.dfci.harvard.edu/tgi/software/) as well as perl scripts developed at the FGCZ. 454 contigs were pre-assembled using the Newbler software (Roche Life Sciences), with default parameters. 2. Material and Methods 49 ______

454 pre-assembled contigs were assembled together with the short Illumina reads at Fasteris SA using Velvet and Oases software packages (Zerbino and Birney, 2008; Young et al., 2010, http://www.ebi.ac.uk/~zerbino/oases/) with default parameters. VELVET builds a hash table of all possible “kmer” (sequence of “k” bases) in the dataset and, through de Bruijn graph construction and repeat resolution, builds de novo contigs. OASES is a tool processing the VELVET output in order to assemble a transcriptome. OASES is designed to deal with sequences not uniformly covered by reads. A minimum size of 100 was required to retain a transcript. To optimize the Velvet/Oases assembly process, to validate the de novo assemblies and to estimate the number of reads assembled, Illumina reads were mapped to the Velvet/Oases assembly. A maximum number of Illumina reads could be mapped to an assembly which was generated with the Illumina reads cut at 70 bases and then assembled with the 454 pre-assembled contigs using Vevlet/Oases. Choosing a hash value of 63 in the Velvet algorithm proved to result in an assembly to which the highest percentage of Illumina reads mapped. In the Velvet input, the 454 contigs were used as long templates (-long -fasta option).

2.16.4 Short Read Mapping and Transcript Annotation

The quality controlled and trimmed Illumina reads were mapped to the transcripts from the optimized Velvet/Oases assembly using the CLCGenomics Workbench (CLC bio, Aarhus, Denmark) with default parameters. The reads for each biological replicate were separately mapped to the transcripts. Transcripts were imported into Blast2Go (version 2.3.6; http://www.blast2go.org/). Blast2Go is an automated tool for the retrieval of BLAST hits and the assignment of gene ontology terms (GO-terms). It was designed for use with novel sequence data (Conesa et al., 2005; Götz et al., 2008). Transcripts were annotated by running local BLASTx searches against the NCBI nr-Database (All non-redundant GenBank CDS translations+PDB+SwissProt+PIR+PRF) as well as against the TAIR9 Arabidopsis thaliana (Swarbreck et al., 2008) annotated protein database. An e-value threshold of 10 -5 was used. Assignment of GO-terms to transcripts with the BLASTx matches from the NCBI nr-Database was performed by Blast2go with default parameters. For gene by gene analyses of transcripts encoding proteins involved in starch metabolism, BLAST hits against the TAIR9 database were inspected manually and the best BLAST hits were exported into excel-files.

2. Material and Methods 50 ______

2.16.5 Read Count Normalization, Identification of Differentially Expressed Transcripts and GO Enrichment Analysis

Read counts for each transcript were normalized according to Robinson and Oshlack, (2010). The procedure uses the raw data to estimate appropriate scaling factors. The normalization is part of the Bioconductor package ‘edgeR’ (http://www.bioconductor.org/help/bioc- views/release/bioc/html/edgeR.html (Gentleman et al., 2004; Robinson et al., 2010). Differentially expressed transcripts were identified by ‘edgeR’, based on a log2 fold change (FC) cutoff of 2 resp -2, a p-value smaller than 0.001 and a minimum of combined total 30 reads per transcript. Lists of transcripts uniquely expressed in leaves or MBs or transcripts over-expressed in one or the other tissue were exported and combined with the BLAST hits in excel. Statistical assessment of GO-term enrichments among the uniquely expressed or over-expressed transcripts was performed by comparing them to the total annotated transcripts as reference group. This functionality was introduced in Blast2Go by the Gossip package (Blüthgen et al., 2005). Gossip computes a ‘Fisher's Exact Test’ applying robust false discovery rate (FDR) correction for multiple testing and returns a list of significant over- or under- represented GO-terms. One-sided ‘Fisher's Exact Tests’ were performed, and only GO-terms significantly over-represented in the different transcript lists were returned, based on a cut-off value of 5% FDR. Lists from the GO enrichment were consolidated and visualized using the REVIGO web server (http://revigo.irb.hr/, (Supek et al., 2010)). A similarity of 0.7 was allowed for the grouping of GO-terms, and the numbers of transcripts in each GO-term from the ‘Fisher's Exact Test’ made in Gossip were used to assign circle sizes in semantic-similarity based scatter plots of the GO-terms. 3. Protein-Protein Interactions between Proteins Involved in Starch Degradation 51 ______

3. Protein-Protein Interactions between Proteins Involved in Starch Degradation

3.1 Introduction

Starch synthesis is a process that requires the closely coordinated activity of a number of enzymes and it is therefore not surprising that protein-protein interactions between the enzymes involved in starch synthesis are common (Tetlow et al., 2004b; Hennen-Bierwagen et al., 2008a; Tetlow et al., 2008; Hennen-Bierwagen et al., 2009)). No protein-protein interactions have been described so far for any of the enzymes known to be involved in starch degradation (Kötting et al., 2010; Zeeman et al., 2010). The current models for starch degradation value the hydrolytic degradation of the glucose chains by BAMs to be one of the major pathways of starch degradation. Even though β-amylases are crucial for starch degradation, they require other enzyme activities for the efficient conversion of amylopectin to maltose. α-1,6 linkages cannot be hydrolyzed by β-amylases and need to be removed by debranching enzymes (Wattebled et al., 2005; Delatte et al., 2006). Likewise, the phosphorylation on C3 and C6 residues of glucose units which is initially required to loosen the semi-crystalline granule structure represents an obstruction to β-amylase activity, as these enzymes cannot work past phosphorylated glucose residues (Takeda and Hizukuri, 1981). The removal of phosphate from glucan chains is therefore essential, reflected by the sex phenotype of the mutant lacking the phospho-glucan phosphatase SEX4 (Kötting et al., 2009). The aforementioned enzymes are interdependent and their activities are increased by preceding actions of other enzymes. Phosphorylation of glucan chains by GWD for instance, is increased during starch degradation and enhanced by parallel β-amylolysis in vitro (Ritte et al., 2004; Edner et al., 2007). Inversely, phosphorylation and dephosphorylation of starch granules increase the rate of β-amylolytic degradation of amylopectin (Edner et al., 2007; Kötting et al., 2009). Even debranching enzymes, which enable efficient β-amylase activity and are required for the complete degradation of amylopectin preferentially work on substrates previously digested by β-amylases. Given the high degree of interdependence of starch degrading enzymes, complex formation between different enzymes involved in starch degradation might be a way to coordinate and control starch degradation (Edner et al., 2007; Kötting et al., 2009).

It is intriguing that even though starch degradation is among the central processes in a plant leaf during the night very little information is available on the precise mechanisms controlling onset or rate of starch degradation. Earlier it was suggested that starch degradation might not be regulated at all and that starch accumulation solely was controlled by changes in the rate of starch synthesis (Preiss, 1982). However, 3. Protein-Protein Interactions between Proteins Involved in Starch Degradation 52 ______several lines of evidence argue in favour of more direct control of starch degradation (Zeeman et al., 2004a). Among them is the fact that during starch synthesis, very little breakdown is detectable (Kruger and Ap Rees, 1983; Li et al., 1992) even though the enzymes involved in starch degradation are present. Leaf starch degradation at night is tightly regulated in Arabidopsis so that almost all starch is degraded at the end of the night. This regulation is at least partially controlled by the circadian clock, as shown by the altered starch degradation dynamics of mutant lines lacking clock components (Graf et al., 2010). The rate of starch degradation can be adjusted immediately. When plants, generally grown in a 12-h day/12-h night regime were subjected to an early night, the rate of starch degradation was reduced to make the carbon storage reserves last through the night (Graf et al., 2010). The flexibility of its starch metabolism allows Arabidopsis to grow in extreme short-day conditions. Growth is possible, albeit at a reduced rate, in photoperiods down to 2h (Gibon et al., 2009). Interestingly, in order for the carbon reserves to last through the night in Arabidopsis , they need to be present as granular starch. This is suggested by the fact that the WSP produced in the isa1 and isa2 DBE mutants of Arabidopsis is degraded before the onset of the light-period (Delatte et al., 2005). This observation makes the glucan-water dikinases which initiate starch degradation by loosening the tight packing of glucose chains through phosphorylation a likely site of regulation for the starch degradation pathway. However, this does not exclude the possibility that amylolytic enzymes (e.g.: β-amylases), are also subject to regulation. The lack of information on the regulatory mechanisms of starch degradation and the between enzymes involved in degrading the starch granule led us to search for protein-protein interactions which could potentially orchestrate and control the activities of single enzymes in the starch degradation pathway.

3.2 Results

3.2.1 Gel Filtration Chromatography – Column Calibration

Gel filtration chromatography (GFC) was used to separate proteins of crude leaf extracts according to their native molecular weights (NMW). GFC is buffer flow driven and the mobility of a protein or protein-complex is independent of its electric charge. Using Sephacryl 300HR™ (GE Healthcare) as column medium and a buffer containing 150mM KCl, substrate binding effects should be minimized but cannot be completely excluded. By running a set of known molecular weight standards, the self-made Sephacryl 300HR™ column was calibrated so that the fraction or the elution volume in which a protein or protein-complex eluted could be assigned to a molecular weight range (Fig2-1 A and B). As small molecules tend to be retarded in the pores of the column media, large protein complexes elute in earlier fractions than small molecules. The elution of standards, as well as of the actual Arabidopsis proteins is distributed over several fractions collected by a fraction collector. An estimation of molecular weight 3. Protein-Protein Interactions between Proteins Involved in Starch Degradation 53 ______using GFC can at best be based on the molecular weight assigned to the fractions or the elution volume in which a maximum of the protein in question elutes. GFC column calibration assumes that proteins and protein complexes are approximately globular, thus shape effects could potentially cause the elution profile to deviate from the genuine NMW of proteins and protein complexes. Despite these limitations, GFC offers a relatively rapid estimation of NMW for proteins which can subsequently be detected by western blotting or activity assays.

The main focus of work described here was to identify protein-protein complexes which could be involved in starch degradation. Therefore, the proteins extracted from wild-type leaves harvested at the end of the night were separated by GFC and specific antibodies, available against a number of proteins involved in starch degradation, were used to probe the elution fractions. The experiments presented below were conducted on two separate protein extracts, which were separated over the GFC column at a time interval of 4 months. The extraction conditions for the proteins were identical; the column was calibrated prior to each run. The fraction collection was slightly different between the two runs. The performance of the column did not change significantly in the time between the experiments so only one out of two calibration experiments is shown in Fig2-1 (A and B). Both runs were calibrated using the same set of standards and therefore are directly comparable to each other by the peak elutions of the standards.

Figure 2-1. Calibration of an XK16 Sephacryl 300HR™ Gel Filtration Chromatography Column (A) Known size standards were run in two combinations, according to manufacturer’s instructions, to calibrate a self-made Sephacryl 300HR™ column operated by an ÄKTA Explorer100 controlled by the Unicorn software. Combination 1 contained a mix of the thyroglobulin, aldolase and ovalbumin proteins (red trace). Combination 2 contained a mix of ferritin and conalbumin proteins (green trace). The absorbance at 280nm was monitored (mAU: milli-Absorbance Units). (B) The relation between the elution volume of a given standard protein and its molecular weight was calculated and is used in subsequent experiments to determine the native molecular weights of proteins involved in starch degradation.

3. Protein-Protein Interactions between Proteins Involved in Starch Degradation 54 ______

3.2.2 Gel Filtration Chromatography of Proteins Involved in Starch Degradation

An order approximately following the assumed series of events during starch degradation is used to present the elution profiles of the different proteins analyzed. The glucan kinases, GWD and PWD (Fig2- 2 E and F respectively) both have comparably large observed monomeric weights on the SDS-PAGE (approx 130kDa and 100kDa respectively). While the vast majority of the PWD and GWD proteins elutes in fractions corresponding to approximately their monomeric weights, a significant portion of the GWD protein is present in fractions larger than 158kDa. It may thus be possible, that GWD associates in high molecular weight (HMW) complexes judging from these GFC experiments. The phospho-glucan phosphatase SEX4 is a comparably smaller protein of 43kDa predicted mass. Due to the narrow choice of fractions, the western blot in Fig2-2 A fails to identify the peak elution of SEX4 but it seems clear from the increased protein elution after the 75kDa marker that this peak might match with the predicted monomeric weight of SEX4. It is however remarkable, that the SEX4 protein is present in a wide range of molecular weights, extending up to above 448kDa. This finding shows the potential for SEX4 to be present in different HMW complexes. Due to the insufficient resolution of the GFC it is however not possible to separate and further characterize such different complexes according to their size. The debranching enzymes ISA3 and LDA (Fig2-2 C and J) are both not detectable in fractions significantly larger than their monomeric weights (approximately 78kDa for ISA3 and 100kDa for LDA). As no antibody was available for LDA, the enzyme activity was detected using a coloured substrate (red- pullulan, α(1,4) maltotriose units α(1,6) linked with each other, with a red stain approximately every 30 glucose units), specific to LDA-activity, which results in clear bands on an activity gel. At least four β- amylases are localized to the Arabidopsis chloroplast: BAM1, BAM2, BAM3 and BAM4 (Fulton et al., 2008). Of these, BAM2 and BAM3 were shown to be mainly present in the low molecular weight fractions (Fig2-2 I and D).

For AMY3, the only known chloroplast localized α-amylase (Yu et al., 2005), the majority of the protein seems to be present in fractions larger than 158kDa, which is significantly above the monomeric weight of AMY3 which is around 100kDa (Fig2-2 B). An interesting elution profile could be observed for DPE2, the cytosolic disproportionating enzyme (Fig2-2 H). Due to a narrow selection of fractions analyzed, its entire elution profile could not be observed. However, it is clear that DPE2 elutes with a NMW of above 450kDa. Thus DPE2 is present in one or several large HMW complexes. Two proteins whose functions in starch metabolism are unclear so far could also be detected in HMW complexes. One of them was a protein originally identified as a starch binding protein from pea seeds and had been named Esterhaze (ESH; Dr S. C. Zeeman, unpublished results). The gene coding for the ESH protein showed a diurnal expression pattern, peaking towards the end of the day, similar to a number of genes known to be 3. Protein-Protein Interactions between Proteins Involved in Starch Degradation 55 ______involved in starch degradation (e.g: GWD, PWD, ISA3, AMY3, DPE1, DPE2; (Smith et al., 2004). It has therefore been studied as a potential new element involved in starch degradation. The Arabidopsis protein is present in a double band at around 60kDa and can be detected using a specific antibody (Fig2-2 H). Interestingly, the elution profile of ESH revealed its presence almost exclusively in HMW fractions around and above 448kDa.

Next to BAM2 and BAM3, another catalytically active β-amylase, BAM1, has not been mentioned yet. BAM1 was largely present in HMW fractions of 100-150kDa (Chapter 4, Fig 4-4A) and the protein complex including this β-amylase will be discussed in detail in the next chapter. Likewise, another protein which has so far not been attributed a clear function in starch metabolism is LSF1. The SEX4 homologue could be shown to be present in a broad range of native molecular weights up to more than 158kDa in size. Like for BAM1, a description of the interaction partners and potential functions of LSF1 in starch metabolism will be given in the next chapter (Chapter 4, Fig 4-4B).

3.3 Discussion

3.3.1 Enzymes Involved in Starch Degradation Are Present as HMW Forms

The high degree of cooperation between different classes of enzymes involved in starch synthesis is reflected by the fact that several of them are present in HMW forms. This work is the first step in testing whether complex formation is equally widespread among the proteins involved in starch degradation.

Using GFC, a range of proteins was tested for their potential to be present in HMW forms and thus to potentially form stable complexes with other proteins. From the results, it appears unlikely that a broad interaction network exists between the enzymes of starch degradation. With the limited resolution obtained with the GFC system used, it can be concluded that most complexes would be relatively small in size, around 150kDa (compare elution profiles of BAM1, LSF1, or AMY3) or that proteins are present in their monomeric state (compare elution profiles for LDA, ISA3, BAM2, BAM3 and PWD). However, some proteins are present in potentially very large HMW complexes. DPE2 for instance is present in fractions corresponding to over 448kDa as is the so far uncharacterized starch binding protein ESH. The electrophoretic mobility of DPE2 on non-denaturing PAGE made with total leaf extracts from Arabidopsis is also very slow, adding another level of confidence for the presence of DPE2 in HMW complexes (Chia et al., 2004). In Arabidopsis , DPE2 is involved in a downstream reaction of maltose degradation (see Section 1.7.5) that occurs in the cytosol (Chia et al., 2004) and is thus unlikely to form complexes with the chloroplast localized enzymes involved in the initial steps of starch degradation. 3. Protein-Protein Interactions between Proteins Involved in Starch Degradation 56 ______

Nonetheless, the potential DPE2-complex formation in the cytosol is an interesting topic for further investigations (Fettke et al., 2009).

ESH represents another interesting candidate for further investigations. The diurnal fluctuation of mRNA as well as protein levels over the 24 hours time-course (Mettler, 2006) and its potential to bind to starch granules (Dr S.C. Zeeman, personal communication and Edner et al. (2007) make this protein a candidate for a function in protein complexes involved in starch metabolism. The localization of GWD in fractions larger than its monomeric weight is corroborated by a proteomics approach, where the oligomeric stromal proteome of Arabidopsis was investigated (Peltier et al., 2006). In this study, the NMW of GWD has been suggested to be approximately 320kDa, which correlates with the results presented above, even though the resolution of the GFC is not optimal. Interestingly, the NMW for PWD has been estimated to be 280kDa in the same study. Even though PWD has not been clearly identified as having a NMW larger than its monomeric weight in my investigation, complex formation of PWD with other partners should not be excluded.

AMY3 is present in fractions of above 158kDa and might thus interact with one or several proteins to form heteromers but possibly also form homomers. α-Amylase purified from Vigna mungo seeds has been shown to form homo-dimers, -trimers and further homo-multimers (Koshiba and Minamikawa, 1981) which may be the case of AMY3 in Arabidopsis , although, for larger multimers, this would mean that the GFC substantially underestimated the AMY3 NMW. Interestingly, AMY3 levels are severely reduced in the Arabidopsis sex4 phospho-glucan phosphatase mutant (Zeeman et al., 1998b; Yu et al., 2005; Niittylä et al., 2006). This allows to speculate about an interaction between SEX4 and AMY3, such that AMY3 is destabilized in the absence of SEX4. Consistent with such an idea is the observation that the SEX4 protein is present across the GFC profile, even though most of it is in the low molecular weight (LMW) fractions.

SEX4 has also been identified to be present in MW fractions larger than 500kDa in a proteomics approach looking at megadalton complexes in the stroma of Arabidopsis chloroplasts (Olinares et al., 2010). Such large complexes would not have separated well in our approach but as for some other ‘starch degrading enzymes’, a second look with an improved GFC system should be performed. A classical biochemical purification approach starting from isolated chloroplasts, with an enrichment for starch binding proteins and protein-complexes followed by GFC and MS/MS based identification of proteins in different GFC-fractions might contribute significantly to a broad identification of complex formation relevant for starch degradation and extend the preliminary work presented here.

3. Protein-Protein Interactions between Proteins Involved in Starch Degradation 57 ______

448kDa 158kDa 75kDa A Gel Filtration Chromatography Fractions 31 32 34 35 36 37 38 40 42 50kDa SEX4 37kDa 448kDa 158kDa 75kDa B Gel Filtration Chromatography Fractions 31 32 34 35 36 37 38 40 42 100kDa 75kDa AMY3 50kDa 448kDa 158kDa 75kDa C Gel Filtration Chromatography Fractions 31 32 34 35 36 37 38 40 42 75kDa ISA3 50kDa 448kDa 158kDa 75kDa D Gel Filtration Chromatography Fractions 29 31 32 33 34 35 36 37 38 39 40 41 43 45

BAM3 50kDa 448kDa 158kDa 75kDa 43kDa E Gel Filtration Chromatography Fractions 23-26 27-30 31-34 35-38 39-40 41-42 43-44 45-46 47-48 49-50 51-52 53-56 150kDa 100kDa GWD 75kDa 448kDa 158kDa75kDa 43kDa F Gel Filtration Chromatography Fractions 23-26 27-30 31-34 35-38 39-40 41-42 43-44 45-46 47-48 49-50 51-52 53-56 150kDa 100kDa PWD 75kDa 448kDa 158kDa75kDa 43kDa G Gel Filtration Chromatography Fractions 23-26 27-30 31-34 35-38 39-40 41-42 43-44 45-46 47-48 49-50 51-52 53-56 150kDa 100kDa DPE2 75kDa 448kDa 158kDa75kDa 43kDa H Gel Filtration Chromatography Fractions 23-26 27-30 31-34 35-38 39-40 41-42 43-44 45-46 47-48 49-50 51-52 53-56 75kDa 50kDa ESH 37kDa 158kDa 75kDa 43kDa I Gel Filtration Chromatography Fractions 28 32 36 39 41 43 45 47 49 51 54 59 75kDa BAM2 50kDa 448kDa 158kDa 75kDa J Gel Filtration Chromatography Fractions 31 32 34 35 36 37 38 40 42 LDA

3. Protein-Protein Interactions between Proteins Involved in Starch Degradation 58 ______

Figure 2-2. Native Molecular Weight Determination of Proteins Involved in Starch Degradation Total protein extracts of wild-type leaves were applied to a Sephacryl 300HR™ XK16 column operated by an ÄKTA Explorer100 at a flow rate of 0.4mL min -1. Elution fractions from the column were collected, pooled as indicated and analysed for the presence of the glucan phosphatase SEX4 (A), α- amylase AMY3 (B), isoamylase ISA3 (C), β-amylase BAM3 (D), glucan-water dikinase GWD (E), phospho-glucan water dikinase PWD (F), disproportionating enzyme DPE2 (G), esterhaze ESH (H) and β-amylase BAM2 (I) proteins using specific antibodies and western blotting. The distribution of limit- dextrinase LDA (J) was monitored using a native gel containing a coloured substrate uniquely degraded by LDA. Gel permeation chromatography does not reveal presence of ISA3 or LDA in high molecular weight fractions. Only the monomeric forms of these proteins can be observed. SEX4 and AMY3 are at least partially present in high molecular weight fractions. The elution spectrum of SEX4 is broad, extending into high molecular weight fractions but the majority of the protein is present in lower molecular weight fractions. AMY3 is present in fractions larger than its monomeric weight. Gel permeation chromatography of wild-type extracts does not reveal that either BAM3 or BAM2 are present in a high molecular weight form.

3.3.2 14-3-3 Proteins - Potential Interaction Partners of Proteins Involved in Starch Degradation

14-3-3 proteins are a family of highly conserved proteins, present in animals and plants (Oecking and Jaspert, 2009). In Arabidopsis , 15 genes are annotated to encode 14-3-3 proteins. For 12 of them mRNAs have been detected (Rosenquist et al., 2001). Direct modulation of target enzyme activity has been observed as well as a scaffolding function of the 14-3-3 proteins (Chang et al., 2009). 14-3-3 proteins are implicated in the regulation of distinct biological processes by phosphorylation-dependent protein-protein interaction. Among these processes, starch metabolism has been identified, as 14-3-3 proteins have been identified bound to starch granules (Sehnke et al., 2001)and because the antisense downregulation of the ε-subgroup of 14-3-3 proteins led to a slight sex phenotype (Sehnke et al., 2001). As 14-3-3 proteins interact with SSIII in Arabidopsis and with GBSS, SSI, SSII and SBEIIa in barley (Sehnke et al., 2001; Alexander and Morris, 2006; Chang et al., 2009), the sex phenotype has been attributed to an effect on starch synthesis. However, an impact of 14-3-3 proteins on starch degradation cannot be excluded, as starch degrading enzymes have also been identified as 14-3-3 clients ( β-amylase and α-amylase from barley (Alexander and Morris, 2006) and GWD from Arabidopsis (Chang et al., 2009)). The HMW form of GWD observed in the GFC experiment shown here might reflect the interaction of GWD with 14-3-3 proteins and thereby with further interaction partners.

It is interesting to note that 14-3-3 proteins generally but not exclusively, require phosphorylation of their target for binding (Ottmann et al., 2007). The AMY3 and DPE2 proteins have been observed to be present in HMW fractions in the experiments described above, even though they do not possess the typical 14-3-3 binding motifs, phosphorylated peptides belonging to AMY3 and DPE2 have been identified (Heazlewood et al., 2008; Durek et al., 2010), allowing one to speculate about potential phosphorylation- 3. Protein-Protein Interactions between Proteins Involved in Starch Degradation 59 ______dependent interactions. The fact that some of the key enzymes in starch degradation are present in HMW weight complexes and are confirmed or potential targets of 14-3-3 proteins warrants a thorough investigation of 14-3-3 function in modulating starch degradation.

3.3.3 Limitations of a GFC Based Approach

The approach used in these experiments is limited by a number of factors which could either mis-identify or fail to detect HMW complexes. One limitation is related to the number and size of the fractions collected from the filtrations and later on analyzed by western blotting. A selection has to be made for the analysis of fractions by western blotting and the elution peak of a particular protein can be missed (e.g.: DPE2 or BAM3). The first western blots may only serve as a guideline, for further investigations as soon as the protein or protein complex to focus on has been identified. The use of plant crude extracts may be a factor favoring the identification of false positives. Mixing chloroplast and cytosol localized proteins, associations between proteins from different compartments are possible in vitro which might not genuinely occur in planta. Increasing the concentration of the total protein extract might also lead to the aggregation of proteins which would in planta not interact with each other. Conversely, it is plausible that a given protein complex assembles only under certain circumstances in the chloroplast (e.g.: in the presence of starch granules/amylopectin or under high maltose conditions) but disassembles upon extraction under the buffer conditions used in the presented GFC experiments.

3.3.4 A Knowledge-Based Decision to Focus on BAM1 and LSF1 to Confirm and Characterize Complex Formation

The aforementioned reasons for failure to detect a protein complex or for detection of false positive interactions require that an independent verification of a suspected interaction (or also non-interaction) be made. As such follow up experiments are more time consuming, I decided to focus on two proteins identified in HMW fractions; BAM1 and LSF1. I confirmed previous observations suggesting that BAM1 was present in HMW fractions from GFC experiments conducted in earlier work (Delatte et al., 2005; Fulton et al., 2008) and thus might form complexes with so far unrecognized interaction partners. Reasons for choosing LSF1 were mainly that lsf1 mutants in Arabidopsis display a starch excess phenotype but its function in starch metabolism remains unclear (Comparot-Moss et al., 2010). Such a guilty-by-interaction approach may not provide definite answers about a protein’s function but it certainly offers a starting point for a thorough characterization. In addition it offers the potential to further extend the interaction- network of proteins involved in starch degradation. 4. The Interaction Network of BAM1 and LSF1 60 ______

4. The Interaction Network of BAM1 and LSF1

4.1 Introduction

Among the BAMs present in the chloroplast, BAM3 and BAM1 are currently considered the major catalytic β-amylases involved in starch degradation. Plant lines lacking BAM3 are unable to fully degrade their starch during the night and display a sex phenotype. While bam1 mutants do not have a sex phenotype, the heterologously expressed BAM1 protein has high catalytic activity and loss of BAM1 in addition to BAM3, enhances the bam3 sex phenotype more than two-fold (Fulton et al., 2008). In GFC of plant crude extracts performed by Delatte et al. (2005), the activity, later attributed to BAM1 by Fulton et al. (Fulton et al., 2008) co-eluted with the ISA1/ISA2 DBE complex activity. This DBE complex has, in several plant species, been attributed a NMW of between 350 and 500kDa (Fujita et al., 1999; Dauvillee et al., 2001b; Delatte et al., 2005; Wattebled et al., 2005; Takashima et al., 2007). The presence of BAM1 in a HMW form was therefore quite likely and made BAM1 an interesting candidate for further investigations.

BAM1 stands out among the other chloroplast localized BAMs as it is redox regulated. The heterologously expressed BAM1 protein is only active in its processed form, with the predicted chloroplast transit peptide removed (Sparla et al., 2006). The enzyme activity has a pH optimum of 6 to 8, depending on whether the activity was tested using an artificial chlorogenic substrate (PNPG5, see Section 2.14) or with amylopectin. The heterologously expressed protein was monomeric, with an apparent molecular mass of 60kDa. The reductive activation of BAM1 requires the reduction of a disulfide bridge between Cys-32 and Cys-470 and was most efficiently mediated by the chloroplast localized thioredoxin f1 (Sparla et al., 2006; Valerio et al., 2010). Reductive activation is counterintuitive for an enzyme involved in starch degradation, as the chloroplast is thought to be a more reducing environment when photosynthesis is occurring, hence the enzyme should be activated during the day. Interestingly, BAM1 is not the only enzyme involved in starch degradation which is reductively activated. The glucan-kinase GWD as well as the phospho-glucan phosphatase SEX4 have been shown to be activated under reducing conditions in vitro (Mikkelsen et al., 2005; Sokolov et al., 2006). Despite the in vitro data, BAM1 and SEX4 have indirectly been shown to be active during the night (Fulton et al., 2008; Kötting et al., 2009} and phosphorylation of starch is increased during night-time starch degradation (Ritte et al., 2004). These findings suggest that either the redox conditions might be locally heterogeneous in chloroplasts or that redox regulation of GWD, SEX4 and potentially also BAM1, not necessarily follows the light/dark activation/inactivation paradigm. 4. The Interaction Network of BAM1 and LSF1 61 ______

Expression studies with the BAM1 promoter driving expression of the β-glucuronidase or green fluorescent proteins (GFP), have suggested that BAM1 is expressed mainly in the guard cells of leaves of young plants where it is involved in the degradation of starch during the day; a process which has been related to stomatal opening (Lawson, 2009). Compared to young, unstressed plants, the expression of BAM1 is increased older as well as in osmotically stressed plants (Valerio et al., 2010). As starch is degraded under osmotic stress to produce osmolytes (Kempa et al., 2008), it has been concluded that BAM1, besides its function in guard cells, might activate a starch degradation pathway in illuminated mesophyll cells under osmotic stress (Valerio et al., 2010). Despite this work, the insight into the functions of the different BAMs in starch degradation and the relative contributions of BAM1 and BAM3 to starch degradation are still limited. One of the goals of this work was the identification of interaction partners of BAM1 to provide further insight into the regulation of this enzyme and its role in starch degradation.

Since BAMs are not able to work past a phosphorylated glucose residue in a glucose chain, the removal of phosphate from glucan chains is essential for starch degradation. The importance of this process is reflected by the sex phenotype of mutants lacking the phospho-glucan phosphatase SEX4 (Kötting et al., 2009). The two genes LSF1 and LSF2 (Like-SEX-Four 1 and 2) in the Arabidopsis genome which are homologous to SEX4 both encode proteins that localize to the chloroplast. LSF2 is an active glucan phosphatase, acting in parallel to SEX4 with a partially redundant function (Dr D. Santelia, personal communication). LSF2 is different in its domain structure as compared to SEX4; it lacks the CBM that SEX4 possesses.

The overall domain architecture is more similar between LSF1 and SEX4 (Fig4-1). The two proteins share 35% identity and 54% similarity between their phosphatase domains (dual-specificity phosphatase domain, DSPc) and 35% identity and 53% similarity between their CBMs (Comparot-Moss et al., 2010). LSF1 has also an N-terminal extension of about 200 amino acids, which is not present in SEX4. Within this extension, a so called PDZ-domain has been recognized (aa75-162). PDZ is an acronym combining the first letters of the names of the three proteins (Post synaptic density protein, Drosophila disc large tumor suppressor, and Zonula occludens-1 protein), where PDZ domains were first discovered. PDZ domains are 80 to 90 amino acids long modular protein interaction domains which can be found in a large number of species, ranging from bacteria and yeast to multicellular organisms like worms, mammals and plants (Ponting, 1997; Sheng and Sala, 2001). In a sequence specific way, PDZ domains mainly bind to specific short sequences at the C-terminus of their target proteins but also recognize internal sequences in some cases (Elkins et al., 2010). A high diversity of PDZ binding partners is achieved by variable amino acids lining the peptide-groove of the PDZ domain (Sheng and Sala, 2001). PDZ containing proteins have 4. The Interaction Network of BAM1 and LSF1 62 ______often been found to form homo-multimers where distinct surfaces of the PDZ domain may be used for multimerization and target peptide binding.

The N-terminal extension is not the only difference between LSF1 and SEX4. Another one is found in the conserved amino acid residues of the DPSc domains of LSF1 and SEX4. LSF1 contains an atypical DPSc motif, in that the histidine preceding the catalytic cysteine is exchanged for a threonine (Fig 4-1). This histidine is thought to be important for catalytic activity as it lowers the pKa of the catalytic cysteine during (Zhang and Dixon, 1993). However, it remains to be established whether the change to a threonine preceding the catalytic cysteine indeed abolishes LSF1 phosphatase activity. Although not a core question of the research presented here, I addressed the problem using tagged LSF1 protein partially purified from plant leaves.

Figure 4-1. Domain Structure of LSF1 and SEX4 LSF1 and SEX4 have predicted N-terminal chloroplast transit peptides (cTP) of 61 and 54 aa respectively. Both proteins contain a carbohydrate binding module (CBM) and a predicted dual specificity phosphatase catalytic (DSPc) domain. The core residues for phosphatase activity (CX 5R) are conserved in both proteins. A histidine (H) preceeding the catalytic cysteine (C) is exchanged for a threonine (T) in LSF1, possibly making LSF1 an inactive phosphatase. In contrast to SEX4, LSF1 has an N-terminal extension, containing a PDZ (an acronym of Post synaptic density protein, Drosophila disc large tumor suppressor and Zonula occludens-1 protein) domain. PDZ domains mediate protein-protein interactions. Figure redrawn based on Comparot-Moss et al. (2010) and Vander Kooi et al. (2010).

The investigation of the lsf1 mutant phenotype suggests that while LSF1 is homologous to SEX4, its function in starch degradation might be different. Plants lacking SEX4 accumulate phosphorylated oligosaccharides. These are mainly released by the ISA3 DBE and the α-amylase AMY3. Due to the lack of glucan-phosphatase activity the phospho-oligosaccharides (P-oligos) cannot be dephosphorylated or degraded and thus accumulate during the process of starch degradation (Kötting et al., 2009). Like sex4 mutants, lsf1 mutants display a sex phenotype but unlike sex4 , do not accumulate P-oligos (Comparot- Moss et al., 2010). While sex4 leaf extracts display a reduced capacity to release phosphate from purified and in vitro phosphorylated starch granules as compared to wild-type extracts, lsf1 extracts do not. The sex phenotype of lsf1 is lower than that of sex4. However, the lsf1 /sex4 double mutant displays a 25% increase of the sex phenotype as compared to the sex4 mutant. It thus appears unlikely that the function of 4. The Interaction Network of BAM1 and LSF1 63 ______

LSF1 in starch degradation is similar to SEX4 and questionable whether LSF1 would dephosphorylate oligosaccharides at all. If LSF1 was a phospho-glucan phosphatase, its substrates would have to be different to those dephosphorylated by SEX4 and would have escaped detection so far (Comparot-Moss et al., 2010). It remains possible that LSF1 requires a specific structural setting on the granule surface (e.g.: very short external chains), a specific interaction partner or cellular conditions in order to dephosphorylate glucan chains. Another possibility is that LSF1 is not a phospho-glucan phosphatase but rather a protein phosphatase, like many of the other dual-specificity phosphatases (Luan, 2003). Given that the function of LSF1 is not clearly discernible looking at the phenotype of lsf1 plants, investigating its interaction partners might help to shed light on its role in starch degradation.

To investigate the potential protein interaction network of BAM1 and LSF1, initial experiments were conducted to find possible interactors of BAM1. Intriguingly, these experiments have suggested that BAM1 and LSF1 might interact with each other. A set of different approaches was used to confirm and characterize this interaction and to extend the potential interaction network. Exploratory experiments were also conducted to investigate the function and significance of the interaction of BAM1 and LSF1 and to investigate interaction of LSF1 with other proteins independently of BAM1.

4.2 Results

4.2.1 bam1 and lsf1 Mutants Lack One Common β-Amylase Activity

To date, BAM1 had already been suggested to be present in HMW complexes but the nature of the BAM1 complex and its composition was unclear. To reveal BAM1 interaction partners and to screen for further possible protein interactions, I looked for alterations in the electrophoretic mobility of native BAM1. Soluble protein extracts from leaves of different mutants, harvested at the end of the night, were separated on non-denaturing (native) polyacrylamide gels containing amylopectin, glycogen or β-limit- dextrin (amylopectin digested with an excess of commercial β-amylase). After electrophoresis and incubation, enzyme activities modifying the substrate in the gels are identified by staining the gels with an

I2KI-solution (Lugol-Solution). Enzymes can degrade or modify the substrates in the gel resulting in clear or coloured bands depending on substrate and enzyme. In native gels, proteins are separated according to their electric surface charge at the given pH, their substrate binding properties and according to their NMW.

On native gels containing amylopectin, two bands of amylopectin degrading activity (clear bands) are missing in the bam1 mutant (Fig 4-2A, compare also Fig 4-3). One well focussed band of slow electrophoretic mobility (a) and a more diffuse region of activity of faster electrophoretic mobility (a’). 4. The Interaction Network of BAM1 and LSF1 64 ______

No difference between wild-type and bam1 mutant extracts was visible on gels containing β-limit-dextrin, on which β-amylases are not active (Fig 4-2C). Probably owing to the generally lower activity on the glycogen containing gels only a single band of activity was visibly missing in the bam1 mutant (Fig 4-2B, a). Interestingly, lsf1 lacks the slow-migrating BAM1 activity, while the faster migrating BAM1 activity was unchanged, if not increased (Fig 4-2A, a and a’).

No other mutant line analysed, displayed a change in amylopectin modifying activity, other than that affected by the mutation. However, the fact that a large number of enzyme activities involved in starch degradation cannot be detected on these native gels (e.g.: SEX4 or AMY3) may limit the power of such an approach. Furthermore, the electrophoretic conditions used in the native gel system might cause instability of a protein complex and thus failure to detect it. Despite its limitation, this approach has provided evidence that in the lsf1 mutant, the electric surface charge, the substrate affinity or the NMW of BAM1 are altered as compared to the wild type.

4. The Interaction Network of BAM1 and LSF1 65 ______

Figure 4-2. The lsf1 Mutant is Deficient in a Slow-Migrating BAM1 Activity Soluble protein extracts from leaves of different mutants were separated on non-denaturing polyacrylamide gels, containing amylopectin (A), glycogen (B) or β-limit-dextrin (C). Gels were incubated to allow for enzyme activity and alterations in the substrate were revealed by staining gels with Lugol-solution. β-Amylase activities of different electrophoretic mobility (a and a’) are missing in the bam1 mutant. The slower migrating β-amylase activity (a) is missing in the lsf1 mutant as well, while the fast migrating activity (a’) is present. No changes in β-amylase activity can be detected in mutants lacking other enzymes involved in starch degradation or synthesis (b: LDA activity, c: ISA3 activity, d: DPE1 activity, e: ISA1/2 activity, f: SBE2 activity). Abbreviations: Wt: wild type, bam : β-amylase; lsf1 : Like- SEX4-1; sex : starch excess; amy : α-amylase; lda: limit-dextrinase; isa : isoamylase; pwd : phospho-glucan- water dikinase; gwd : glucan-water dikinase; dpe : disproportionating enzyme; s be : starch branching enzyme.

4. The Interaction Network of BAM1 and LSF1 66 ______

4.2.2 The Electrophoretic Mobility and Native Molecular Weight of BAM1 and LSF1 Depend on LSF1 and BAM1 Respectively.

To investigate the link between the changes in BAM1 mobility and LSF1 protein, western blotting of amylopectin containing native gels was performed, followed by immunodetection of BAM1 and LSF1 respectively. These gel blots revealed that the BAM1 protein co-localizes with the two bands of BAM1 activity missing in bam1 as expected. The LSF1 protein partially co-localizes with the low-mobility BAM1 activity (Fig 4-3). In the absence of LSF1, BAM1 is exclusively present in the fast migrating form. A similar increase in electrophoretic mobility of LSF1 can be observed in the absence of BAM1.

By mixing and pre-incubation of bam1 and lsf1 extracts, the slow migrating, BAM1 activity can be restored and both, BAM1 as well as LSF1 display a reduced electrophoretic mobility. These findings suggest that BAM1 and LSF1 might interact. However, as electrophoretic mobility is influenced by several factors and not only the NMW, a mobility shift might as well result from changes in secondary modifications or substrate binding. I have used GFC to exclude these factors and to further investigate the putative interaction of BAM and LSF1.

Blotted Native Gel bam1 bam1 bam1 Wt bam1 lsf1 +lsf1 Wt bam1 lsf1 +lsf1 Wt bam1 lsf1 +lsf1

α-BAM1 α-LSF1

Figure 4-3. The Slow Migrating BAM1 Activity is Restored by Mixing bam1 and lsf1 Mutant Extracts Total protein extracts of leaves of wild type, bam1 and lsf1 were run on native gels containing amylopectin together with a mixture of equal amounts of bam1 and lsf1 extract. Mixing bam1 and lsf1 mutant extracts restores the slow-migrating BAM1 activity (asterisks). Western blotting of native gels and immunodetection of BAM1 (middle) and LSF1 (right) reveals that BAM1 and LSF1 are present in distinct locations on the gel and that BAM1 co-localizes with the slow- and fast-migrating BAM1 activities. In the absence of LSF1, BAM1 is shifted to the faster migrating band. In the absence of BAM1, the electrophoretic mobility of LSF1 is increased. In bam1 mutant extracts, LSF1 does not co-localize with an amylolytic activity. Protein concentrations were determined for each extract and 40µg total proteins are loaded in each lane.

4. The Interaction Network of BAM1 and LSF1 67 ______

4.2.2.1 BAM1 and LSF1 are Present in High Molecular Weight Complexes A self-made Sephacryl 300HR™ (GE Healthcare) column in an XK16 casing operated by an ÄKTAexplorer100 (GE Healthcare) was calibrated using molecular standards as described in Chapter 3. When total soluble protein extracts made from wild-type leaves were separated over the column, BAM1 and LSF1 were both present in HMW forms of below 150kDa (Fig 4-4 A and B, top panels). The elution profile of LSF1 is considerably broader than that of BAM1 and an estimation of the molecular weight corresponding to the peak of LSF1 elution is difficult. Although the elution profile of BAM1 is broad, a peak of BAM1 elution can be assigned to a molecular weight of between 100 and 150kDa. A more precise estimation is inappropriate given the broad elution spectra. However, upon loss of LSF1, the elution profile of BAM1 markedly shifts towards lower molecular weight fractions, which are approaching the monomeric weight of BAM1 (Fig 4-4A, lower panel). Likewise, the elution profile of LSF1 is shifted to lower molecular weights in the bam1 mutant (Fig 4-4B, lower panel).

Thus the GFC experiments provide further evidence that BAM1 and LSF1 are present in HMW complexes and that their NMW decrease in the absence of one another. This corroborates the idea that they interact with each other. Owing to the broad elution spectra several questions remain unsolved from these experiments. The stochiometry of a potential BAM1-LSF1 complex cannot unambiguously be determined and other interaction partners might be present in the BAM1-LSF1 complex. BAM1 or LSF1 might independently form complexes with other proteins and exist to a certain amount in their monomeric forms.

4. The Interaction Network of BAM1 and LSF1 68 ______

Figure 4-4. BAM1 and LSF1 Depend on Each Other to Be Present in High Molecular Weight Forms Gel filtration chromaotgraphy of wild-type and mutant leaf extracts. Total leaf extracts were applied to a Sephacryl300™ HR XK16 column, operated by an ÄKTAExplorer100. Elution fractions from the column were collected and analysed for the presence of BAM1 or LSF1 by SDS-PAGE and immunoblotting using specific antibodies. (A) Anti-BAM1 western blots of a separation of wild-type (top) and lsf1 (bottom) extracts. The relative abundance of BAM1 in the fractions was estimated by comparing band intensity of every lane to the sum of band intensities. In wild-type extracts (diamonds), BAM1 is present in a high molecular weight form of up to 150kDa as judged from the elution profile of known molecular weight standards. In lsf1 extracts (squares), BAM1 is present as lower molecular weight form. (B) Extracts of wild type (triangles) and bam1 (circles) were applied to gel filtration chromatography and the fractions collected were analysed by anti-LSF1 western blots. The relative abundance of LSF1 in the fractions is expressed as a fraction of the sum of band intensities. LSF1 is present in a broad spectrum of molecular weights in wild-type extracts. In the absence of BAM1, a shift towards lower molecular weights is observed. 4. The Interaction Network of BAM1 and LSF1 69 ______

4.2.2.2 BAM1 and LSF1 are Present in the Same Complex The changes of NMW of BAM1 and LSF1 in the respective mutants as revealed by GFC provide only indirect evidence for the presence of BAM1 and LSF1 in the same complex (es). Therefore I conducted co-immunoprecipitation assays, using antibodies recognizing BAM1 or LSF1 (Fig 4-5). The LSF1 protein was pulled down by the BAM1 antibody from wild-type extracts but not from bam1 mutant extracts (Fig 4-5A). This is indicative of the presence of LSF1 and BAM1 in the same complex (es) and excludes non- specific interactions with the resin used in the co-immunoprecipitation. In reverse, the BAM1 protein was pulled down from wild-type extracts but not from lsf1 extracts using the LSF1 antibody (Fig 4-5B). Thus, these reciprocal co-immunoprecipitation experiments confirm that BAM1 and LSF1 occur in the same complex.

A BAM1-CoIP BAM1-CoIP Input Elution Input Elution Wt bam1 Wt bam1 Wt bam1 Wt bam1 95kDa 95kDa

72kDa 72kDa

55kDa 55kDa

α-BAM1 α-LSF1

B LSF1-CoIP LSF1-CoIP Input Elution Input Elution Wt lsf1 Wt lsf1 Wt lsf1 Wt lsf1 95kDa 95kDa

72kDa 72kDa

55kDa 55kDa

α-LSF1 α-BAM1

Figure 4-5. Reciprocal Co-Immunoprecipitation Reveals Physical Interaction of BAM1 and LSF1 Antibodies specifically recognizing BAM1 and LSF1 were used to co-immunoprecipitate LSF1 and BAM1 respectively. (A) Using the BAM1 antibody, the BAM1 protein (left panel, arrowhead) and the LSF1 protein (right panel, arrowhead) were pulled down from wild-type extracts but not from bam1 extracts. (B) Using the LSF1 antibody, the LSF1 (left panel, arrowhead) and the BAM1 protein (right panel, arrowhead) were pulled down from wild-type extracts but not from lsf1 extracts. 4. The Interaction Network of BAM1 and LSF1 70 ______

4.2.3 Reciprocal TAP-tagging Confirms the BAM1-LSF1 Interaction

To further confirm the BAM1-LSF1 interaction and to identify potential new interaction partners of BAM1 and/or LSF1, a tandem-affinity-purification (TAP) approach was used. Tandem affinity purification allows a two-step, affinity-based purification of protein-protein complexes under conditions which help to retain weak interactions. TAP-tags were fused to the C-termini of the LSF1 and BAM1 coding regions and stably expressed under the control of a double 35S CaMV (cauliflower mosaic virus) promoter in mutant lines lacking the respective proteins (see Section 2.9 for details and Fig 4-6 for the TAP-tag structure). For control and comparison, BAM2, BAM3 and BAM4, were also tagged and transformed into the respective null mutants. BAM2 and BAM3 are catalytically active, while BAM4 is catalytically inactive (Fulton et al., 2008). To investigate BAM1 interacting proteins independently of LSF1, the BAM1-TAP construct was also expressed in the lsf1 mutant background.

3C Cleavage 2x35S 9xmyc-Tag Site Nos-Ter Coding Region TMV 6xHis-Tag 2x IgG-BD U1Ω

Figure 4-6. pC-TAPa Structure The coding regions of BAM1, BAM2, BAM3, BAM4 and LSF1 without the stop codons were cloned into the pC-TAPa vector (Rubio et al., 2005) and transformed into the respective single mutants. The pC- TAPa vector would allow a three-step-purification with the proteinA IgG binding domain (IgG-BD), the His-Tag and the myc epitope. The IgG-BD is separated from the His-tag and the myc-tag by a 3C- protease cleavage site, for elution of tagged proteins from the IgG-Sepharose beads during the first purification step. Expression of the product is driven by a double 35S promoter from the cauliflower mosaic virus. Translation is enhanced by the tobacco mosaic virus (TMV) U1 Ω translational enhancer. The nopaline synthase terminator (Nos-Ter) sequence is located downstream of the expression cassette.

4.2.3.1 Complementation of Activities on Native Gels The BAM1-TAP construct was able to restore the slow migrating BAM1-LSF1 activity on amylopectin containing native gels (missing in both, bam1 and lsf1 mutants) while hardly any activity could be observed for the faster migrating BAM1 activity (Fig 4-7, bands a and b). In the lsf1 mutant, the BAM1- TAP construct could not complement the lack of the slow migrating BAM1 activity (Fig 4-7, bands a and b). The LSF1-TAP construct also complemented the slow migrating BAM1-LSF1 activity. These results confirm that the complex formation between BAM1 and LSF1 occurs, even if one of the interaction partners is TAP-tagged (Fig 4-7 band c). For bam2 , bam3 and bam4 mutant extracts, no changes of activity can be detected on amylopectin containing native gels as compared to extracts made from wild- 4. The Interaction Network of BAM1 and LSF1 71 ______type leaves (see Fig 4-2). Within the range of biological variation observed on native gels, leaf extracts of lines expressing the BAM2-TAP and the BAM3-TAP construct do not differ in their activity on amylopectin containing native gels, from the wild type. Additional bands are, however, visible in the BAM4-TAP expressing line. These bands did not co-localize with the BAM4-TAP tagged protein as investigated by western blotting (Fig 4-7, bands f and g) and might be due to secondary effects or experimental artefacts. The fact that TAP-tagged BAM4 protein runs in two distinct locations on the native gel is interesting but has not been investigated further.

For the BAM1-TAP expressing lines, western blotting and immunodetection of the tagged protein revealed correlation of protein and activity localisation on native gels (Fig 4-7, bands a’ and b’). The pattern of the LSF1-TAP protein on amylopectin containing native gels partially overlaps with the observed BAM1 activity (Fig 4-7 band c’). A large portion of the tagged LSF1 is present in bands not correlating with the BAM1-LSF1 complex activity (Fig 4-7, bands d and e). Band d could either correspond to LSF1-TAP in complex with another protein (other proteins) or the monomeric LSF1-TAP fusion protein. Band e could either correspond to the monomeric LSF1-TAP protein or a degradation product thereof. From the experiments conducted so far, it is impossible to discriminate between the two possibilities.

4. The Interaction Network of BAM1 and LSF1 72 ______

Wt bam1 lsf1

a c

b

a’ c’ f d

b’ e g

Figure 4-7. TAP-tagged BAM1 and LSF1 Restore Slow and Fast Migrating BAM1 Activities Mutant lines expressing the respective BAM-TAP or LSF1-TAP transgenes were selected and checked for restoration of the BAM activities on amylopectin containing native gels (top). Western blotting of native gels run in parallel and immunodetection of the tagged proteins using an α-myc antibody shows localization of the tagged proteins in the gels (bottom). BAM1 in the bam1 mutant and LSF1 in the lsf1 mutant restore the BAM1-LSF1 complex activity (a and c). Western blotting of the native gels (bottom) confirms that the BAM1-TAP and LSF1-TAP co-localize with bands of BAM1 activity (a’ and c’). In the absence of LSF1, BAM1-TAP localizes to the fast migrating BAM1 activity (b and b’). Only a fraction of the LSF1-TAP protein co-localizes with the BAM1-LSF complex activity. LSF1-TAP is present in different bands on native gels (c’, d and e). While BAM2-TAP localizes in a single fast migrating band (g) BAM3-TAP and BAM4-TAP are present in at least two regions on the native gels (f and g).

4. The Interaction Network of BAM1 and LSF1 73 ______

4.2.3.2 Complementation of sex Mutant Phenotypes The functionality of the TAP-tagged proteins was further evaluated by quantitative starch measurements of different mutant lines transformed with the BAM-TAP and LSF1-TAP constructs. The quantifications of starch at the end of the night and at the end of the day for the best complementing lines are shown in Fig 4-8. The bam3 sex phenotype could be complemented by 50% at the end of the night and by 53% at the end of the day by the BAM3-TAP construct. As the loss of the BAM1 protein does not cause a starch excess phenotype in a wild-type background but enhances the starch excess phenotype in a bam3 mutant background (Fulton et al., 2008), the BAM1-TAP construct was stably expressed in the bam1/bam3 double mutant. Compared to the bam3 phenotype, the bam1/bam3 phenotype was reduced by 48% and 34% at the end of the night and at the end of the day, respectively, through the expression of the BAM1- TAP construct. The sex phenotype of bam4 could be reduced by 68% and 61% in two BAM4-TAP tagged lines at the end of the night. Surprisingly, the starch levels were only reduced by 18% and 49% as compared to the wild-type levels at the end of the day. In the best complementing line expressing the LSF1-TAP transgene, starch levels were reduced by 74% at the end of the night and by 98% at the end of the day when compared to the lsf1 mutant sex phenotype. The bam2 mutant has a wild-type phenotype (Fulton et al., 2008), so no test for the functionality of the BAM2-TAP construct was possible. Together with the complementation of the BAM1 activities on native gels, these experiments suggest that the TAP- tag does not abolish the functionality of the fusion protein. However, a reduction in functionality is likely, as complete complementation or even over-complementation have not been observed.

4. The Interaction Network of BAM1 and LSF1 74

40 End of Night End of Day FW) -1 30

20 20 20 Starch Contents 10 10 10 (mg Glc. Equivalents g

0 0 0 Wt bam1 bam3 bam1/ BAM3-TAP BAM1-TAP Wt bam4 BAM4-TAP BAM4-TAP Wt lsf1 LSF1-TAP bam3 (bam3) (bam1/bam3) (bam4 )#6-1 (bam4 )#5-3 (lsf1 )

Figure 4-8. TAP-tagged BAM1, BAM3, BAM4 and LSF1 Can Complement sex Mutant Phenotypes Mutant lines, transformed with the respective TAP-fusion construct were grown on selective MS plates and resistant plants were transplanted onto soil, alongside with the wild-type and mutant controls. Starch was extracted in perchloric acid from whole rosettes harvested at the end of the day or end of the night and immediately frozen in liquid N 2. Starch was measured after enzymatic hydrolysis to glucose and starch levels are expressed as mg glucose equivalents per g fresh weight (FW). Measurements were done in three independent experiments for the different TAP-tag transformations and are therefore shown in separate panels, including the wild-type and mutant controls. Only results for the best complementing lines are shown, lines of two independent transformation events are shown for BAM4-TAP. Values are means of five replicate samples ±SE. All transformed lines showed a significant reduction of the mutant starch excess phenotype at the end of the night (t-test; P<0.01).

4. The Interaction Network of BAM1 and LSF1 75

4.2.3.3 Confirmation of Interaction of BAM1 and LSF1 In a two-step affinity purification, the tagged proteins and their potential interaction partners were partially purified from crude extracts of leaves. This process involved binding of the tagged proteins to an IgG-binding resin in a first purification step, followed by proteolytic cleavage to release tagged proteins from the IgG-binding resin. In a second step, tagged proteins were further purified by means of IMAC (immobilized metal affinity chromatography; see Section 2.10 for details) and eluted in imidazole containing media. The dilute elutions from the second purification step were concentrated and proteins co-purifying with BAM1 and LSF1 visualized on silver stained SDS-PAGES (Fig 4-9).

In order to identify proteins unspecifically interacting with the affinity resins used in the purification, crude wild-type leaf extracts were treated the same way as extracts from transformed plants. From the purifications of wild-type extracts as well as from transformed plants, it was obvious that the partial purifications still contained a fair number of contaminating proteins, with the large subunit of Rubisco, just below the 55kDa size marker and the protease used in the purification procedure at approximately 46kDa being the most prominent ones. Thus, proteins identified from wild-type controls were considered unspecific interactors with the resin and were removed from the list of interactors for the different TAP- tagged proteins.

In order to identify co-purifying proteins, gels were cut into sections and proteins present in each section were identified using tandem mass-spectrometry (MS/MS) after in-gel tryptic digests. MS/MS spectra were analyzed using ‘Mascott’ (Matrix Science, London, UK) and validated by ‘Scaffold’ (Proteome Software Inc., Portland, USA). Protein identifications were only accepted if they could be established at greater than 95% probability (Nesvizhskii et al., 2003), based on at least 2 peptides.

Bands corresponding to the tagged BAM1 protein (Fig 4-9 A and B, region a) and the tagged LSF1 protein (Fig 4-9 A, bands d and d’) were clearly discernible and matched the predicted protein sizes (the native protein plus the part of the TAP-tag remaining after the proteolytic cleavage in the purification procedure). Both, BAM1-TAP and LSF1-TAP fusion proteins were present in double bands. Separate tandem MS/MS analyses of the two bands for LSF1 suggest that the lower band might represent a degradation product, missing an N-terminal moiety of the full length protein. The doublette for the BAM1-TAP was not separated prior to MS/MS and its nature is unclear so far. Throughout the tandem- affinity-purification procedure the BAM1-LSF1 complex from either BAM1-TAP or LFS1-TAP purifications retained amylolytic activity and electrophoretic mobility on native gels comparable to the unpurified BAM1-LSF1 complex in crude extracts (Fig 4-9A, region f). The purifications were free from amylolytic activities detectable on native gels, except that attributable to the BAM1-LSF1 complex. 4. The Interaction Network of BAM1 and LSF1 76

Purifications of wild-type extracts, not containing any TAP-tagged protein, as well as elutions of the BAM2-, BAM3- and BAM4-TAP tagged proteins did not contain detectable amylolytic activity. At least for BAM3, this is surprising as it is an active BAM. However, no activity of BAM3 has so far been detected on native gels; they might not be suitable to reveal BAM3 activity.

In the BAM1-TAP purification, the LSF1 protein was identified in the region of the gel corresponding to approximately 70kDa (Fig 4-9A, region b). In the LSF1-TAP purification, not only the expected BAM1 protein was identified in a region of the gel corresponding to about 50-60kDa (Fig 4-9A, region e) in size but also BAM3.

Table 4-1 summarises the interactions among the TAP-tagged proteins which were purified from three independent leaf extracts. Proteins identified from wild-type extracts processed through all steps of the purification procedures were removed from the list of proteins interacting with the TAP-tagged proteins. Using the TAP tagging approach, the BAM1 and LSF1 interaction was confirmed again. In addition, BAM3 has been found to interact with LSF1 in three independent experiments (two two-step tandem affinity purifications and one one-step purification making use of the myc-tag, which is part of the TAP- tag; Fig 4-6). LSF1 was also found to interact with the TAP-tagged BAM3 in two out of three replicate purifications. In the third (the myc-CoIP), it was not found. However, even the BAM3 protein was only just detectable, suggesting a poor purification yield in this replicate. Thus, these data confirm the LSF1- BAM3 interaction. BAM2 or BAM4 were not detected among the proteins interacting with LSF1 or BAM1. No peptides matching BAM3 have been found in any of the BAM1, BAM2 or BAM4 purifications, nor have peptides of LSF1 been identified in any BAM2 or BAM4 purifications, suggesting that the interaction of LSF1 with BAMs is specific and exclusive for BAM1 and BAM3. The fact that BAM1 does not interact with BAM3 and vice versa suggests that at least two distinct complexes exist: one containing BAM1 and LSF1 and the other one containing BAM3 and LSF1.

4. The Interaction Network of BAM1 and LSF1 77

A Wt bam1 lsf1 250kDa

95kDa d f a d’ 72kDa b 55kDa e RbcL 3C-GST 38kDa c c

26kDa GST

Wt bam1 B 250kDa

d 95kDa a a 72kDa b 55kDa e 38kDa c

26kDa

Figure 4-9. Partial Purification of BAM1-TAP and LSF1-TAP (A) Leaves were harvested at the end of the night, immediately frozen in liquid N 2 and ground to leaf powder. TAP-tagged proteins were purified from frozen leaf powder using a two-step affinity purification. Proteins co-purifying with BAM1-TAP and LSF1-TAP were concentrated and separated on a 10% SDS-PAGE. Extracts made from wild-type leaves were subjected to the same procedure to check for proteins interacting with the affinity resins. Silver staining (left) reveals the set of proteins co- purifying with BAM1-TAP and LSF1-TAP as well as unspecific interactors from wild-type extracts. Gels were cut into sections and proteins therein identified by tryptic digestion followed by tandem mass spectrometry. Major contaminants were the large subunit of RuBisCO (RbcL), the protease used in the purification process (3C-GST) and a part of the GST (glutathione S-) tag. Gel section(s) containing a: BAM1-TAP, b: LSF1, c: p-NAD-MDH, d and d’: LSF1-TAP and a degradation product of LSF1-TAP, e: BAM1 and/or BAM3 were identified. Right: amylopectin containing native gel showing amylolytic activity of the partially purified BAM1-LSF1 complex (region f). (B) Partial purification of BAM1-TAP from bam1 and lsf1 mutant backgrounds. The amount of leaf powder extracted was reduced by ½, while all other purification conditions were as for (A). TAP-tagged BAM1 is present in both purifications (section a), while LSF1 and p-NAD-MDH are only present in the purification from the bam1 mutant background (gel sections b and c). As revealed by an amylopectin containing native gel (right), BAM1 activity is present both in its slow migrating form (i.e. as complex with LSF1, d) and in its fast migrating form (e) if purified from the bam1 mutant. BAM1 activity is only in the fast migrating form if purified from the lsf1 mutant background.

4. The Interaction Network of BAM1 and LSF1 78

Table 4-1. LSF1-TAP Interacts With BAM1 and BAM3 in Three Independent Co-Purification Experiments BAM4 - BAM1-TAP BAM2-TAP BAM3-TAP LSF1-TAP TAP myc- myc- myc- myc- TAP- TAP-tagging TAP-tagging TAP-tagging TAP-tagging CoIP CoIP CoIP CoIP tagging Unique 10 18 5 0 0 0 0 0 0 7 13 10 0 Peptides % Sequence BAM1 17 33 8.5 0 0 0 0 0 0 14 28 17 0 Coverage % of Total 0.15 0.41 0.05 0 0 0 0 0 0 0.12 0.21 0.1 0 Spectra Unique 0 0 0 2 9 10 0 0 0 0 0 0 0 Peptides % Sequence BAM2 0 0 0 3.5 19 22 0 0 0 0 0 0 0 Coverage % of Total 0 0 0 0.04 0.19 0.09 0 0 0 0 0 0 0 Spectra Unique 0 0 0 0 0 0 31 44 2 6 5 6 0 Peptides % Sequence BAM3 0 0 0 0 0 0 55 65 5.1 14 12 13 0 Coverage % of Total 0 0 0 0 0 0 1.1 1.1 0.02 0.07 0.04 0.05 0 Spectra Unique 7 17 4 0 0 0 9 25 0 14 18 17 0 Peptides % Sequence LSF1 12 21 7.1 0 0 0 16 35 0 19 20 26 0 Coverage % of Total 0.08 0.39 0.04 0 0 0 0.09 0.25 0 0.24 0.37 0.14 0 Spectra Unique 0 0 0 0 0 0 0 0 0 0 0 0 4 Peptides % Sequence BAM4 0 0 0 0 0 0 0 0 0 0 0 0 7.1 Coverage % of Total 0 0 0 0 0 0 0 0 0 0 0 0 0.083 Spectra In two two-step (TAP-tagging) and one one-step (myc-CoIP) purification TAP-tagged proteins and co- purifying proteins were isolated from leaf extracts. Co-purifying proteins were separated by SDS- PAGE and subjected to tryptic in-gel digests. Tryptic peptides were identified by tandem mass spectrometry (MS/MS). MS/MS spectra, recorded on an LTQ-Orbitrap, were analyzed using Mascot. Protein identifications were validated by Scaffold and accepted if they could be established at greater than 95.0% probability, based on at least 2 peptides. The table summarizes the co-purification of BAM and LSF1 proteins with TAP-tagged BAM and LSF1 proteins. The number of unique peptides, the sequence coverage and the percentage of total spectra (number of spectra matching a protein in relation to the total number of spectra identified from the purification) are shown for each combination. In all purifications, the tagged protein itself can be found in the elution (orange). BAM1 consistently co- purifies with LSF1-TAP and vice versa (green). BAM3 can be found to co-purify with LSF1-TAP in all experiments. In two out of three purifications, LSF1 co-purifies with BAM3-TAP (blue). No peptides matching LSF1 can be found in the BAM2-TAP or BAM4-TAP purifications or vice versa .

4. The Interaction Network of BAM1 and LSF1 79

4.2.3.4 BAM1 and LSF1 Interact with a Plastid Localized NAD-Dependent Malate-Dehydrogenase Apart from validating the BAM1-LSF1 interaction and extending the interaction network of LSF1 to BAM3, the TAP-tagging approach is suitable to detect interactions with proteins so far not implicated in starch metabolism. Given the large number of proteins contaminating the TAP-tagged protein purificantions, a comparison of all LSF1 and BAM1 interactors across three affinity purification experiments was made and only proteins interacting with LSF1 or BAM1 in all experiments were taken into consideration for possible further interaction with LSF1 and BAM1 (Fig 4-10). Besides BAM1, BAM3 and LSF1 itself, this cautious approach identifies 3 other proteins consistently interacting with LSF1. In Table 4-2, an overview of these proteins is given, including the number of peptides which have been found in other TAP purifications, if applicable. The three proteins consistently found in LSF1-TAP purifications besides LSF1, BAM1 and BAM3 were a chloroplast localized putative fructose- bisphosphate aldolase, a mitochondrial pyruvate dehydrogenase and a plastid localized NAD-dependent malate dehydrogenase (p-NAD-MDH)(Backhausen et al., 1998; Imsande et al., 2001). The three proteins consistently interacting with BAM1-TAP were BAM1, LSF1 and the same p-NAD-MDH found in LSF1- TAP purifications (compare Table 4-2). The numbers of peptides found in LSF1-TAP purifications matching to p-NAD-MDH are high. In addition, the presence of a 42kDa protein, fitting the predicted size of p-NAD-MDH, among the proteins interacting with LSF1 is obvious from silver stained gels (Fig 4-9A, band c). The fact that the p-NAD-MDH was also found to consistently interact with BAM1-TAP suggests that a trimeric complex of LSF1, BAM1 and p-NAD-MDH exists. It was interesting to see that in a separate experiment, where the BAM1-TAP tagged protein was purified from an lsf1 mutant background and not from the bam1 mutant background, no peptides matching the p-NAD-MDH could be detected (Fig 4-9B and Table 4-3). This suggests that BAM1 requires the presence of LSF1 in order to interact with p-NAD-MDH.

4. The Interaction Network of BAM1 and LSF1 80

Table 4-2. Proteins Co-Purifying with LSF1-TAP in Three Independent Experiments LSF1-TAP BAM1-TAP BAM2-TAP BAM3-TAP BAM4-TAP Accession Protein Description M myc- myc- myc- myc- w TAP-tagging TAP-tagging TAP-tagging TAP-tagging TAP-tagging Number CoIP CoIP CoIP CoIP Chloroplast localized, Unique Peptides 3 3 3 0 0 2 0 5 3 6 0 2 0 43 AT2G21330 putative Fructose- % Sequence Coverage 8 9.3 8.5 0 0 3 0 13 9 17 0 3 0 kDa Bisphosphate-Aldolase % of Total Spectra 0.03 0.03 0.03 0 0 0.02 0 0.05 0.04 0.04 0 0.02 0 Chloroplast localized, Unique Peptides 14 18 17 7 17 4 0 0 0 9 25 0 0 66 AT3G01510 putative phosphatase, % Sequence Coverage 19 20 26 12 21 7.1 0 0 0 16 35 0 0 kDa LSF1 % of Total Spectra 0.49 0.37 0.14 0.08 0.39 0.04 0 0 0 0.09 0.25 0 0 Unique Peptides 7 13 10 10 18 5 0 0 0 0 0 0 0 Chloroplast Localized β- 64 AT3G23920 % Sequence Coverage 14 28 17 17 33 8.5 0 0 0 0 0 0 0 amylase, BAM1 kDa % of Total Spectra 0.12 0.21 0.1 0.15 0.41 0.05 0 0 0 0 0 0 0 NAD-dependent Unique Peptides 15 15 10 5 13 4 0 0 3 10 15 2 0 chloroplast localized 42 AT3G47520 % Sequence Coverage 43 46 36 11 34 12 0 0 6 28 43 5 0 malate dehydrogenase, p- kDa NAD-MDH % of Total Spectra 0.47 0.35 0.12 0.1 0.2 0.07 0 0 0.04 0.09 0.13 0.03 0 Unique Peptides 6 5 6 0 0 0 0 0 0 31 44 2 0 Chloroplast Localized β- 61 AT4G17090 % Sequence Coverage 14 12 13 0 0 0 0 0 0 55 65 5.1 0 amylase, BAM3 kDa % of Total Spectra 0.07 0.04 0.05 0 0 0 0 0 0 1.1 1.1 0.02 0 Pyruvate-dehydrogenase Unique Peptides 2 2 2 0 0 0 0 0 3 4 4 0 0 activity. Located in 39 AT5G50850 % Sequence Coverage 5.2 5.2 6 0 0 0 0 0 10 14 14 0 0 mitochondrion, nucleolus, kDa plasma membrane. % of Total Spectra 0.02 0.02 0.02 0 0 0 0 0 0.04 0.04 0.03 0 0 Proteins co -purifying with TAP -tagged proteins were separated by S DS -PAGE, subjected to tryptic in -gel digests and tryptic peptides were identified by tandem mass spectrometry (MS/MS). Protein identifications were accepted if they could be established at greater than 95.0% pr obability, based on at least 2 peptides. This table lists all proteins consistently co-purifying with LSF1-TAP in two independent two-step purifications as well as one one-step purification, after unspecific interactors have been subtracted. The numbers of unique peptides identified by MS/MS, the perc ent sequence coverage as well as the percent of total spectra (number of spectra matching a protein in relation to the total number of spectra identified from the purific ation) for a given protein are indicated.

4. The Interaction Network of BAM1 and LSF1 81

Figure 4-10. Proteins Interacting with LSF1-TAP and BAM1-TAP in Three Experiments After subtraction of unspecific interactors, the lists of proteins interacting with LSF1-TAP and BAM1- TAP were compared from three independent purification experiments. Venn-diagrams illustrate the number of proteins identified in each of the three experiments. Six proteins co-purify with LSF1-TAP in all purifications: LSF1, BAM1, BAM3, a plastid localized malate dehydrogenase (p-NAD-MDH), a fructose-bisphosphate-aldolase and a pyruvate-dehydrogenase. Three proteins consistently co-purify with BAM1-TAP: BAM1, LSF1 and the p-NAD-MDH.

Table 4-3. The BAM1-MDH Interaction Depends on LSF1 BAM1-TAP ( bam1 ) BAM1-TAP ( lsf1 ) Unique Peptides 18 16 BAM1 % Sequence Coverage 33 32 % of Total Spectra 0.41 0.39 Unique Peptides 17 0 LSF1 % Sequence Coverage 21 0 % of Total Spectra 0.39 0 Unique Peptides 13 0 p-NAD-MDH % Sequence Coverage 34 0 % of Total Spectra 0.2 0 Proteins co-purifying with TAP-tagged BAM1 purified from the bam1 or lsf1 mutant background were separated by SDS-PAGE, subjected to tryptic in-gel digests and tryptic peptides were identified by tandem mass spectrometry (MS/MS). The number of unique peptides identified by MS/MS, the percent sequence coverage as well as the percent of total spectra for BAM1, LSF1 and p- NAD-MDH are indicated. LSF1 and p-NAD-MDH were co-purifying with BAM1-TAP, when BAM1-TAP was purified from the bam1 mutant background and missing when BAM1-TAP was purified from the lsf1 mutant background.

4. The Interaction Network of BAM1 and LSF1 82

4.2.4 LSF1 lacks an NAD-dependent MDH activity

To corroborate the idea that LSF1 and BAM1 (see Table 4-2) could interact with an MDH, activity assays based on native gels were conducted. Native gels were used to determine the activities of different isoforms of malate dehydrogenases in crude leaf extracts of wild type, bam1, bam3 and lsf1 mutants. After electrophoresis, the MDH activity was assayed based on the detection of reduction of NAD + (and NADP +) concurrent with the oxidation of malate supplied in the native gel incubation medium.

When in the lab of Dr Jychian Chen (Academia Sinica, Taipei) glycogen containing native gels were incubated in a medium containing NAD + and malate, two bands of MDH activity were missing in the lsf1 mutant as compared to wild type (Fig 4-11A, region a). When the gels were incubated with NADP +, no difference between the wild-type and lsf1 was visible (not shown). It was thus concluded that in lsf1 mutants, a strictly NAD + dependent MDH activity was missing. The NAD-MDH activity missing in lsf1 was a minor activity as compared to the total NAD-MDH activity detectable on native gels. The major MDH activity (Figure 4-11A, band b) appeared unchanged in all mutants investigated. On the glycogen containing native gels, no changes in MDH activities were visible in the bam1 mutant (Figure 4-11A). When native gels containing amylopectin were run and incubated to reveal amylolytic and MDH activity in parallel, two bands of activity, a slow migrating and a fast migrating one, were missing in lsf1 (Figure 4-11B, right picture, bands a and b). The slow migrating band co-localized with the BAM1-LSF1 complex, as revealed by its amylolytic activity in the Lugol stained amylopectin gel (Figure 4-11B, compare band b in the ‘left’ and ‘right’ picture). This slow migrating MDH activity band was also missing in the bam1 mutant. The faster migrating activity band did not co-localize with the BAM1-LSF1 complex and was unaffected in bam1 . The MDH activity co-localizing with the BAM1-LSF1 complex appeared weaker than the faster migrating band. Similar to the glycogen containing gels, the bulk of MDH activity was unaffected by lsf1 and bam1 mutations (Figure 4-11B, band c). No change could be observed in the bam3 mutant as compared to the wild type.

These data corroborate the interaction of LSF with an NAD-MDH, they also support the previous finding that BAM1 interacts with NAD-MDH as well. The native gels do not clarify whether the NAD-MDH activities missing in lsf1 and bam1 mutants are due to altered electrophoretic mobility, protein stability or enzymatic activity of the NAD-MDH. Only a minor fraction of the total NAD-MDH activity detectable on native gels is affected by the lsf1 and bam1 mutations. However, as the gels have been performed using total leaf extracts, it remains possible that the NAD-MDH changes would be more pronounced if only chloroplast extracts were used.

4. The Interaction Network of BAM1 and LSF1 83

A B Wt bam1 lsf1 sex4 Wt lsf1 bam1 bam3 Wt lsf1 bam1 bam3 a a b

c b

Figure 4-11. An NAD-dependent Malate Dehydrogenase Activity is Missing in lsf1 and bam1 Mutant Extracts. Soluble protein extracts from leaves of wild-type and mutant plants were made and investigated for alterations in NAD +/NADP + dependent malate dehydrogenase (MDH) activity. (A) Glycogen-containing native gel, to visualise NAD-dependent MDH activity. Two bands of MDH activity are missing from the lsf1 mutant (region a), while a faster migrating, strong activity is comparable between wild-type and mutant extracts (b). No obvious differences between the bam1 mutant, the sex4 mutant and the wild type are visible (courtesy of Dr Jychain Chen, Academia Sinica, Taipei). (B) Amylopectin-containing gels, incubated and stained for activity of amylolytic enzymes (left) or assayed for MDH activity (right). Two bands of activity are missing from the lsf1 mutant (a, b). The stronger of the two does not co-localize with the BAM1-LSF1 amylolytic activity and is present in the bam1 mutant (b). The weaker band which co-localizes with the BAM1-LSF1 complex amylolytic activity is missing in lsf1 and bam1 mutants but not in bam3 mutants (a). The strongest band of MDH activity appears unchanged between the wild-type and the different mutant lines (c).

4.2.5 Functional Characterization of the BAM1-LSF1 Complex

Besides identifying, confirming and extending the information on an interaction of BAM1 and LSF1, I attempted to functionally characterize the BAM1-LSF1 complex. Different hypotheses can be put forward regarding a potential function of the BAM1-LSF1 interaction. 1) LSF1 might act as phospho-glucan phosphatase to de-phosphorylate glucan chains, as β-amylolysis by BAM1 (and BAM3) is in progress. Analyses of the lsf1 mutant phenotype suggest that this is unlikely but cannot fully exclude the possibility. 2) LSF1 has a CBM and can bind to starch granules. Therefore, it might help to target BAM1 (and BAM3) to the starch granule. 3) LSF1 might serve the function of scaffolding protein interactions important in a so far unrecognized process relevant to starch degradation. Furthermore, it remains possible that LSF1 acts as a protein phosphatase. A number of exploratory experiments were conducted to address the question of the relevance and function of the BAM1-LSF1 interaction.

4.2.5.1 BAM1 and LSF1 Interact in the Chloroplast In the lab of Dr A. Smith (‘John Innes Center’, Norwich, UK), Dr A. Graf has applied a split YFP approach to confirm and localize the BAM1-LSF1 interaction in planta. The N- and C-terminal parts of YFP were fused in frame to the c-terminal ends of LSF1 and BAM1 and transiently expressed in 4. The Interaction Network of BAM1 and LSF1 84

Nicotiana benthamiana epidermal cells by Agrobacterium mediated co-expression (Fig 4-12). The YFP signal resulting from the reconstitution of the YFP protein following tight local interaction of BAM1 and LSF1 co-localised with the chlorophyll autofluorescence, suggesting direct or close contact of BAM1 and LSF1 in the chloroplast. Controls for homo-multimerisation of BAM1 or LSF1 or for self-assembly of the YFP were negative (see Appendix I). To corroborate the chloroplast localization of the BAM1-LSF1 interaction, Arabidopsis wild-type protoplasts were isolated and separated into chloroplast and cytosol fractions (Fig 4-13). The contamination of the chloroplast fraction with cytosolic enzymes was quantified measuring phosphenolpyruvate-carboxylase (PEP-carboxylase, a cytosolic marker enzyme, Fig4-13A). The chloroplast contamination of the cytosol fraction was quantified by measuring NADP-dependent glyceraldehyde 3-phosphate dehydrogenase (NADP-GAPDH, a chloroplast marker enzyme; Fig4-13A). Amylopectin containing native gels (Fig4-13B) were loaded based on equal chloroplast marker or equal cytosolic marker enzyme activities. Activities mainly present in the chloroplast should be equally present between cytosolic and chloroplast fractions when loaded based on equal chloroplast marker enzyme activity and chloroplastic proteins should be highly abundant in the chloroplast fraction when loaded based on equal cytosolic marker enzyme activity. The BAM1-LSF1 complex activity on native gels followed this distribution pattern, as did the starch debranching enzyme complex, ISA1/2, a known chloroplast enzyme (Zeeman et al., 1998c; Delatte et al., 2005). This biochemical approach thus confirms the BAM1-LSF1 interaction in the chloroplast.

4. The Interaction Network of BAM1 and LSF1 85

LSF1-YC BAM1-YC LSF1-YC BAM1-YC LSF1-YN BAM1-YN BAM1-YN LSF1-YN YFP Chlorophyll Overlay

Figure 4-12. Bimolecular Fluorescence Complementation Shows BAM1-LSF1 Interaction in the Chloroplast Bimolecular fluorescence complementation to localize the BAM-LSF1 interaction was performed by Dr A. Graf (‘John Innes Center’, Norwich, UK). The N- and C-terminal parts of YFP were fused in frame to the c-terminal ends of LSF1 (LSF1-YC and LSF1-YN) and BAM1 (BAM1-YC and BAM1-YN) proteins, under the control of a 2x 35S promoter. Transient co-expression of the fusion proteins in N. benthamiana epidermal cells was achieved by co-infiltration of epidermal cells with Agrobacterium strains transformed with the constructs. Close physical proximity of the tagged proteins allows for reconstitution of a functional YFP fluorophore and consequently, the detection of fluorescence using confocal microscopic analysis. When the N- or C-terminal parts of YFP were fused to BAM1 and LSF1 respectively the YFP signal could be detected. Whenever a YFP signal could be detected, the signal co-localized with the autofluorescence from the chloroplasts. Low levels of homo-multimerization could be observed if BAM1- YC and BAM1-YN were co-transformed.

4. The Interaction Network of BAM1 and LSF1 86

B Equal Chloroplast Equal Cytosol A Marker Marker chloro cyto chloro cyto 0.16 0.012 0.14 0.01 0.12 0.008 0.1 0.08 0.006

PEP-C PEP-C (AU) 0.06 0.004 0.04 NADP-GAPDH (AU) NADP-GAPDH

0.002

0.02

0 0

Figure 4-13. The BAM1-LSF1 Complex is Localized in the Chloroplast Wild-type Arabidopsis mesophyll protoplasts were isolated and separated into chloroplast- and cytosol- enriched fractions. (A) NADP-glyceraldehyde 3-phosphate dehydrogenase (NADP-GAPDH) and phosphoenolpyruvate- carboxylase (PEP-C) activity were measured as chloroplast and cytosol marker activities respectively, in both fractions (see Section 2.6 for details). (B) Equal chloroplast and equal cytosol marker activities from both fractions were assayed for amylolytic activity on amylopectin containing native gels. The BAM1-LSF1 complex activity was present mainly in the chloroplast fraction (grey arrow-head), like the known chloroplast enzyme activity of the ISA1/2 debranching enzyme (black arrow-head).

4.2.5.2 Redox Dependence of the BAM1-LSF1 Complex BAM1 is a redox activated protein (Sparla et al., 2006). To investigate the redox dependency of the BAM1-LSF1 interaction, soluble proteins were extracted from leaves harvested at the end of the night and end of the day. The extraction medium either included 1mM DTT as reducing agent or did not. When amylolytic activities were analysed using native gels containing amylopectin, the presence of a reducing agent in the extraction medium stimulated some enzyme activities (Fig 4-14A). Differences in enzyme activities between the two time-points of harvest were not visible. Among the activities strongly dependant on DTT were the ISA1/2 DBE activity (Fig 4-14A band a), the BAM1-LSF1 activity (Fig 4- 14A band b) and another as-yet unidentified activity (Fig 4-14A band c). To check whether the formation of the BAM1-LSF1 complex is affected by the redox-state, wild-type and lsf1 plant extracts, made with

1mM DTT, were oxidised using 300µM CuCl 2 and subsequently re-reduced by adding 10mM DTT (Fig 4-14B). Both BAM1 activities (Fig 4-14B, bands b and b’ respectively) were diminished in the oxidised state and could be re-established upon re-reduction. The ISA1/2 activity showed a similar trend (Fig 4- 14B, band a), while the BAM5 activity at the very bottom of the native gels was unaffected by oxidation and re-reduction. When identical native gels run in parallel, were blotted and probed with an anti-BAM1 4. The Interaction Network of BAM1 and LSF1 87

antibody, it appeared that in oxidised samples, BAM1was no longer in complex with LSF1. Upon re- reduction, the slow-migrating form of BAM1 was restored, suggesting that BAM1-LSF1 complex formation was re-established (Fig 4-14B, band b). I investigated whether there was any change in the BAM1-LSF1 complex during the day/night cycle. In extracts made with an extraction medium containing 0.5mM DTT, no differences in BAM1 activities or presence of the BAM1 protein in the BAM1-LSF1 complex were discernable (Fig 4-14C, bands b and b’).

Figure 4-14. The BAM1-LSF1 Complex Formation is Redox Dependent and does not Change in the Diurnal Cycle (A) Total soluble protein extracts were made from leaves harvested at the end of the night (EON) or at the end of the day (EOD) with (+) or without (-) 1mM DTT in the extraction buffer. Protein extracts were analysed using amylopectin containing native gels to reveal the different amylolytic activities. DTT in the extraction buffer strongly enhanced the ISA1/2 DBE activity (a), the BAM1-LSF1 activity (b) and another, slightly faster migrating activity (c). No differences were visible between samples extracted with the same extraction buffer at EOD or EON. (B) Native gel containing amylopectin (left) and native gel western blot probed for BAM1 (right) to investigate the dependence of BAM1-LSF1 complex formation on redox conditions. Leaf extracts made in 1mM DTT containing buffer were oxidized for 30 min with 300µM CuCl 2 and subsequently re-reduced for 30min with 10mMDTT. 40µg of total protein are loaded in each lane. Fast- and slow-migrating BAM1 activities (b and b’ respectively) are reduced in oxidized samples. BAM1 is no longer present as slow-migrating form. The migration and activity pattern of BAM1 is restored upon re-reduction. (C) Protein extracts from leaves harvested at different time points in the diurnal cycle, extracted in buffer containing 0.5mM DTT. No major changes in BAM1-LSF1 (b) as well as BAM1 (b’) activity or protein amounts were detectable on either activity native gels (left) or blotted native gels probed with the BAM1 antibody (right). 4. The Interaction Network of BAM1 and LSF1 88

4.2.5.3 BAM1-LSF1 Cannot Dephosphorylate Phosphorylated Oligosaccharides. The interaction between a β-amylase and a glucan-phosphatase could be beneficial for a closely correlated glucan de-phosphorylation and β-amylolysis, since de-phosphorylation is crucial for complete β- amylolysis of a glucan chain. I tested whether the BAM1-LSF1 complex could de-phosphorylate and degrade phosphorylated oligosaccharides (P-oligos). P-oligos were supplied by Dr O. Kötting in our lab. In brief, they were purified from potato amylopectin, which had been extensively degraded by BAMs, AMYs and DBEs. P-oligos were separated from their non-phosphorylated counterparts by anion- exchange chromatography and consisted of linear glucan chains, with the majority of chains ranging from DP 6 to 15, as revealed by HPAEC-PAD (Fig 4-15, inset). Continued β-amylolytic degradation of the P- oligos is blocked by phosphorylated glucose resides on the non-reducing ends of the glucan chains.

To test for glucan-phosphatase activity, the partially purified BAM1-LSF1 complex (from the BAM1- TAP purification; BAM1-TAP) and recombinant BAM1 (BAM1(H), expressed and purified from E.coli by Dr H. Reinhold in our lab), were incubated with P-oligos. Only dephosphorylation of the P-oligo would allow β-amylolysis and thus release of maltose. When P-oligos were incubated with the BAM1- TAP elution or the heterologously expressed BAM1, no release of maltose could be detected by HPAEC- PAD (Fig 4-15). Only when recombinant SEX4 was added in addition to the BAM1-TAP or BAM1(H), were the P-oligos degraded. The glucan chains of the P-oligos were shortened and a clear accumulation of maltose could be detected (Fig 4-15). Thus, in this experiment, no glucan phosphatase activity of the BAM1-LSF1 complex could be detected. However, it can not be completely excluded that very low levels of de-phosphorylation are possible and had not been detected by the aforementioned assay.

4. The Interaction Network of BAM1 and LSF1 89

Figure 4-15. P-Oligo Digestion by the BAM1-LSF1 Complex Phosphorylated oligosaccharides (P-oligos) were produced and purified by Dr O. Kötting, as described in Section 2.12. P-oligos consist of linear glucan chains, the majority of chains is of DP6 to DP16 (inset, P- oligo+SEX4). To test the BAM1-LSF1 complex for glucan phosphatase activity, P-oligos were incubated with partially purified BAM1-TAP, containing the co-purifying LSF1 (dark and light blue). As negative control, recombinant BAM1 (BAM1(H)) was used (dark and light red). As positive control for de- phosphorylation, recombinant SEX4 was added in parallel with the β-amylases (light blue and light red). P-oligos were incubated with the enzyme combinations for 3hrs at 37°C, boiled to inactivate β-amylases and dephosphorylated prior to HPAEC-PAD analysis to report the chromatograms shown. BAM1-TAP and BAM1(H) could not release maltose (G2) from the P-oligos. Only when recombinant SEX4 was added in addition to BAM1-TAP or BAM1(H), maltose was released.

4.2.5.4 Identification of BAM1 Phospho-Peptides Based on sequence comparison, the glucan phosphatase SEX4 and its homologue LSF1 have been suggested to be potential protein phosphatases (Niittylä et al., 2006). Even though no protein substrate has been identified and SEX4 has been shown to act as a glucan phosphatase (Comparot-Moss et al., 2010), the possibility that LSF1 is a protein phosphatase remains. This is an interesting possibility because BAM1 and BAM3 have both been identified as phospho-proteins in large-scale proteomic screens (Heazlewood et al., 2008; Lohrig et al., 2009; Durek et al., 2010). Even though the tandem-affinity- purifications I have done were not specifically enriched for phospho-peptides, phospho-peptides have been identified for BAM1 (Table 4-4). The exact phosphorylation site has only been discernible for one of the peptides and is consistent with that published in the PhosphAt database (http://phosphat.mpimp- golm.mpg.de, (Heazlewood et al., 2008)). Comparable numbers of phosphorylated peptides matching 4. The Interaction Network of BAM1 and LSF1 90

BAM1 have been found irrespective of whether BAM1-TAP was purified from the bam1 or lsf1 mutant background. No phospho-peptides have been identified in any of the BAM3-TAP purifications. These findings do not exclude the possibility that LSF1 may work as a protein phosphatase but they corroborate the finding that BAM1 is a phospho-protein and suggest that the presence of LSF1 does not negatively influence the phosphorylation status of BAM1.

Table 4-4. Identification of Phosphorylated Peptides of BAM1

Peptide Sequence

BAM1-TAP AHGTDP(s)PPM(s)PILGA(t)R (bam1 ) SGEMTDSSLLSI(pS)PPSAR BAM1-TAP AHGTDP(s)PPM(s)PILGATR (lsf1 ) SGEMTDSSLLSI(pS)PPSAR SGEMTDSSLLSI(pS)PPSAR PhosphAt GEGGAAADHHSEAS(pS)PSSLSAGAGNGAK Database AHGTDP(pS)PPM(pS)PILGATR Peptides obtained from tryptic digests of gel regions containing BAM1 -TAP purified form the bam1 or lsf1 background were analyzed by MS/MS. Spectra were searched allowing for phosphorylation of Ser (pS), Thr (pT) and Tyr (pY) as variable modifications. For peptides highlighted in orange, the phosphorylation site could be determined according to Reiland et al., (2009). For the other peptide, the phosphorylation site could not be determined unambiguously. Amino acids where the phosphorylation site is ambiguous are enclosed in brackets and are in lower case. The PhosPhAt database (http://phosphat.mpimp-golm.mpg.de; Heazlewood et al. (2008)) was searched for phosphorylated peptides from BAM1 and these are shown with the phosphorylation sites indicated.

4.2.5.5 LSF1 Increases the β-Amylolytic Activity of BAM1-TAP A possible function of LSF1 in the BAM1-LSF1 complex could be to influence the substrate binding properties of BAM1 and thereby enhance the efficiency of BAM1 (potentially also BAM3) in degrading starch. The effect of LSF1, interacting with BAM1, on the β-amylolytic activity of BAM1 can be investigated by purifying the TAP-tagged BAM1 from the bam1 mutant and lsf1 mutant background. While most of the BAM1-TAP in the bam1 mutant is in complex with LSF1 (compare Fig4-9A, band f or Fig 4-9B, band d), all BAM1 is in its monomeric state when purified from the lsf1 mutant. Due to the contamination of each preparation with other proteins, a comparison of BAM1-TAP activities based on the total protein levels was inappropriate. Instead, the levels of BAM1-TAP tagged protein were relatively quantified in each of the two samples by separation of proteins using SDS-PAGE and immunodetection 4. The Interaction Network of BAM1 and LSF1 91

of BAM1-TAP on western blots. The intensities of the bands on western blots were quantified from different loadings, to ensure linearity of immunodetection of BAM1-TAP (Fig 4-16A). It was thus possible to compare the amount of BAM1 present in the two samples. The proportions of BAM1 in complex with LSF1 or present as monomer were not quantified in the BAM1-TAP purified from the bam1 mutant in this experiment. However, based on previous purifications, most of the activity was likely in the complexed form. LSF1 had a stimulatory effect on BAM1-TAP activity against amylopectin. In two independent purification experiments, the BAM1-TAP purified from the bam1 background significantly outperformed the monomeric BAM1-TAP purified from the lsf1 background in terms of maltose release from amylopectin (Fig 4-16B).

A

3000 2760 2000 1175 1513

AU 672 1000 291 607 0 BAM1-TAP BAM1-TAP BAM1-TAP BAM1-TAP BAM1-TAP BAM1-TAP (bam1 ) (lsf1 ) (bam1 ) (lsf1 ) (bam1 ) (lsf1 ) Loading 5µL 10µL 20µL

B 2.5 6

2 5 4 1.5 3 1 2

0.5 1 µM Maltose/ProteinµM Unit µM Maltose/Protein µM Unit 0 0 0 60 120 0 60 120 Minutes Minutes

Figure 4-16. Degradation of Amylopectin by BAM1 and the BAM1-LSF1 Complex (A) BAM1-TAP purified from bam1 and lsf1 mutant backgrounds in two replicate experiments was relatively quantified by SDS-PAGE, western blotting and immunodetection of the myc-epitope (top panel, only results for one experiment are shown). Band intensities were quantified using ImageJ and are expressed as arbitrary units (AU; bottom panel). Different loadings of BAM1-TAP confirm linearity of immunodetection. (B) Partially purified BAM1-TAP from the bam1 (black diamonds) or the lsf1 mutant background (grey squares) was incubated with amylopectin and the release of maltose was measured over time using HPAEC-PAD. Maltose release was adjusted to equal BAM1-TAP protein input, as determined in (A). In two independent purifications (left and right panel), the release of maltose was consistently higher when BAM1-TAP was purified from the bam1 mutant background than from the lsf1 mutant background. Each value represents the mean of three replicate measurements ±SE. 4. The Interaction Network of BAM1 and LSF1 92

4.2.6 Characterization of bam1 /lsf1 , bam3 /lsf1 and bam1 /bam3 /lsf1 Multiple Mutants

Attempts to characterize the function of the BAM1-LSF1 complex have provided evidence for its chloroplast localization and its sensitivity to oxidizing environments. However, no glucan-phosphatase activity could be shown for the BAM1-LSF1 complex, while a stimulating effect of LSF1 on the β- amylolytic activity of BAM1 could be observed in preliminary experiments. However, although these comparisons of activity between BAM1 and the BAM1-LSF1 complex used the proteins and complexes purified from plants, the degradation of solubilized amylopectin in isolation from other activities required for the degradation of starch represents an artificial situation. Such assays may fail to identify indirect effects of LSF1 on starch degradation. One way to look at the in planta actions of BAM1, LSF1, BAM3 and their direct or indirect interactions is to investigate the single and multiple mutant lines lacking these proteins.

To investigate whether the lsf1 phenotype might be due to impaired β-amylase activity, total soluble β- amylase activity was measured in the bam1 , bam3 and lsf1 single mutants as well as the bam1 /3 double mutant (Fig 4-17). Previous studies have revealed that bam1 mutants have significantly lower β-amylase activity than wild type (Fulton et al., 2008). In my analyses, the mean β-amylase activity was lower in bam1 than in wild type but the difference was not statistically significant, probably due to the relatively large standard error. In the bam3 and the bam1 /bam3 double mutant, the β-amylase activity was significantly reduced. Interestingly, even though the additional loss of BAM1 in a bam3 mutant background causes an increase in the sex phenotype, this is not reflected by an additional loss in the total soluble β-amylase activity. This discrepancy could be explained by a variation in BAM5 activity. BAM5 is a phloem localized β-amylase (Wang et al., 1995) and has been shown to be highly variable between mutants affected in starch metabolism (Caspar et al., 1989; Monroe and Preiss, 1990)

The starch levels of the single bam1 , bam3 and lsf1 mutants have already been described (Fulton et al., 2008; Comparot-Moss et al., 2010). The double mutants bam1 /lsf1 and bam3 /lsf1 were made by crossing the respective single mutants and screening the segregating F2 population using PCR-based screens published by Fulton et al. (2008) and Comparot-Moss et al. (2010). The triple mutant bam1 /bam3 /lsf1 was screened from a cross between the double bam1 /bam3 (Fulton et al., 2008) and lsf1 mutants. Starch was extracted from plants grown in a 12-h day/12-h night regime, which were harvested at the end of the day, two and four hours into the night and at the end of the night (Fig 4-18A and B). In contrast to bam1 mutants, the bam3 and lsf1 single mutants had sex phenotypes as compared to the wild type. The starch levels were similar in the bam3 and lsf1 single mutants. These data are consistent with previously published findings (Fulton et al., 2008; Comparot-Moss et al., 2010). The bam1 /lsf1 double mutant starch content was not statistically significant from the lsf1 single mutant. Starch levels in the bam3 /lsf1 double 4. The Interaction Network of BAM1 and LSF1 93

mutant, however, were significantly increased as compared to the lsf1 single mutant. Even though the difference to the bam3 or lsf1 single mutants was small, the difference could be corroborated by looking at starch levels of the mutants during starch degradation. For all time-points measured, the bam3 /lsf1 mutant had increased starch levels as compared to the single mutants (Fig 4-18B). As stated already earlier, the additional loss of BAM1 in a bam3 mutant background results in a highly increased sex phenotype compared to the bam3 single mutant, pointing out the partial redundancy of the two enzymes. Interestingly, the triple mutant, lsf1 /bam1 /bam3 had no increased sex phenotype as compared to the bam1 /bam3 double mutant (Fig 4-18A).

Maltose, generated by β-amylases is the major starch degradation product. Deficiency in β-amylolytic attack of the starch granules should therefore be reflected in the amount of maltose released. As maltose is metabolized concomitantly with its production, maltose levels measured represent steady state levels and have to be interpreted carefully. When maltose levels were analyzed from plants harvested four hours into the night – a time when starch degradation is in full progress - a significant reduction could be observed in the lsf1 and bam3 mutants. The maltose levels in the bam1 mutant were not significantly different from the wild type. Maltose levels in the double mutants bam1/bam3 , bam1 /lsf1 and bam3 /lsf1 were significantly reduced as compared to the wild type (Fig 4-19).

1.6

1.4

1.2

protein 1 -1

µg **

-1 0.8 hr 0.6 410nm 410nm

∆OD 0.4

0.2

0 Wt bam1 bam3 bam1/ lsf1 bam3

Figure 4-17 β-Amylase Activity in bam and lsf1 Mutant Plants. Total soluble β-amylase activity in crude extracts from leaves of the wild type, the bam1 , bam3 and lsf1 as well as the bam1/bam3 mutant was determined using the betamyl assay kit from Megazyme. Values are means of four replicate samples ± SE. The bam3 and the bam1/bam3 mutants display significantly lower β-amylase activities than the wild type, while total β-amylase activity is not significantly different in the lsf1 or bam1 mutants as compared to the wild type (t-test; p<0.01).

4. The Interaction Network of BAM1 and LSF1 94

A B 50 50 End of Night WT bam1 End of Day bam3 lsf1 bam1/bam3 bam1/lsf1 40 40 bam3/lsf1 bam3/lsf1

FW) d d FW) -1 -1

30 30

20 c 20 Starch Contents Starch Contents b b b (mg Glc. Equivalents g (mg Glc.10 Equivalents g 10

a a 0 0 Wt bam1 bam3 lsf1 bam1/ bam3/ bam1/ bam1/bam3/ lsf1 lsf1 bam3 lsf1 0 2 4 6 8 10 12 Hours into the Night

Figure 4-18. Starch Levels in bam1 , bam3 and lsf1 Single and Multiple Mutant Plants. Whole rosettes of 3-4 week-old plants were harvested at the end of the day, 2 or 4 hours into the night and at the end of the night and immediately frozen in liquid N 2. Starch was extracted in perchloric acid, enzymatically hydrolyzed to glucose and quantified. Starch was extracted from leaves of single bam1 , bam3 and lsf1 mutants, the bam1/lsf1 and bam3/lsf1 double as well as the bam1/bam3/lsf1 triple mutant. (A) Starch levels in the rosettes of all mutant combinations at the end of the day and at the end of the night. Values are means of 5 replicate samples ±SE. Starch levels in the bam1 mutant are not significantly different from the wild type. Starch levels in lsf1, bam3 and the bam1/lsf1 double mutant at the end of the night are significantly increased as compared to the wild type but not significantly different from each other. Starch levels in the bam3 /lsf1 double mutant are significantly higher than in lsf1 or bam3 mutants. Starch levels in bam1/bam3 and bam1/bam3/lsf1 are significantly increased as compared to all other mutants but not significantly different from each other. Significant differences are indicated by different letters (t-tests with p<0.01). (B) Starch levels of wild type, bam1 , bam3 , lsf1 , bam1/bam3 , bam1/lsf1 and bam3/lsf1 at different time points into the night. Values are means of 5 replicate samples ±SE.

4. The Interaction Network of BAM1 and LSF1 95

40

30 FW -1

20 µg Maltose g 10

0 Wt bam1bam3 lsf1 bam1/ bam1/ bam3/ bam3 lsf1 lsf1

Figure 4-19. Maltose Levels in bam and lsf1 Single and Multiple Mutants. Soluble sugars were extracted from plants harvested 4 hours into the night. Plants were the same as in Fig 4-18 and soluble sugars were extracted in parallel with starch. Maltose levels were measured using HPAEC-PAD. Maltose levels in bam1 mutants were not significantly different as compared to the wild type. Maltose levels in bam3, lsf1, bam1/bam3, bam1/lsf1 and bam3/lsf1 mutants were significantly reduced as compared to wild-type plants. Values are means of four replicates, ±SE (t-test; p<0.01).

4.3 Discussion

4.3.1 BAM1 and LSF1 Are Present in a Common Complex

The primary goal of the experiments presented above was to identify the interaction partners of BAM1 and LSF1 and thereby specify their functions in starch degradation. In a number of different experiments I provide compelling evidence that BAM1 and LSF1 interact with each other. The fact that lsf1 mutants lack one of the two BAM1 activities which can be observed on native gels, suggested that BAM1 and LSF1 interact with each other, although the altered mobility might also have been an indirect or pleiotropic effect of the lsf1 mutation. A variety of techniques, including co-immunoprecipitation, TAP- tagging and bimolecular fluorescence complementation (BiFc) allow me to unequivocally conclude that BAM1 and LSF1 are present in a common complex in the chloroplast. In addition, the BiFc approach suggests that BAM1 and LSF1 must be in very close proximity to each other in this complex such that the YFP fluorophore can be reconstituted. It has been estimated that fluorescence complementation can occur if the fluorophore fragments are fused to protein fragments approximately 10nm apart, provided there is enough flexibility to reconstruct the YFP fragments (Hu et al., 2002; Bhat et al., 2006). This means that even though BiFC favors the detection of direct interactions, it remains an open question as to whether BAM1 and LSF1 interact directly or only via a third party protein (e.g.: p-NAD-MDH). Questions about a

4. The Interaction Network of BAM1 and LSF1 96

direct interaction of BAM1 and LSF1 and about the topology of the BAM1-LSF1 complex are difficult to answer using the in vivo approach presented in this work.

Looking for BAM1-LSF1 interaction in a yeast-2-hybrid approach would be a suitable approach to answer the question. In addition, experiments using heterologously expressed proteins for in vitro mixing and binding experiments as well as potential crystallization studies would be suited to study the complex at an ultrastructural level. Preliminary experiments in which heterologously expressed and partially purified BAM1 and LSF1 proteins have been mixed and run on acrylamide containing native gels have already been conducted by Dr Oliver Kötting in our laboratories. These experiments have so far not been successful and the BAM1-LSF1 complex has not yet been reconstituted in vitro . Many reasons for this could be imagined, including the fact that BAM1 and LSF1 might not interact directly. The NMW of the BAM1-LSF1 complex was estimated to be around 150kDa using GFC. This suggests that apart from BAM1 and LSF1 other proteins or another protein might be present in the complex. The TAP-tagging approach has been very useful in the identification of the p-NAD-MDH as a third complex partner in addition to BAM1 and LSF1. The absence of p-NAD-MDH in the in vitro mixing experiments might have been the reason for failure to reconstitute the BAM1-LSF1 complex in vitro .

With the rough size estimation obtained by the GFC, conclusions about the stoichiometry of the complex can only be speculative. With the approximate monomeric observed weights of BAM1 (60kDa), LSF1 (65kDa) and p-NAD-MDH (36kDa), a combination of one BAM1 protein, one LSF1 protein and one p- NAD-MDH proteins could be imagined. However, given that all mammalian and plant NAD-MDH enzymes characterized so far are present in dimers, with subunit masses of 32.5-37kDa (Ocheretina and Scheibe, 1997), it may be that two MDH subunits are present in the BAM-LSF1 complex.

The apparent dependency of the BAM1-LSF1 complex on reducing conditions offers the potential for a regulation of complex formation. My data suggest that oxidation of the plant extract abolishes complex formation. The mechanism for this is unclear but formation of the described internal cysteine bridge between Cys-32 and Cys-470 (Sparla et al., 2006), upon oxidation of BAM1, might lead to a conformational change that both, renders it inactive and disrupts the complex with LSF1. The influence of redox state on complex formation makes an in vivo characterization of complex formation dynamics (i.e.: over time) difficult, as the in vivo redox state of the environment of the BAM1-LSF1 complex would have to be conserved. A preliminary attempt to look at the complex dynamics over the diurnal cycle using low amounts of DTT in the extraction buffer has failed to identify any dynamic changes. Rapid in vivo crosslinking of proteins might help to conserve the BAM1-LSF complex state. Crosslinking might abolish amylolytic activity but the complex status of BAM1 (and LSF1) could still be assessed by western blotting.

4. The Interaction Network of BAM1 and LSF1 97

4.3.2 LSF1 Interacts With BAM3

As the single bam1 mutant does not display a deficiency in starch degradation, the sex phenotype of lsf1 could not be explained solely through its interaction with BAM1. Finding BAM3 as a consistent interaction partner of LSF1 in the TAP-tagging experiments thus added weight on the importance of an LSF1-BAM interaction as bam3 mutants display a sex phenotype. The fact that LSF1 could also be found to interact with the TAP-tagged BAM3 protein adds confidence to the LSF1-BAM3 interaction. It should be noted here, however, that a TAP-tag independent verification of the BAM3-LSF1 interaction is still missing and required. In fact, there is evidence arguing against an interaction of LSF1 and BAM3. Firstly, the NMW of BAM3 has been found to be in the LMW range, corresponding well with the predicted monomeric weight of BAM3 (see Fig 3-2 D). This was not changed even if bam1 mutant extracts were used in the GFC separation, where BAM3 would not be competing with BAM1 for interaction with LSF1 (not shown). Another problem relates to the fact that BAM3 activity has so far not been detectable using native gel approaches, despite BAM3 being an active β-amylase and total β-amylase activity being reduced in bam3 mutants (Fig 4-17). This hampers attempts to determine the NMW of BAM3 and observe possible changes in relation to LSF1 (or BAM1) levels. Easy disruption of the putative LSF1- BAM3 complex during electrophoretic- and GFC-based separations might explain the failure to detect BAM3 on native gels and failure to observe it as a HMW complex. The LSF1 BAM3 interaction thus awaits confirmation and experiments comparing the isolated BAM3 activity with the LSF1-BAM3 activity on amylopectin (as done here for BAM1) might be required. Better still, assays on phosphorylated starch granules might help to confirm the significance of the interactions of BAM1 and BAM3 with LSF1.

4.3.3 A Plastid Localized NAD-dependent Malate Dehydrogenase is Present in the BAM1-LSF1 Complex

Even though the different NAD-dependent MDH forms from plastid, mitochondria and microbodies are quite similar to each other (the plastid localized isoform shares 80% amino acid sequence identity with the other two), the peptides identified in the LSF1-TAP and BAM1-TAP ( bam1 ) purification clearly identified the LSF1, BAM1 (and BAM3) interacting protein as the plastid localized NAD-dependent isoform by virtue of a C-terminal, extension unique to this very enzyme. The lack of an NAD-dependent MDH activity in lsf1 mutants revealed by native gels substantiates the LSF1-MDH interaction and adds another level of complexity to the lsf1 phenotype. From these experiments, it has to be concluded at the same time, that the larger part of the MDH activity missing in lsf1 mutants is not linked to the BAM1- LSF1 interaction. While a comparably weak MDH activity band co-localizes with the BAM1-LSF1

4. The Interaction Network of BAM1 and LSF1 98

complex, a comparably stronger band does not. This suggests that LSF1 might have functions and interactions which are not related to or dependent on BAM1. Although it has been shown that the MDH activity missing in lsf1 mutants and partially missing in bam1 mutants can use NADH, the identity of the protein responsible for the activity has not yet been revealed. The conclusion that the missing activity is indeed due to the p-NAD-MDH interacting with BAM1 and LSF1 remains likely but requires confirmation. Current work involving the identification of p-NAD-MDH mutant lines and testing MDH- specific antibodies will answer this question and help to directly assign the missing NAD-MDH activity to a protein.

4.3.4 Plastid localized NAD-dependent Malate Dehydrogenase – The Malate-Valve

The interaction of a plastid localized MDH with a BAM and a putative glucan phosphatase may seem strange and certainly requires a deeper knowledge of the functions of malate dehydrogenases to even speculate about the potential functional significance of the BAM1-LSF1-MDH or LSF1-MDH interactions.

Malate dehydrogenases reversibly reduce oxaloacetate (OAA) to malate, thereby oxidising NADPH or NADH to NADP + and NAD + respectively. Isoenzymes of MDH are present in all plant compartments (Scheibe, 2004). In chloroplasts, the NADP dependent MDH activity required for the operation of the malate valve in illuminated plastids is best studied (Fig 4-20). The malate valve is thought to serve to regenerate NADP + from NADPH, by reducing OAA to malate. In exchange for OAA, the malate is exported from the chloroplast in the OAA/malate shuttle (Taniguchi et al., 2002). By this indirect export of reducing equivalents from the chloroplast, the electron acceptor NADP + is regenerated and the production of ATP in photosynthesis and oxidative phosphorylation can continue. Thus, the malate valve serves to balance the NADPH/ATP ratio in illuminated chloroplasts as it uncouples ATP production from NADPH generation (Berkemeyer et al., 1998). The NADP-dependent MDH activity in chloroplasts is inhibited by NADP+ (Scheibe, 2004) which avoids that NADPH is oxidized when it would still be used for assimilaroty processes in the chloroplast. Furthermore, the chloroplast NADP-MDH activity is activated by reduction in the light and completely inactive in the dark. When oxidized, a C-terminal extension of NADP-dependent MDHs, folds over the catalytic site, obstructing access of the substrate (Carr et al., 1999). Berkemeyer et al. (1998) unequivocally identified a chloroplast localized NADH- dependent MDH activity. This activity was shown to be redox independent and thus at least potentially also be active in the darkened plastid. It was suggested that a version of the malate valve was also active in darkened plastids (Backhausen et al., 1998; Scheibe, 2004). In this pathway, triosephosphate oxidation by NAD-GAPDH results in the production of ATP, NADH and 3-phosphoglycerate (3-PGA). While the

4. The Interaction Network of BAM1 and LSF1 99

ATP could be used in a number of processes (including starch degradation), excess reducing equivalents would be removed from the chloroplast by the OAA/malate shuttle known from illuminated plastids. The reduction of OAA to malate would in this case be accomplished by the p-NAD-MDH (Backhausen et al., 1998). The fact that exogenously supplied dihydroxyacetone phosphate (DHAP) was oxidized into 3- PGA when OAA was supplied to isolated intact spinach chloroplasts suggested that the oxidation of triosephosphate is indeed operating and dependent on OAA to remove malate from the plastid (Backhausen et al., 1998). These results have led to the suggestion that a dark-operating malate valve might be important for the balance of the redox status of the plastid and also provides a link to the degradation of starch.

Figure 4-20. The Malate-Valve in Illuminated Chloroplasts During photosynthesis, electrons (e-) become available in the forms of reduced ferredoxin and NADPH. NADP-dependent malate dehydrogenase (NADP-MDH) uses oxaloacetate (OAA) as the acceptor of reducing power from NADPH, converts it to malate and restores the electron acceptor NADP +. In exchange for OAA, malate can be exported to the cytosol, stored in the vacuole or metabolized in mitochondria. This malate-valve uncouples ATP synthesis from NADPH synthesis and helps to balance ATP/NADPH ratios. NADP-MDH is inhibited by NADP +. Thus, when all NADPH is consumed for assimilatory processes, malate is no longer shuttled out of the chloroplast (PET: photosynthetic electron transport; from Scheibe R., 2002).

4. The Interaction Network of BAM1 and LSF1 100

A B Starch Starch

SEX4 SEX4 ISA3 BAM1 LSF1 MDH SEX4 LSF1 MDH BAM3 SEX4 BAM1 BAM1 BAM3 PWD PWD BAM3 GWD AMP + Pi GWD AMP + Pi

ATP ATP ADP BAM3

BAM1 3PGA DHAP Fru-6P NADH NAD + Glc-6P

OAA MDH Malate LSF1 BAM1

Chloroplast Cytosol

Malate OAA

Figure 4-21. Models for LSF1 Function in Starch Degradation LSF1 forms a complex with BAM1 and p-NAD-MDH, potentially also with BAM3 and p-NAD- MDHMDH and p-NAD-MDH alone. Based on these observations, two models for the function of LSF1 in starch degradation can be suggested: (A) LSF1 facilitates binding of BAM1 (BAM3) to starch granules or influences the substrate specificities of the BAMs to promote β-amylolytic starch degradation. The starch binding capacity of LSF1 might increase the amount of BAMs bound to the starch granule and thus enhance starch degradation. It could be imagined that LSF1 helps to target BAMs to regions of the granule surface where phosphorylation of glucan chains has led to increased hydration and accessibility of glucan chains for hydrolytic enzymes. In the absence of LSF1, BAM1 and BAM3 might be insufficiently targeted to the granule. The inefficient β- amylolytic attack of the granule might result in the observed sex phenotype of lsf1 . (B) An alternative model explains the lsf1 sex phenotype by decreased p-NAD-MDH activity. ATP, required for the transient phosphorylation of glucan chains and thus for the degradation of starch can be provided by the oxidation of triosephosphates. Glucose-6-phosphate (Glc-6P) can be converted to fructose-6-phosphate (Fru-6P) and finally to dihydroxyacetone phosphate (DHAP). Oxidation of DHAP generates 3-phosphoglycerate (3-PGA) , ATP and reducing equivalents (NADH). A ‘dark-version’ of the oxaloacetate (OAA)/malate valve, which involves p-NAD-MDH might restore electron acceptors and finally remove reducing equivalents from the chloroplast in the form of malate. The reduction of NAD- MDH activity in lsf1 might result in insufficient regeneration of NAD +, reduced ATP synthesis and thus impaired starch degradation.

4. The Interaction Network of BAM1 and LSF1 101

4.3.5 Models to Explain the Starch Excess Phenotype of lsf1 Mutants – BAM Stimulation versus the Malate Valve in the Dark

So far it has not been possible to relate the sex phenotype of the lsf1 mutant to a function of the LSF1 protein. With the extended knowledge on LSF1 and its interaction partners, three hypotheses can be suggested to explain the excess starch in lsf1 . Firstly, there still remains a possibility, albeit small, that LSF1 acts as glucan or protein phosphatase. The fact that no P-oligos accumulate in the lsf1 mutant was shown earlier (Comparot-Moss et al., 2010). The fact that the TAP-tag purified LSF1 is not able to dephosphorylate phosphorylated glucan chains was shown in this work. Both pieces of evidence suggest that this possibility is unlikely. This view of LSF1 is corroborated by as-yet unpublished experiments in the lab of Dr G. Moorhead (University of Calgary, Canada), where the initially active SEX4 glucan phosphatase has been rendered inactive by changing the histidine preceding the catalytic cysteine (Zhang and Dixon, 1993) in the DSPc motive to a threonine and thus mimicking the DSPc motive of LSF1 (compare Fig 4-1). In addition, it has been shown in the same lab that the heterologously expressed and purified DSPc domain of LSF1 does not have phosphatase activity. The heterologous expression and purification of sufficient amounts of soluble, full length LSF1 protein has not been achieved yet but would be desirable for functional tests of the LSF1 protein.

The second possible explanation for the sex phenotype of lsf1 is linked with its capability to interact with BAM1 and BAM3 as has been shown in this work. Preliminary results shown in this work suggest a stimulatory effect of LSF1 on the ability of BAM1 to degrade amylopectin. The presence of LSF1 in the BAM1-TAP purification increases the release of maltose from amylopectin by 42% or 31% in two independent experiments as compared to the BAM1-TAP protein purified in the absence of LSF1. These findings are currently limited in the range of substrates against which β-amylase activity has been tested and by the quantification method of the TAP-tagged proteins. The stimulatory effect of LSF1 on BAM1 activity is rather small but might be more pronounced if a similar comparison was possible for BAM3. In addition, not all BAM1 is complexed with LSF1 in the BAM1-TAP purification from the bam1 mutant background. Thus uncomplexed BAM1 might dilute the stimulatory effect of LSF1. Only speculation on how LSF1 might stimulate BAM activity is currently possible. LSF1 has been shown to be able to bind to starch granules in vivo (Comparot-Moss et al., 2010) and might thus target the BAMs to the granule surface (Fig 4-21A). BAM1 and BAM3 heterologously expressed proteins can bind to amylopectin in vitro (Li J, 2009) . The in vivo situation might be different, with the starch granule being the actual substrate of the BAMs and with the transient phosphorylation at the granular surface. The function of LSF1 could be to target the BAMs to the starch granule in a glucan-phosphorylation dependent way. If this was the case, LSF1 might target BAM to the granular surface specifically to the locations where

4. The Interaction Network of BAM1 and LSF1 102

glucan-phosphorylation is loosening the tight packing of glucan chains. This could lead to strongly enhanced and more efficient starch degradation (Fig 4-21A). However, the sequence of events including phosphorylation, dephosphorylation and glucan hydrolysis is quite complex and not fully understood yet. Such models remain speculative at the moment. Restoration of the BAM1-LSF1 and BAM3-LSF1 complexes in vitro and testing their activities as compared to the single β-amylases, in combination with the starch phosphorylating, dephosphorylating as well as debranching machinery will be crucial to put this idea to test.

A third possible explanation of the lsf1 sex phenotype can be suggested owing to the extended knowledge on LSF1 interaction partners. The sex phenotype of lsf1 mutants might be linked with the fact that the plastid localized NAD-dependent MDH activity is reduced in lsf1 mutants. An effect not directly related to BAM1, as the stronger of two MDH activity bands lacking in lsf1 is not co-migrating with the BAM1 protein. We do not have a quantitative measure of the reduction of plastid NAD-MDH activity in lsf1 yet . In addition, it is difficult to predict the effect of a reduction of plastid NAD-MDH activity on a plant’s capacity to degrade starch. No mutants lacking the plastid MDH activity are currently available and only indirect hints at the function of the plastid-NAD-MDH can be found. Providing isolated chloroplasts of Mesembryanthemum crystalinum with OAA for instance, caused a strong decrease in the DHAP/3PGA ratio, together with a stimulation of starch degradation. This stimulatory effect was comparable to the effect achieved by addition of exogenously supplied ATP (Neuhaus and Schulte, 1996). The continuous phosphorylation and dephosphorylation of glucose residues on the granule surface requires ATP which, in darkened, plastids could be provided by triosephosphate oxidation (Fig4-21B). A lack of plastid MDH activity might result in the accumulation of reducing equivalents and a lack of NAD + for the NAD- GAPDH reaction resulting in impaired starch degradation due to low ATP levels. Biochemical approaches to corroborate that the MDH activity missing in the lsf1 mutant indeed corresponds to the plastid NAD-MDH and to quantify the reduction of plastid NAD-MDH activity in lsf1 mutants are currently ongoing and will help to evaluate the relevance of the MDH-LSF1 interaction in the context of starch degradation.

4.3.6 Investigation of Multiple Mutants to Characterize the Function of LSF1 in Starch Degradation

From the results obtained in the LSF1 interaction experiments as well as the preliminary characterization of BAM1 activity with or without interacting LSF1 shown in this work, together with the characterization of the lsf1 mutant in previous work (Comparot-Moss et al., 2010), it has been possible to put forward two models for the LSF1 function in starch metabolism. In the last part of the results I present in this work, I

4. The Interaction Network of BAM1 and LSF1 103

have investigated the effect of multiple bam and lsf1 mutants to evaluate the in vivo significance of the two models. While the loss of BAM1 in the bam3 mutant caused an increase of the bam3 sex phenotype (Fulton et al., 2008), the bam1 /lsf1 mutant did not have increased starch levels as compared to the single lsf1 mutant. This suggests that the function of BAM1 can, at least partially, be taken over by BAM3. A subtle but significant increase in starch excess in the bam3 /lsf1 double mutant as compared to the lsf1 single mutant was observed. Thus, LSF1 provides a function in starch degradation which is not completely dependent on BAM3, potentially the interaction with BAM1. The fact that no reduction of MDH activity could be observed in bam3 mutants, thus favors a model where the interaction of LSF1 with the BAMs is causal for the lsf1 sex phenotype. Yet, it remains to be tested vigorously if indeed MDH activities in bam3 are equal to wild type. The increased sex phenotype in the bam3 /lsf1 mutant suggests that in the absence of LSF1, BAM1 is not able to compensate the loss of BAM3 to the same degree as in the single bam3 mutant. Thus, BAM1 might depend to a larger degree on LSF1 for starch degradation than BAM3.

The simultaneous loss of BAM1 and BAM3 conditions a sex phenotype which is larger than those of lsf , bam1 /lsf1 or bam3 /lsf1. This shows that both β-amylases are able to act independently of LSF1, which at best, provides a stimulation of BAM1 and BAM3 activities. However, loss of LSF1 in the bam1 /bam3 background does not result in an increased sex phenotype. This could be explained by the fact that if no β- amylases were present, stimulation of their activities by LSF1 is obviously not possible and thus favors a model where LSF1 directly influences BAM activity. However, it cannot be excluded that the reduction in transient phosphorylation resulting from impaired ATP provision in the lsf1 mutant has a negative impact on β-amylolytic capacity and thus contributes to a reduction in starch degradation.

4.4 Conclusion and Outlook

Starting from the observation that both, BAM1 and LSF1 are present in HMW complexes, I have shown in this chapter that the two proteins interact. In addition, they form a complex with a plastid localized NAD +-dependent MDH. It is likely but remains to be confirmed that LSF1 also interacts with BAM3. I suggest two new models for the action of LSF1 in starch degradation. In the first one LSF1 stimulates starch degradation by altering BAM1 (and BAM3) activity. In the second model, LSF1 is required for NAD-MDH activity in the chloroplast and thus indirectly for the supply of ATP for starch degradation. While the first model is appealing for its simplicity and the second model appears relatively indirect and complex, the data I have provided to not allow to exclude one or the other.

4. The Interaction Network of BAM1 and LSF1 104

Different approaches can be taken to validate the two models. In vitro work, describing the stimulatory effect of LSF1 on BAMs might help to substantiate the first model but will not disprove the second one. In vivo approaches should be considered at the same time. Investigating starch degradation in mutants lacking the p-NAD-MDH either totally or to certain levels would be suitable to test the second model. If starch degradation was not impaired in plants with reduced p-NAD-MDH levels, the second model could be excluded. In addition, it would be interesting to isolate wild-type and lsf1 mutant chloroplasts and determine the rates of starch degradation depending on whether OAA was supplied in the incubation medium or not. If indeed, a dark-version of the malate valve was required for the provision of ATP for starch degradation, then external OAA might stimulate starch degradation. If the p-NAD-MDH activity reduced in lsf1 was involved in this process and thus the second model for LSF1 action applied, the stimulatory effect of OAA supply on starch degradation should not be visible (or at least be reduced) in chloroplasts isolated from lsf1 mutants.

Attempts to complement the sex phenotype of lsf1 with versions of LSF1 missing the PDZ domain or parts thereof will help to describe the BAM1-LSF1 interaction in more structural detail. If it were possible to complement the lsf1 phenotype with a version of LSF1 not interacting with BAM1 (and BAM3), the second model would be highly favored. However, as the two models for LSF1 action are not mutually exclusive, it remains possible that both of them apply and that it will not be possible to exclude one of them. An evaluation of the relative importance of the two models would still be desirable and achievable with the approaches described above.

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 105 ______

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant

5.1 Introduction

5.1.1 C. peltata as a Model System for Water Soluble Polysaccharide Synthesis in Plants

The Müllerian bodies (MB) of C. peltata , represent an exceptional example of a myrmecophytic food structure produced by a higher plant. The accumulation of a glycogen-like, water-soluble polysaccharide (WSP) in MBs is outstanding, as this has so far not been described in such detail in non-mutant plant lines. The accumulation of WSP in total or partial replacement of starch has, however, been observed from mutants in different plant species lacking isoamylase-type debranching enzymes (James et al., 1995; Zeeman et al., 1998c; Kubo et al., 1999; Dauvillee et al., 2001a; Burton et al., 2002). The debate about the minimal set and combination of enzyme activities (synthases, branching and debranching enzymes) required for the synthesis of a crystallization-competent polyglucan (i.e. starch) is ongoing. MBs offer a unique possibility to make direct comparisons of enzyme activities in a WSP accumulating tissue and a starch synthesizing tissue from the same organism where starch is synthesized. This may help to disentangle the importance of the different starch synthetic enzymes for the synthesis of a crystallization- competent polyglucan. The two tissues, though fundamentally different, have the same genome and comparative analysis of gene expression might reveal as-yet unrecognized factors for the regulation of starch synthetic genes or enzyme activities. In addition, novel factors influencing the crystallinity of starch could be discovered. Although using C. peltata to improve our knowledge on starch metabolism provides unique opportunities, it also is challenging due to the complete lack of genome sequence information and molecular biology tools.

5.1.2 Approaches to Characterize Müllerian Body Metabolism

The synthesis of starch (but also glycogen) is a complex process and requires several enzymatic activities. In plants, multiple isoforms of starch synthesizing enzymes are present, with partially overlapping but also distinct specificities and preferences. A lack or reduction in DBE expression or activity may only be one of the possible explanations for the accumulation of WSP instead of starch. I have used three different approaches to investigate the WSP accumulation in MBs. I have described the WSP accumulating in MBs, as from the structure of a polyglucan, conclusions may sometimes be drawn about the enzymes involved in its synthesis. I have investigated the SSS, SBE and DBE enzyme activities in C. peltata leaves and MBs as changes in these enzyme activities are likely to affect the polyglucan structure. Such an approach is limited by the enzyme activities which can be tested and it often lacks the isoform

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 106 ______resolution which might be required to detect the reason for WSP accumulation in MBs. Furthermore, the interest in MB metabolism is of course not strictly limited to starch synthesis, even though this is the main focus of this work. Large structural changes occur in the plastids of MBs as compared to leaf mesophyll chloroplasts are large and previous work (Marshall and Rickson, 1973) suggests that the primary function of MBs is to accumulate large amounts of polysaccharides. To investigate MB metabolism upstream of WSP production and to obtain isoform resolution of expression of starch synthetic genes, a next generation sequencing (NGS) approach was taken to characterize the MB transcriptome in comparison to the C. peltata leaf transcriptome. In brief, this third approach involved partial sequencing of the C. peltata transcriptome, whereby, the number of sequence tags found for a given transcript was used as a measure for the expression level. The recent developments in sequencing technologies allow such transcriptomics approaches to be conducted in a non-model species. Inevitably, such an approach is technically challenging due to the fact that tools for NGS are still being developed and, in some cases, are not yet well established. The output from such an approach is a broad overview of transcriptional differences between the two tissues. This will include those required for the large plastid reprogramming to take place in MBs and those required for the fine tuning of expression levels of genes involved in starch metabolism. Both, the broad and focused views are desirable and worth the technical challenge.

The methods used for the characterization of MB polysaccharides and the native gel techniques have partially already been described in previous chapters. A short introduction into the NGS technologies used in this work, as well as a consideration of their suitability for NGS of non-model species is given below.

5.1.3 Transcriptome Sequencing in Non-Model Species

5.1.3.1 Next Generation Sequencing Technologies NGS technologies allow for the transcriptional analysis of non-model plants even where no prior sequence information is available. NGS technologies enable the quick, relatively inexpensive and comprehensive analysis of complex nucleic acid populations (Metzker, 2010). NGS technologies produce a massive amount of data that come with all the benefits and caveats of such high throughput approaches. The three main platforms for NGS which are currently available are the Genome sequencer FLX system from 454 Life Sciences/Roche (www.454.com), the Solexa/Illumina Genome Analyser from Illumina (www.illumina.com) and the SOLiD system from Applied Biosystems (www.appliedbiosystems.com). Each of the three technologies has its own benefits and drawbacks. Table 5-1 summarises the main features of these NGS technologies. While the 454 sequencing offers the longest reads with the lowest error rate, the system produces a lower sequence output and is also most

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 107 ______

expensive on a base pair (bp) basis. The Illumina and ABI technologies are more comparable to each other in terms of costs, the amount of sequencing data which can be generated per run, as well as in their error rates. One difference, important for some projects, is the difference in the achievable read lengths, which are currently longer for the Illumina technology than for SOLiD. Read length is an important factor to be considered for the downstream data-processing.

Here, the 454 and the Illumina systems have been used and a short introduction into the sequencing chemistry for these technologies follows. Every NGS technology runs in four different steps: template preparation, the actual sequencing process, imaging and data analysis. Both methods considered here, rely on the detection of light emission or fluorescence upon incorporation of nucleotides. However, as most imaging systems are not sensitive enough to detect single events, the DNA templates to be sequenced need to be amplified clonally to increase the signal in the sequencing step. The 454 and Illumina technologies provide different solutions to this problem and therefore the sample preparation differs.

The 454 technology uses emulsion PCR (emPCR) to achieve template amplification. Here DNA molecules are captured onto beads under conditions favouring one DNA molecule per bead. The beads are then encapsulated singly in aqueous droplets of an oil-aqueous emulsion containing the chemistry required for PCR amplification of the DNA molecules. Once DNA molecules have been clonally amplified on the beads, each bead containing several thousand copies of the same template sequence is deposited in a well of a PicoTiterPlate (www.454.com and Metzker (2010)). The Illumina/Solexa system uses solid-phase amplification to clonally amplify templates. In solid-phase amplification, DNA templates are linked to a glass slide via covalently bound primers. Initial priming and extension of the single stranded template is followed by local multiplication of the template by bridge amplification with immediately adjacent primers, resulting in local clusters on the glass slide (www.illumina.com and Metzker, (2010)).

To detect the incorporation of a given nucleotide in the actual sequencing process, the 454 system relies on pyrosequencing (Ronaghi et al., 1998; Margulies et al., 2005). The wells of the PicoTiterPlate are filled with additional, smaller beads, to which sulphurylase and luciferase are bound. The pyrophosphate, released by DNA polymerase upon nucleotide incorporation, is converted to ATP by the sulphurylase. The ATP is then used by the luciferase to convert luciferin to oxiluciferin in a light emitting reaction. Pyrophosphate release and light emission are proportional to the number of dNTPs incorporated and homopolymeric stretches of up to six nucleotides can be determined accurately. The sequence of nucleotide incorporation for every well is determined by washing the four species of nucleotides over the PicoTitrePlate in consecutive series. High error rates for homopolymeric stretches

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 108 ______

longer than six nucleotides and crosstalk between adjacent wells are technical problems currently limiting this technology. For example, the strong signals generated by the polyA stretches of mRNA interfere with efficient sequencing runs using the 454 technology. To circumvent this problem, polyA tails in the cDNA library used for the C. peltata sequencing, were altered using a degenerate primer approach (see Section 2.16.1 for details).

The Illumina/Solexa system uses a reversible terminator method to detect the sequence of incorporation of nucleotides in the clonally amplified clusters. In this method, DNA polymerase incorporates one fluorescently labelled nucleotide at a time. Incorporation of more than one nucleotide is prevented by a terminating group linked to the nucleotide. After unincorporated nucleotides have been washed off, the slide is imaged to determine the clusters which have incorporated the given nucleotide. Then, the terminator group is cleaved off and the process is restarted. A limitation of this method lies in the inefficiency with which DNA polymerases incorporate the labelled nucleotides.

Table 5-1.: Comparison of Next Generation Sequencing Technologies Library/ Amount of Sequencing Read Cost Error Platform template Mb/day data Pros Cons chemistry length (bp) (US$/Mb) rate preparation (Gb/run) Roche/454 GS Fragmentation, Long Limited Pyrosequencing 300-450 ~750 ~0.45 ~20 10 -3-10 -4 FLX Titanium emPCR reads output Illumina/Solexa Fragmentation, Reversible Large 18-125 ~7000 ~18 ~0.4 10 -2-10 -4 Short reads GA II solid-phase terminator output Fragmentation, Sequencing by Large ABIs SOLiD 25-75 ~7000 ~30 ~0.4 10 -2-10 -4 Short reads emPCR ligation output The comparisons are based on information on www.454.com , www.ill umina.com and www.appliedbiosystems.com in addition to a comparison by Metzker (2010). The exact values for the costs, for the speed, as well as for the amount of data generated are subject to changes in this rapidly developing field. This comparison of NGS technologies is not complete and does not include the emerging small suppliers of NGS technology (emPCR: emulsion PCR, see Section 5.1.3.1).

5.1.3.2 Considerations for Next Generation Sequencing Transcriptome Analysis A typical NGS transcriptome approach (hereafter called RNA-seq approach) sequences transcripts, generating large numbers of short reads. If an RNA-seq experiment is performed on a species for which extensive sequence information is already available, the reads mapping to specific genes can be counted and expression levels can be deduced. The expression levels of different genes are often compared by calculating the number of reads per kilobase exon-model per million mapped reads (RPKM, Mortazavi et al., (2008)). Therefore the RPKM value reflects the concentration of a transcript, normalized for the length of the specific transcript and for the total number of reads in an experiment or sequencing run.

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 109 ______

Comparability of RNA-seq approaches with microarray-based estimation of gene expression levels is good (Marioni et al., 2008; Wall et al., 2009). In addition, the background signal from RNA-seq experiments is lower than for hybridization based arrays. The dynamic range of RNA-seq is over 9’000- fold, as compared to a few-hundredfold for microarrays (Wang et al., 2009). Most RNA-seq approaches published so far, however, have focused on species with sequenced genomes. Even though microarrays or tilling arrays are available for most of these species, RNA-seq approaches have been chosen as they are not biased by an array design. These experiments have provided new insight into differential splicing and have advanced the identification of nucleotide markers (Lister et al., 2008; Mortazavi et al., 2008). For C. peltata , no prior sequence information was available, so methods used for model species were not applicable for my C. peltata RNA-seq approach.

Sequencing the transcriptome of a non-model species comes with important challenges. The most significant are: 1) the lack of a reference genome for the assembly and mapping of reads; 2) the choice of tissue type and cDNA preparation and 3) constraints on the financial side. To tackle these challenges, we have decided to run a mixed approach, which makes use of 454 and Illumina sequencing. In fact, it has been estimated that a combination of 454 sequencing with Illumina/Solexa sequencing provides the most cost-efficient strategy to obtain optimal transcriptome coverage in a RNA-seq approach (Wall et al., 2009). Assembly of short Illumina/Solexa reads without any reference is problematic (Miller et al., 2010). Therefore, to provide a framework for the assembly and mapping of the short reads, longer reads were generated first using the 454 technology. To obtain the maximum information on the C. peltata transcriptome, cDNA made from RNA of leaves and MBs was normalized and combined (Zhulidov et al., 2004). In the normalization procedure, highly abundant transcripts are specifically reduced. Normalization should therefore increase the number of different transcripts detected, increasing the coverage for low and medium abundance transcripts. This allowed me to get a broad view of the C. peltata transcriptome. In a second step, quantitative information about the expression levels of genes in different C. peltata tissues was determined by sequencing non-normalized cDNA libraries from leaves and from MBs. This strategy provided optimal sequence assembly and annotation thanks to long 454 sequencing reads together with quantitative information about expression levels from the large numbers of short Illumina/Solexa reads. Such a strategy is outlined in Fig 5-1 and has recently been suggested by Wall et al. (2009) and in more detail by Bräutigam and Gowik (2010) to be optimal for the analysis of the transcriptome of so far unsequenced species.

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 110 ______

reference transcriptome for non-model species

joint assembly of contigs and short reads

reference transcriptome

Figure 5-1. Next generation Sequencing Strategy for RNA-seq of Non-Model Species A trade-off exists between the generation of comparably few, long NGS reads using the 454/Roche technology and the generation of many short NGS reads using Illumina/Solexa technologies. A dual strategy gives optimal transcriptome characterization. A mix of leaf and MB normalized cDNA libraries was sequenced generating long reads, to capture a broad overview of the C. peltata transcriptome and to improve mapping accuracy of short reads to the reference transcriptome. Long reads were pre-assembled into contigs (right arm of the graph). To obtain information on transcript abundance, replicate non- normalized cDNA libraries from each tissue were sequenced generating short reads (left arm of the graph). Short reads were used to improve and revise contigs generated from long reads. The joint assembly was then used as the C. peltata reference transcriptome. The contigs of the transcriptome were annotated (not shown here) and short reads mapped back to the reference contigs to determine relative transcript levels between C. peltata leaves and Müllerian bodies (graph modified from Bräutigam et al. (2010)).

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 111 ______

5.2 Results Part I – Biochemical and Enzymatic Characterization of Müllerian Bodies

MB metabolism was investigated from three different angles. Firstly, WSP accumulation in MBs was analysed to confirm earlier reports. Secondly, enzyme activities involved in starch synthesis were investigated using enzyme activity gels. Thirdly, a whole transcriptomic approach was applied using NGS technologies. In ‘Part I’ of the results the biochemical approaches are presented and discussed. In ‘Part II’, results obtained from the NGS transcriptomics approach are shown. The transcriptomic analysis using NGS technologies stands apart from the other as it describes the broad transcriptional changes between C. peltata leaves and MBs. Therefore, for clarity, the NGS approach is presented and discussed separately. My conclusions and a synopsis of results can be found in the discussion at the end of this chapter.

5.2.1 C. peltata ; Plant Growth, Propagation and Development of Myrmecophytic Traits

A tree of the species Cecropia peltata grows in the botanical garden in Bern. Head cuttings of newly emerging branches were transferred into potting soil and covered by a plastic bag to maintain humidity. Two head cuttings obtained by this method were transferred to a greenhouse at the ETH Zürich. MBs started to emerge from trichilia, as soon as the first leaf of the trees had reached about three quarter of its final size. Later on, MBs were found on trichilia, even when the leaf was still covered by the stipule. Generally, the trichilia of the top 4 leaves actively produced MBs. For older leaves, MB production ceased. The emergence of a single MB above the surface of an active trichilium is surprisingly rapid and completed within minutes. Approximately 30-40 MBs were produced per day, per trichilium. The MB emergence rate was highest on sunny days, between noon and 6p.m. Generally, MB production decreased during autumn, reaching a minimum in winter and increased in spring to reach a maximum in summer. The dependence of the MB production on light availability has already been published (Folgarait et al., 1994) and could reflect the importance of photosynthesis and carbon fixation in leaves to provide carbon for MB production. Ants present in the greenhouse quickly invaded the C. peltata trees and collected the MBs from the trichilia. The ant species present in the greenhouse was not determined and it is unclear whether it was of the genus Azteca, the ant species generally associated with C. peltata trees in the wild.

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 112 ______

5.2.2 Cellular and Subcellular Structures of Leaves and Müllerian Bodies

To investigate subcellular structures, leaf and MB tissue was harvested towards the end of the day, fixed in glutaraldehyde, stained with osmium (VIII)-oxide and embedded in spur resin. For light microscopy, the sections cut from embedded leaf and MB tissue were stained with toluedene blue, which stains primarily cell wall structures.

Light microscopic analysis of sections from embedded C. peltata leaves revealed a typical leaf structure with a thick epidermal cell layer and clearly defined palisade and spongy mesophyll below (Fig 5-2A). In all chloroplasts investigated by transmission electron microscopy (TEM; Fig 5-2B and C), normal looking starch granules with clear borders could be detected. Starch granules could be observed in the chloroplasts of mesophyll and epidermal cells. Being discoid, with an approximate length of 2 µm, starch granule morphology was comparable to Arabidopsis granules. The number of granules generally ranged from 1-3, which is slightly lower than in Arabidopsis . The chloroplasts in the epidermis and the mesophyll contained well defined thylakoid membranes and grana stacks. No material resembling a soluble polysaccharide could be detected using TEM techniques.

Light microscopic analysis of sections from embedded MBs revealed their multi-cellular structure. MBs are 0.5-1 mm in diameter and covered by a layer of epidermal cells (Fig 5-2D). They contain large numbers of homogenous looking cells beneath the epidermis, (Fig 5-2D). Transmission electron micrographs were made to investigate the subcellular structures present in MBs (Fig 5-2E-G). Cells were packed full of spherical organelles with a diameter of between 2.5 µm and 10 µm. These organelles are surrounded by a double membrane, suggesting they are plastids (Fig 5-2F, black arrows and G, white arrow). This confirms earlier conclusions that the spherical organelles are highly specialized plastids, which differentiated from chloroplasts during MB formation. Remnants of an internal membrane system which could have been part of a thylakoid system (Fig 5-2G, black arrow) were visible in some of the MB plastids. Most of the internal space of the organelles is filled with small electron dense particles of 20-30 nm in diameter. These small particles are comparable in size and shape to glycogen particles in animals or WSP (phytoglycogen) particles accumulating in the Atisa1 and Atisa2 debranching mutants of Arabidopsis (Delatte et al., 2005). No starch granules were detected in any of the plastids in the MBs, neither in the inner cell layers nor the surrounding epidermal cell layer.

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 113 ______

A B epi

pal S sp

S S gc

100µm 2µm C

S

S S pal epi 2µm2µm

DE epi

PG PG LD

PG LD

PG

PG

100µm 5µm FG 1µm 0.5µm

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 114 ______

Figure 5-2. Ultrastructures of C. peltata Leaves and Müllerian Bodies C. peltata leaves and Müllerian bodies were harvested at the end of the day and fixed in glutaraldehyde. After staining using osmium (VIII)-oxide and dehydration in increasing ethanol series, samples were embedded in Spurr, as described in Section 2.15. Sections were cut for light microscopy (LM) and transmission electron microscopy (TEM). (A) LM image showing a section of a C. peltata leaf, with the thick epidermis (epi) and the palisade (pal) and spongy mesophyll (sp) structure, generally known from plant leaves. (B) TEM image showing the starch granules (s) inside a guard cell (gc) of a C. peltata leaf. (C) Representative TEM image, showing epidermal and palisade mesophyll chloroplasts with a clear thylakoid structure and starch granules. Starch granules were present in all chloroplasts of different cell types of C. peltata leaves. No deposition of soluble polysaccharides could be detected. (D) LM image showing a transversal section of an MB. MBs are 0.5-1 mm in diameter, multicellular and surrounded by an epidermal cell layer. (E) TEM micrograph showing that the densely packed cells inside MBs contain spherical plastid-like organelles. Apart from plastids, numerous lipid droplets (LD) and phytoglycogen (PG) are present in the MB cells. (F) TEM micrograph, showing an epidermal MB plastid, surrounded by a double membrane (black arrows in F, white arrow in (G)). (G) Magnification of (F). Compared to leaf chloroplasts, the MB plastids only contain remainders of an internal membrane system (black arrow) and are filled with small particles (20-30µm in diameter).

5.2.3 Soluble and Insoluble Polyglucan Levels

Total starch and WSP were extracted from C. peltata leaves and MBs which had been harvested at the end of the day. The amount of glucose present in water insoluble (starch), total water soluble (WSP and small polyglucans) and water soluble but methanol precipitable fractions (large soluble polyglucans i.e.: WSP only) was determined spectrophotometrically, by enzymatically converting the polyglucans to glucose and measurement of glucose in an NADH-linked assay. The amount of starch in leaves reached 11 mg glucose equivalents /g fresh weight (Fig 5-3) and thus was comparable to the starch levels in Arabidopsis leaves (10-15 mg Glc equivalents g -1 FW). In C. peltata leaves, 2 mg Glc equivalents g -1 FW of total soluble polyglucans were detected. Of these polyglucans, 97% were precipitable by methanol, suggesting that large, soluble branched polyglucans might be present in the leaves of C. peltata, albeit in comparably small amounts.

MBs contained high levels of soluble polyglucans, which accumulated to more than 25% of the MB fresh weight (311 mg Glc equivalents g -1 FW). The majority (88%) of these soluble polyglucans were precipitable by methanol, suggesting they are large, soluble polyglucans (Fig 5-3). However, compared to leaves, large amounts of small, soluble polyglucans might also be present. MBs also contained significant levels of insoluble polyglucans. Even though these levels were low compared to the total polyglucan, MBs contained a comparable level of insoluble polyglucans as leaves (11 mg Glc equivalents g -1 FW). The presence of insoluble polyglucans in MBs was unexpected, as no starch granules were observed in the TEMs. The nature of this insoluble polyglucan is not clear. Likewise, the presence of soluble but

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 115 ______methanol precipitable polyglucans in leaves was surprising; as such an accumulation of soluble polyglucans is generally not seen in Arabidopsis leaves. To better understand the differences between the soluble WSP in MBs and the insoluble starch in leaves but also to characterize the structure of the soluble polyglucans in leaves and the insoluble polyglucans in MBs, chain length analyses were made of the polyglucans.

Mullerian Bodies 350 Leaves

300

250 FW

-1 200 g * 150

12

8

mg Glc equivalents4

0 Total Soluble Soluble Insoluble MetOH Precipitable

Figure 5-3. Soluble and Insoluble Polyglucan Levels in C. peltata Leaves and Müllerian Bodies Soluble and insoluble polyglucans were extracted from leaves and MBs using perchloric acid. The water- insoluble polyglucans were washed 3 times with water to remove soluble glucan contaminations. The soluble polyglucans were precipitated with 75% (v/v) methanol to determine the contibution of large polyglucans (WSP) to the total soluble polyglucans. The results are the means ± SE of 4 samples. Two independent repetitions of the experiment yielded highly comparable results. Note the split Y-axis.

5.2.4 Structural Characterization of Polyglucans

Chain length distributions (CLDs) of branched polyglucans provide some insight into the architecture of the molecules. The relative proportion of chain lengths can determine solubility and crystallization- competence of a branched polyglucan. For CLD analysis, the branched polyglucans are extracted and the α(1,6) branch points hydrolyzed using commercial debranching enzymes. The resulting population of linear chains can then be analyzed by high-performance-anion-exchange chromatography coupled to

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 116 ______pulsed-amperometric detection (HPAEC-PAD), a method suitable for the separation and detection of uncharged glucose chains, according to their degree of polymerization (DP, i.e.: the number of glucose units present in a given chain).

CLDs only provide information about the length of the chains. Information about the distribution of branch points can, however, be obtained by first digesting outer chains, (i.e.: chains not bearing a branch point), using commercial β-amylases. β-Amylases remove maltosyl units from the exposed, non-reducing ends of glucan chains to within two or three glucose units from a branch point. The resulting β-limit glucans are debranched and their CLD is determined. Chains of DP2 and DP3 are removed from such a β- CLD analysis as they mainly originate from external chains. The β-CLD profile provides information about the length of internal chain segments and thus about how close branch points are located to each other.

The CLD of the starch accumulated in C. peltata leaves was compared to the CLD obtained from Arabidopsis starch (Fig 5-4A). Arabidopsis starch contains an increased proportion of chains of DP5 – DP15, whereas C. peltata starch is enriched in chains of DP16 – DP30. These differences reach up to 2% for some chain lengths (e.g.: DP6, DP10 or DP18) but do not result in differences in the overall appearance of the starch granules as can be seen from the TEM pictures of C. peltata leaves.

A comparison of the CLD of the insoluble polyglucan in the leaves with the soluble polyglucan in MBs, revealed an increase of short chains in the MB WSP (Fig 5-4B). Chains with DP5-DP10 were over- represented in the soluble polyglucan and reached maximal differences of up to 4% (e.g.: DP6 or DP8). Chains with a DP11 to DP21 were more abundant in the insoluble polyglucans of the leaves and reached differences of up to 2%. CLD comparisons between the soluble and insoluble polyglucans from leaves (Fig 5-4C) revealed only minor differences with the insoluble fraction containing slightly more chains of DP6 to DP10. The differences were even less pronounced when the insoluble and soluble polyglucan from MBs were compared. Only chains of DP2 and DP5 were increased in the soluble relative to the insoluble polyglucans. All other chain lengths were comparable (Fig 5-4D).

To obtain information about the internal organization of the polyglucans, β-CLDs were produced as described above. External chains are reduced to DP2 or DP3 during the production of the β-limit glucan. Therefore, these chain-lengths are not included in the analysis presented in Fig 5-5. When the β-CLDs were compared between the soluble polyglucan from MBs and the insoluble polyglucan (i.e.: starch) from leaves (Fig 5-5A), striking differences were obvious for the very short chains of DP4 and DP5. Internal chains of DP4 and DP5 were the most abundant internal chains in the soluble polyglucans from the MBs. Not surprisingly, DP4 and DP5 are strongly over-represented in comparison to the internal chains of starch from leaves. In contrast to the short chains, longer chains of DP ≥10 were more abundant in the

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 117 ______insoluble polyglucans from leaves. Comparisons of the β-CLD structures of insoluble and soluble polyglucans from leaves (Fig 5-5B) revealed comparable trends, with an increased portion of DP4, DP5 and DP6 chains in the soluble polyglucan. The differences of internal chains between soluble and insoluble polyglucans from MBs (Fig 5-5C) revealed a similar increase of chains of DP4 and DP5 in the soluble polyglucan. None of the differences between polyglucans from the same tissue were as striking as the difference between the soluble polyglucan from MBs and the insoluble polyglucan from leaves.

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 118 ______

A 10 4 Arabidopsis Leaf Insoluble - Cecropia Leaf Insoluble

8 Arabidopsis Leaf Insoluble 2 Cecropia Leaf Insoluble 6 0 5 10 15 20 25 30 35 40 45 50 4 -2 Relative % Relative Relative % Difference % Relative 2 -4

0 -6 5 10 15 20 25 30 35 40 45 50 4 B 10 Cecropia Leaf Insoluble – MB Soluble

8 MB Soluble 2 Cecropia Leaf Insoluble 6 0 5 10 15 20 25 30 35 40 45 50 4 -2 Relative % Relative Relative % Difference % Relative 2 -4

0 -6 5 10 15 20 25 30 35 40 45 50 10 C 4 Cecropia Leaf Insoluble – Cecropia Leaf Soluble Cecropia Leaf Soluble 8 2 Cecropia Leaf Insoluble 6 0 5 10 15 20 25 30 35 40 45 50 4 -2 Relative % Relative Relative % Difference % Relative 2 -4

0 5 10 15 20 25 30 35 40 45 50 -6 4 D 10 MB Soluble – MB Insoluble

8 MB Soluble 2 MB Insoluble 6 0 5 10 15 20 25 30 35 40 45 50 4 -2 Relative % Relative Relative % Difference % Relative 2 -4

0 -6 5 10 15 20 25 30 35 40 45 50 Retention Time (min)

Figure 5-4. CLD of Soluble and Insoluble Leaf and Müllerian Body Polyglucans Soluble and insoluble polyglucans were purified from MBs and C. peltata leaves. Insoluble polyglucans were washed with water to remove soluble contaminants and soluble polyglucans were precipitated in 75% (v/v) MetOH to remove small sugars. The resultant samples were debranched with Pseudomonas isoamylase and Klebsiella pullulanase and analyzed by HPAEC-PAD. Peak areas were summed, and the areas of individual peaks were calculated as a percentage of the total. Values are means ±SE of four replicates. The relative % difference between the two CLDs compared in the left panels is shown in the right panels. For the difference plots, SE of the two CLD profiles compared have been summed. Where error bars are not visible, they are smaller than the label. (A) Comparison of CLD profiles of the insoluble polyglucans (i.e.: starch) isolated from C. peltata and Arabidopsis leaves. (B) Comparison of CLD profiles of the insoluble polyglucans from C. peltata leaves and soluble polyglucan from MBs. (C) Comparison of CLD profiles between C. peltata leaf insoluble and soluble polyglucan. (D) Comparison of CLD profiles between MB soluble and insoluble polyglucans.

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 119 ______

A 40 15 Cecropia Leaf Insoluble – MB Soluble 5 30 MB Soluble Cecropia Leaf Insoluble -5 5 10 15 20 25 30 35 40 20

Relative % Relative -10 10 Relative % Difference % Relative -15

0 -35 5 10 15 20 25 30 35 40

B 40 15 Cecropia Leaf Insoluble – Cecropia Leaf Soluble

5 30 Cecropia Leaf Soluble Cecropia Leaf Insoluble 10 15 20 25 30 35 40 -5 20 Relative % Relative -10

10 Difference % Relative -15

0 -35 5 10 15 20 25 30 35 40

C 40 15 MB Insoluble – MB Soluble

5 30 MB Soluble MB Insoluble -5 5 10 15 20 25 30 35 40 20 Relative % Relative -10

10 Difference % Relative -15

0 -35 5 10 15 20 25 30 35 40 Retention Time (min)

Figure 5-5. β-CLD of Soluble and Insoluble Leaf and Müllerian Body Polyglucan Soluble and insoluble polyglucans were purified from MBs and C. peltata leaves as described in Fig 5-4. Soluble and insoluble polyglucans were treated with β-amylase, debranched with Pseudomonas isoamylase and Klebsiella pullulanase and analyzed by HPAEC-PAD. Peak areas were summed, and the areas of individual peaks were calculated as a percentage of the total. Chains of DP2 and DP3 were excluded from the analysis. Values are means ±SE of four replicates. The relative % difference between the two β-CLDs compared in the left panels is shown in the right panels. For the difference plots, SE of the two β-CLD profiles compared have been summed. (A) Comparison of β-CLD profiles of the insoluble polyglucans from C. peltata leaves and soluble polyglucan from MBs. MB soluble polyglucans contain by far more very short chains of DP4 and DP5, while the remainder of chains is relatively unchanged. (B) Comparison of β-CLD profiles between C. peltata leaf insoluble and soluble polyglucans. The differences are comparably small, minor increase in chains of DP4 and DP5 can be seen in the insoluble polyglucan. (D) Comparison of β-CLD profiles between MB insoluble and soluble polyglucans shows differences in chains of DP4 and DP5 which are increased in the soluble polyglucan. Chains of DP6 are more abundant in the insoluble polyglucan.

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 120 ______

5.2.5 Soluble Sugar Levels

Soluble sugars were extracted from leaves and MBs. The high molecular weight polyglucans present in leaves and MBs were removed by precipitation with methanol and the remaining small sugars were analyzed using HPAEC-PAD. Peaks were normalized to an internal standard to account for unequal losses during extraction. Glucose, fructose, sucrose and maltose were quantified by comparison to known standards. Glucose, fructose and sucrose levels were significantly increased in MBs as compared to leaves by 2- to 4-fold (Table 5-2). Maltose levels were very high in the MBs and more than 280-fold increased as compared to leaves.

Table 5-2: Sugar Levels in Leaves and Müllerian Bodies of C. peltata.

-1 (mg *g FW) Leaves Müllerian Bodies Factor

Glucose 2.3 ± 0.37 4.9 ± 0.18 2.1

Fructose 0.7 ± 0.13 2.3 ± 0.07 3.3

Sucrose 5.9 ± 0.67 23.2 ± 1.32 3.9

Maltose 0.02 ± 0.003 4.8 ± 0.46 287.3 Soluble sugars were isolated from C. peltata leaves and MBs by perchloric acid extraction. Glucose, fructose, sucrose and maltose were separated and detected by HPAEC-PAD. Peaks were normalized by an internal standard and quantified based on known sugar standards. Values are means of 4 replicate samples, -1 indicated in mg g FW, ± SE. The levels for all the sugars were 2- to 287-fold increased in MBs as compared to leaves (t -test; p -value ≤ 0.05).

5.2.6 Starch Modifying Enzyme Activities in C. peltata Leaves and Müllerian Bodies

To investigate the enzymatic alterations leading to the accumulation of large amounts of WSP in MBs, activities of enzyme known to be involved in the synthesis of starch in Arabidopsis and other plant species were analyzed. Native polyacrylamide gels containing amylopectin or glycogen which can be degraded or modified by enzymes separated electrophoretically in the gel, offer a rapid and semi- quantitative method to investigate those enzymes.

Protein extraction conditions suitable for Arabidopsis leaves were unsuitable for C. peltata and had to be adjusted to account for the high levels of phenolic compounds in C. peltata leaves. Furthermore, a comparison on the basis of equal FW, as it is often done for Arabidopsis , comparing leaves of mutant and wild-type plants, was not suitable for the highly distinct tissues of C. peltata leaves and MBs. Comparisons are thus based on equal total protein input.

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 121 ______

On amylopectin-containing native gels, stained with Lugol-Solution, a variety of activities can be seen from MB and from leaf extracts. Although the extraction medium has been adjusted for C. peltata leaves, ammonium sulfate precipitation improved the overall quality of the leaf extract and activity bands were clearer and sharper (Fig 5-6A, top panel). When equal amounts of total protein were loaded, a strong band of amylolytic activity (Fig 5-6A b) can be identified in MB and leaf extracts. A weak double band of glycogen degrading activity with electrophoretic mobility comparable to the strong amylolytic activity can be seen in leaf extracts on glycogen containing gels (Fig A, bottom panel, e). Another activity present in leaves and MBs can be seen at the very top of amylopectin-containing gels (Fig 5-6A a). While the band is very faint but visible in MBs, it is clearly discernible in leaves. The position of the band, its slight bluish nature and the fact that a dark band appears at the same position in glycogen containing gels (Fig 5-6A a’), suggests that this activity could be attributed to an ISA1/2-type of DBE activity (compare with Fig 4-2). The dark band on glycogen containing gels is thought to result from removal of excessive branches in the highly branched glycogen, which allows for more glucan chains to form double helices and thus show increased staining with Lugol-Solution. Based on an equal total protein basis, the amylopectin degrading or modifying activities are more active and diverse in leaves than in MBs. Some bands visible from leaf extracts cannot be seen in MB extracts (Fig 5-6A c, c’ and d, d’).

To investigate starch synthase activity, native gels containing low levels of glycogen were incubated in a medium containing ADP-Glc, the activated sugar used by starch synthases to elongate glucan chains (Fig 5-6B). The elongation of chains and thus extended double-helix formation is again detected by staining the gel with Lugol-Solution after the incubation. In extracts of Arabidopsis leaves, one weak and one strong dark staining band can be detected if ADP-Glc is present in the gel incubation medium but not in control gels where no ADP-Glc was present in the incubation medium(Fig 5-6B, top panel). The major starch synthase activities detected in C. peltata leaves and MBs migrate differently to each other and only partially with that from Arabidopsis . These differences might arise due to expression of different isoforms of the starch synthase enzymes but might also be related to the differences in solute composition between extracts of leaves and MBs. Quantitative comparisons of starch synthase activity between the two tissues is difficult based on native gels. However, total starch synthase activity detectable on native gels seems to be higher in MBs than in leaves.

Branching enzyme activity can be detected in native gels when the enzyme rabbit-muscle phosphorylase α and high levels of Glc-1P and AMP are present in the gel incubation medium. Generally, phosphorylase is thought to phosphorolytically release Glc-1P from glucan chains by using ATP. High levels of Glc-1P and AMP, however, result in the elongation of glucan chains through phosphorylase. Branching of newly synthesized chains provides new starting points for the elongation of chains by the phosphorylase, greatly

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 122 ______stimulating its activity. By this process, the activity of branching enzymes in a gel can subsequently be revealed by Lugol staining to show the branched-polyglucan accumulation where branching enzymes act (Fig 5-6C). Quantification of the enzyme activity is once again difficult, owing to the narrow dynamic range of the assay and a stochastic component in the initiation of glucan chain elongation. By loading 5- fold higher total protein amounts for C. peltata leaf extracts than for MB extracts, it was, however, possible to obtain equal SBE activities for MB and leaf extracts, suggesting that the SBE activity levels are approximately 5-fold higher in MBs as compared to leaves.

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 123 ______

C. peltata A MB Leaf B 20µg 40µg (NH 4)2SO 4 0% 80% 0% 80% MBLeaf MB Leaf a a c b c

d b

+ADP-Glc a’ e

c’

d’

-ADP-Glc

C MBLeaf MB Leaf MBLeaf MB Leaf µg Total Protein 0.06 0.3 0.12 0.6 0.2 1 0.25 1.25

100 MB SBE Activity Leaf SBE Activity

75

Tausende 50

25 Arbitrary Units Arbitrary

0

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 124 ______

Figure 5-6. Starch Modifying Enzyme Activities in C. peltata Leaves and Müllerian Bodies. Total soluble proteins were extracted from leaves and MB s using an extractionmedium adjusted for the high levels of polyphenols in leaves. Proteins were either precipitated by 80% ammonium sulphate [(NH 4)2SO 4] or directly used for native gels. Unless indicated otherwise, 20 µg total protein was loaded for leaves and MBs. (A) Amylopectin- and glycogen-degrading activities from crude or ammonium precipitated extracts were investigated using native gels containing amylopectin (top panel) or glycogen (bottom panel). Amylolytic or glycogen degrading activities are stronger and more diverse in leaves than in MBs. One strong amylolytic activity (b) is present in MBs and leaves. A weak activity at the top of the gel (a) is present in leaves and MBs on amylopectin gels and a dark band (a’) is present at the same location in glycogen containing gels. Amylolytic or glycogen degrading bands can be seen in leaf extracts but not in MB extracts (c, d, c’, d’, e). (B) Starch synthase activity is assayed using native gels containing glycogen, incubated in an ADP-Glc containing medium (upper panel). Elongation of glucose chains by starch synthase activity is revealed as dark bands when stained with Lugol-Solution. As negative control, gels were incubated in a medium without ADP-Glc (bottom panel). A weak band (a) and a strong band (b) of starch synthase activity can be seen in Arabidopsis extracts. Starch synthase activities from C. peltata are weaker and have different electrophoretic mobilities (c). (C) Branching enzyme activity is assayed by phosphorylase chain elongation stimulation. Native gels were incubated in a Glc-1P and AMP containing medium with rabbit-muscle phosphorylase α. Glucan chain synthesis/elongation by phosphorylase is enhanced by concurrent branching enzyme activity. Synthesis of branched polyglucan is revealed by staining with Lugol-Solution. Band intensities from different loadings of MB and leaf extracts were quantified using ImageJ (lower panel).

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 125 ______

5.2.7 Conclusions

5.2.7.1 Müllerian Bodies Accumulate Large Amounts of WSP My results strongly indicate that C. peltata indeed accumulates high levels of WSP in its MBs, corroborating and extending the results published by Marshall and Rickson (1973). The plastids present in MBs are highly different from chloroplasts in the leaves and likely dedicated to the synthesis and storage of WSP. As the thylakoids are missing, photosynthesis is not possible to drive carbon fixation for WSP synthesis. The MBs are thus likely to be a heterotrophic tissue, which relies on the import of sugars.. No connections of MBs to the vascular system of the plant have been described. However, such connections should exist as the MBs have to be supplied with the large amount of carbon required for the synthesis of the soluble polyglucans. The high levels of sucrose in MBs might originate from sucrose import from leaves and reflect the MBs function as a sink tissue.

5.2.7.2 CLD Profiles are Insufficient to Explain Müllerian Body WSP Accumulation The CLD profile of the starch accumulated in C. peltata leaves differs significantly from that of Arabidopsis starch. Despite the clear CLD differences between the two glucose polymers, both form semi-crystalline water-insoluble starch granules. There is a considerable degree of flexibility in the total CLD of a polyglucan which still allows crystallization. The distribution of short chains (DP6 to DP17) is characteristic for the amylopectins from different sources and has also been termed ‘fingerprint chains’ (Koizumi et al., 1991; Perez and Bertoft, 2010). It is thus not surprising to see differences in the CLD profile of C. peltata and Arabidopsis starch.

If the CLD of the WSP in MBs and the starch in the leaves was compared, the soluble polyglucan from MBs was clearly enriched for short chains. Especially for chains of DP6, DP7 and DP8, this difference was more pronounced than the difference between Arabidopsis and C. peltata starch. The WSP (termed phytoglycogen) accumulating in the Atisa1 and Atisa2 mutants of Arabidopsis shows a similar increase in short chains (mainly of DP5 to DP8) as compared to wild-type Arabidopsis starch (Delatte et al., 2005). Glycogen isolated from fungal, bacterial or animal sources generally has a shorter average chain length than starches from different sources (Manners, 1991).

It is intriguing that the CLD profile of the C. peltata leaf WSP is highly similar to the one from leaf starch. Likewise, the CLD profile of MB WSP and MB insoluble polyglucan is similar. It may thus be that CLD analyses alone are not sufficient to investigate the structural factors making a polyglucan soluble. A CLD profile gives an overview of the composition of chains but does not describe the organization of these chains. Distances between branch points may be as important to determine the

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 126 ______solubility of a polyglucan as the overall chain length and β-CLD profiles have been shown to be more conclusive to compare the structures of soluble and insoluble polyglucans than CLD profiles (Delatte et al., 2005)

5.2.7.3 Branch Points are Located Close to Each Other in Müllerian Body WSP Pronounced differences can be seen if the β-CLDs are compared between soluble and insoluble polyglucans of C. peltata . An increase greater than 15% in the relative proportion of the chains of DP4 in the WSP from MBs as compared to starch from leaves was observed. The differences between soluble and insoluble polyglucans from the same tissue are also more obvious than for the CLD. A general over- representation of short chains in the β-CLD profile is a feature found in the WSP (phytoglycogen) produced by the Atisa1 and Atisa2 mutants in Arabidopsis as well (Delatte et al., 2005; Streb et al., 2008). The specific massive over-representation of internal chains of DP4 found in WSP from C. peltata MBs is, to my knowledge, unprecedented. It suggests that branch points in the MB WSP are located very close to each other. It has been postulated that the length of internal chain segments may determine the degree of local crystallinity. Thus it is likely that the dense distribution of branch points interferes with the crystallization competence of a polyglucan and that this could be a reason for the soluble nature of the MB WSP (O'Sullivan and Perez, 1999; Perez and Bertoft, 2010).

5.2.7.4 Starch Branching Enzyme and Starch Synthase Activities are higher in Müllerian Bodies as Compared to Leaves. Starch synthetic enzyme activities have been investigated in a native gel based approach. The approach is limited to known enzyme activities which can be resolved and identified on native gels. In addition, absolute quantification of enzyme activities on gels suffers from poor linearity. C. peltata leaves and MBs are very diverse tissues and it is difficult to find protein extraction conditions optimal for both.

The number of amylolytic activities which could be revealed using amylopectin containing native gels was higher in leaves than in MBs. Although I have not tested it, it appears sensible that the starch synthesized in C. peltata leaves is degraded at night to fuel the plant’s metabolism. MBs, however, are harvested by ants or desiccate and fall off the trichilium once they have emerged above the trichomes. Again, although I have not formally tested it, it appears likely that the WSP accumulated in MBs is not remobilized by the plant. The higher diversity of amylolytic activities in leaves might reflect the plants need to degrade the starch again, as compared to MBs, where fewer amylolytic activities were observed.

From the analysis of SSS activities, I can conclude that overall SSS activity is higher in MBs than in leaves. However, due to the different migration pattern in leaves and MBs I cannot say whether the same isoform has different activities or whether different isoforms are active in MBs and leaves. The increased

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 127 ______

SBE activity which can be detected in MBs as compared to leaves might partially explain the accumulation of WSP in MBs. Tanaka et al. (2004) have increased the expression levels of SBEIIb in rice. A duplication of the amount of SBEIIb protein resulted in the accumulation of excessively branched and water-soluble polysaccharides. The soluble polysaccharide comprised approximately 10% of the total polyglucan and granular starch was still present. Thus the phenotype was comparable to but not as pronounced as the WSP accumulation in MBs. It was not possible to assign the changes in the SSS or SBE activities to specific SSS or SBE isoforms, even though this would be beneficial to understand the relative contributions of the different isoforms. To obtain quantitative information about expression levels of different isoforms of enzymes potentially involved in polyglucan synthesis, a more comprehensive, transcriptomic approach was developed. A comparison of the transcripotmes of the starch synthesizing leaves and the soluble glucan synthesizing MBs should not only reveal differences in the expression levels of starch synthesizing enzymes but might provide general information about the metabolic state of MBs.

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 128 ______

5.3 Results Part II – Characterization of the C. peltata Transcriptome

RNA-seq experiments for species where no genome (even of a closely related species) is available are challenging. Short reads to provide quantitative expression data are difficult to assemble into contigs de novo , but mapping the short reads to specific transcripts or genes is key to determine the expression levels of genes. Therefore, data was generated to support the assembly and annotation process of short reads. In order to establish a framework of sequences which could be annotated with high accuracy and to which short Illumina sequencing reads could be mapped, long sequencing reads were obtained using the 454 sequencing from Roche.

5.3.1 454 Sequencing of a Normalized C. peltata cDNA Library and Pre-Assembly

RNA was isolated from leaf and MB tissue and used to prepare cDNA. To mask the long homopolymer stretches in polyA tails, which can cause low quality reads in 454 sequencing, tails were modified using degenerate primers in the course of the cDNA preparation (seeSection 2.16.1). To obtain the broadest possible information about the C. peltata transcriptome, the cDNA preparations from leaves and MBs were normalized to increase the relative abundance of rare transcripts. Normalized leaf and MB cDNAs were pooled before library preparation and the GS FLX sequencing were conducted at the Functional Genomics Center in Zürich (FGCZ). From this sequencing, 997’683 reads were obtained, which had an average length of 380bp (Fig 5-7A). After the reads had been controlled for quality and adapter sequences had been removed in silico , the ‘GS De Novo Assembler’ software from 454 LifeSciences/Roche was used for assemble the reads into 38’972 contigs. The total length of the contig assembly was 26’343’561bp, the average length of contigs was 676bp and the N50 length was 948bp (Fig 5-7B). N50 is the length of the smallest contig in a set of contigs, whose combined length represents at least 50% of the bases in the assembly. This means that more than half of all the bases sequenced in the 454 sequencing were in contigs longer than 948bp. Contigs are built from several reads and the reads can partially (or totally) overlap. A given position in a contig can be sequenced several times, therefore have a coverage of a multiple of one. Thus, the coverage can be calculated per base pair and an average coverage can be calculated for a contig. Many of the 454 contigs have a low coverage, which can be seen if the average coverage of contigs is plotted against the number of contigs (Fig 5-7C). This suggests that we do not have a high overall coverage of the C. peltata transcriptome and despite our normalization effort, we might not have sequenced some transcripts.

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 129 ______

A B 3500 200 3000 150 2500 2000 100 1500 Number of Reads of Number

1000 Contigs of Number 50 500 0 0 100 200 300 400 500 600 700 0 500 1000 1500 2000 2500 3000 3500 Read Length (bp) Contig Length (bp) C 10000

1000

100

Number of Contigs of Number 10

1 1 10 100 1000 10000 100000 1000000 Average Coverage

Figure 5-7. 454 Sequencing of Normalized C. peltata cDNA Normalized cDNA libraries were prepared from RNA extracted from three C. peltata leaf and three MB samples, pooled and sequenced on two big regions of the GS Titanium kit of the Roche/454 GS FLX system. (A) Distribution of read length (after removal of sequencing adapters) of 997’683 reads. (B) A pre-assembly of the 454-reads was made using the ‘GS De Novo Assembler’ from 454Sequencing/Roche. Assembly of the reads resulted in 38’972 contigs with an average read length of 676bp. (C) Coverage of the assembled contigs was variable. Contigs with low average coverage (1-10) were most abundant, suggesting that the sampling of the C. peltata transcriptome is incomplete.

5.3.2 Sequencing of Non-Normalized Leaf and Müllerian Body cDNA Libraries

To obtain quantitative information about the expression levels of different genes in MBs and leaves, Illumina/Solexa sequencing was used to generate large numbers of short reads. Three replicate samples of leaf tissue and three of MB tissue were harvested. Care was taken to sample MBs which were newly emerging from trichilia, in order to harvest them in a state where they would still be metabolically active. MBs were harvested directly into liquid nitrogen and MBs from the same trichilium were pooled. Leaf material was harvested from the three leaves with the active trichilia from which the MBs had been harvested. RNA was extracted for the six samples and further processed at FASTERIS SA , where polyA

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 130 ______transcripts were purified and cDNA synthesis was performed using random primers with an invariant end. By the ligation of specific adapters, the six samples were ‘barcoded’ to be sequenced on two channels of a Genome Analyzer GAIIx. Reads of 76bp in length were obtained. The reads were attributed to the samples by their barcodes and 99.1% of all reads could unambiguously be assigned to a sample. Unassigned reads were not used further. A total of 29’424’574 reads were obtained for the three leaf samples and 28’588’287 for the MB samples. The percentage of reads attributed to each sample ranges from 17.3% to 54.5% of the total of reads in the corresponding channel.

5.3.3 Mixed Sequence Assembly

The final sequence assembly was generated in two steps. In the first step, the Illumina reads were assembled together with the pre-assembled contigs from the 454 sequencing. Assembling reads and contigs from different sources and of different length is not a routine procedure and still requires optimisation. The assembly process was done by FASTERIS, using the freely available programs Velvet and Oases. Velvet is an algorithm for de novo short read assembly using de Bruijn graphs. It takes in short read sequences, removes errors and then produces high quality unique contigs. It can use the long read information from the 454 contigs to retrieve the areas between contigs (Zerbino and Birney, 2008). Oases is an additional module to Velvet, specialized in transcription assembly, able to deal with the unequal coverage of reads expected in transcriptome sequencing. It uses a preliminary assembly produced by Velvet and exploits long read information, when available, to construct transcript isoforms (http://www.ebi.ac.uk/~zerbino/oases/). The sequences obtained after the Velvet/Oases assembly are termed ‘transcripts’ by the Oases package and this terminology is used hereafter.

The parameters, used to compute the mixed assembly, were adjusted so that the maximum number of Illumina reads could be mapped to the final transcripts. The final assembly produced by the combination of Velvet and Oases contained a total length of 4’549’006 bp in 5’992 contigs. The average length of the contigs was 758bp, with an N50 of 990bp. While 63.3% of all Illumina reads could be mapped to the pre- assembled 454 contigs, 72.6% of all Illumina reads mapped to this mixed assembly.The number of contigs in the mixed assembly was greatly reduced as compared to the pre-assembly of the 454 reads. This phenomenon was caused by a coverage cutoff built into the Oases assembly algorithm and is reflected by the fact that only approximately 45% of the pre-assembled 454 contigs are represented in the transcripts. For the analysis of read counts of the different transcripts, this was desired, as only transcripts with matching reads can be used in the quantitative analysis of gene expression. However, to describe the C. peltata transcriptome qualitatively, a reduction in the number of contigs/transcripts is not necessarily desired. The MUMmer package (Kurtz et al., 2004) was used at FASTERIS SA to retrieve contigs from

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 131 ______the 454 pre-assembly which were not present in the final Velvet/Oases assembly. Combining the two assemblies resulted in a final assembly with a total number of 23’695 contigs. While not required for the transcriptomic study presented in the following part, this assembly will be useful to support a proteomic approach to characterize the set of proteins present in C. peltata leaves and MBs (see Section 6.5).

5.3.4 Mapping Reads to the Reference Transcripts and Identification of Differentially Expressed Transcripts

5.3.4.1 Read Mapping and Quality Control To assign expression levels to the different transcripts, Illumina reads were mapped back onto the Velvet/Oases assembly transcripts using the CLC Genomics Workbench (CLCBio). Reads assigned to each biological sample were mapped individually to the transcripts.. Each transcript was thus assigned a read count for each of the 6 samples. To visualize the distribution of read counts, the read count for a transcript in a sample was log2 transformed and transcripts were categorized according to their log2 (read count) values. To take into account transcripts with no read counts in one or several samples, these transcripts were assigned a read count value of 0.25, which transforms into -2 on a log2 scale. The numbers of transcripts in categories were plotted in Fig 5-8. Transcripts with no matching reads in one of the replicates were scarce but more abundant in the ‘MB’ samples than in the ‘Leaf’ samples. This suggests that a smaller set of genes is expressed in MBs than in leaves. Most transcripts had read counts between 32 and 1024. However, a transcript with a maximum of 1’314’357 reads matching in a ‘Leaf’ sample and 1’138’687 reads matching another transcript in a ‘Müllerian Body’ sample can be observed. Even though large variations in read counts of individual transcripts exist between tissues and, to a lower extent, between biological replicas, the general distribution of read counts versus transcript frequency is comparable between tissues and biological replicas. Another simple way to compare the replicate samples is by looking at the dendrogram from complete linkage clustering. The correlation between samples was calculated and 1- the correlation between samples was used as the distance measure to cluster the samples (Fig 5-9). Samples cluster into two groups according to the tissue type. While the correlation between samples from the same tissue is close to 1, it is close to 0 between samples from different tissues. This analysis confirms that technical or biological variations between samples from the same tissue are much smaller than the differences between samples from different tissues. To compare the transcriptomes of leaves and MBs, the sums of all reads for a given transcript from either all leaf replicates or all MB replicates were log2 transformed and plotted against each other (Fig 5-10). The weak correlation between leaves and MBs is corroborated by this comparison. Many transcripts have high read counts from one but not the other sample. Biological replicates compared using the method (see Appendix II) reveal tight

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 132 ______correlations. Thus it can be concluded that the number of transcripts which are differentially expressed between leaves and MBs is large.

Figure 5-8. Distribution of Read Counts per Transcript Illumina reads obtained from each individual sample were mapped onto the transcripts obtained in the Velvet/Oases assembly. Transcripts are categorized according to their log2 transformed read counts (log2 Read Counts, X-axis). Transcripts with no matching reads were assigned a read count value of 0.25, which transforms into -2 on the log2 scale. These transcripts are thus represented by the very left hand side bar in the graph. The number of transcripts in a category is indicated on the Y-axis. Most transcripts had between 32 and 1024 matching reads.

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 133 ______

Figure 5-9. Complete Linkage Clustering for ‘Müllerian Body’ and ‘Leaf’ Samples The pairwise correlation between the 6 different samples (MB1-3 and Leaf 1-3) was calculated using the program ‘R’. 1- the correlation was used as the distance between the samples in complete linkage clustering. Samples cluster into groups according to tissue type. While the correlation between samples from the same tissue is close to 1, it is close to 0 between samples from different tissues. The differences between samples of different tissues are larger than technical or biological variations between samples of the same tissue.

‘Leaf’ versus ‘Müllerian Body’

20

15

10 in ‘Leaf’ in

5

Log2(read counts for each transcript) transcript) each for counts Log2(read 0 0 5 10 15 20 Log2(read counts for each transcript) in ‘Müllerian Body’

Figure 5-10. Comparison of Read Counts between ‘Leaf’ and ‘Müllerian Body’ Reads for a given transcript from all leaf of from all MB samples were summed and log2 transformed. These measures of abundance in each tissue were plotted against each other. The general correlation is weak. Many transcripts have high read counts from one but not the other tissue. Note that transcripts with no reads from either leaf samples or MB samples are not displayed in this graph. The blue line represents equal read count in the two tissues.

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 134 ______

5.3.4.2 Normalization of Read Counts A bias in read counts stems from the different number of reads retrieved for each sample in the Illumina sequencing. Normalization of read counts is therefore required. Often this is achieved by correcting for the total number of reads for a given sample; so called library size normalization. This principle has been widely used (e.g.: Mortazavi et al. (2008)). Using library size for normalization makes intuitive sense, given that it is expected that sequencing a sample to half the depth will give, on average, half the number of reads mapping to each gene (Robinson and Oshlack, 2010). However, if the total number of expressed genes is highly different between tissues or conditions, library size normalization is not appropriate. If some genes are highly expressed in one tissue but not expressed at all in the other, these consume a substantial proportion of the sequencing capacity causing other genes to be undersampled. In order to account for this problem, Robinson and Oshlack (2010) have developed a normalization procedure which takes into account that the proportion of reads attributed to a given gene (or transcript in my case) in a library depends on the expression of the whole sample rather than only on the expression level of the specific gene/transcript. The normalization according to Robinson and Oshlack has been applied to the C. peltata data set before the transcripts were investigated for differential expression.

5.3.4.3 Identification of Differentially Expressed Transcripts using ‘edgeR’ To identify transcripts which are differentially expressed (DE) between leaf and MB samples, the R Bioconductor package ‘edgeR’ was used (Robinson et al., 2010). This package provides routines for determining differential expression in digital expression data. Log2 fold changes (log2FC) between leaf and MB tissue were calculated as well as p-values. The log2 fold changes for all transcripts can be displayed using plots analogous to MA plots known from microarray analysis, where for all genes, the expression ratio between different conditions (or tissues) (M) is plotted on the Y-axis and the average expression (A) on the X-axis. The plots produced with ‘edgeR’ are comparable to standard MA plots in that they plot the transcripts based on their log2FC of a MB versus leaf comparison against their average read count in both tissues (Fig 5-11). The plots by edgeR differ from standard MA plots in that transcripts with zero expression in one tissue are plotted as a vertical ‘smear’ of points at the left-most edge of the plot (Fig 5-11, shaded region). The log2FC scale (Y-axis) is not applicable to these transcripts as their log2FC is technically infinite. Instead, the value on the Y-axis for these transcripts is log2(read counts) if no reads are present in leaf samples or -log2(read counts) if no reads are present in the MB samples. Although this is artificial and not optimal, it allows all transcripts to be visualized in the same graph.

Transcripts with a p-value < 0.001 are colored in red in Fig5-11 (3’320 transcripts meet this threshold). A strict definition of DE was used for the work presented here: the log2(FC) has to be smaller -2 or larger

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 135 ______than 2 and the p-value below 0.001. In addition, transcripts must have a minimum total of 30 read counts combined from all samples to be considered. Applying these thresholds, a total of 2’264 transcripts were identified as DE, more than 37% of all transcripts (Fig 5-12). Two hundred and ten DE transcripts were uniquely expressed in leaves and 40 uniquely in MB s (Fig 5-12). Of all DE transcripts, 946 were more highly expressed in MB s than in leaves and 1’318 were more highly expressed in leaves (Fig 5-12).

Figure 5-11. Fold Change Plot of ‘Müllerian Body’ and ‘Leaf’ Expression The ‘R’ Bioconductor package ‘edgeR’ (Robinson et al., 2010) was used to plot all the transcripts with their log2FC from a comparison of ‘Müllerian Body’ read counts and ‘Leaf’ read counts against their average expression in both tissues (log Conc). Transcripts with a log2FC of 0 are equally expressed in leaves and MBs. Transcripts with a positive log2FC are more highly expressed in MBs, transcripts with a negative log2FC are more highly expressed in leaves. Transcipts with zero read counts in either of the two tissues are represented by the vertical smear of dots at the left-most side of the graph (shaded area). Dots in red represent transcripts with a p-value smaller 0.001. To be defined as DE, transcripts must have more than 30 read counts in total, a p-value smaller than 0.001 and a log2FC above 2 or below -2, values indicated by the blue lines. The diagonal `lines' of points arise from the discrete nature of the count data (i.e.: in the line of points furthest away from the main body of points, the sum of all the counts in either leaf or MB samples is one).

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 136 ______

Figure 5-12. Detection of Differentially Expressed Transcripts Of a total of 5’992 transcripts, 2’264 were differentially expressed between MBs and leaves (see Fig5- 11). In leaves, 1’318 transcripts were more highly expressed than in MBs (log2FC<-2) and 946 transcripts were more highly expressed in MBs than in leaves (log2FC>2). The number of transcripts only expressed in leaves was more than five-fold higher than the number of transcripts only expressed in MBs.

5.3.5 Annotation of Transcripts and Gene Ontology Enrichment Analysis

5.3.5.1 Transcript Annotation by BLAST Searches Annotation of transcripts is crucial for the interpretation of the differential expression analysis by ‘edgeR’. However, as no reference genome or transcriptome are available, transcripts from the Velvet/Oases assembly were annotated de novo . Annotation was performed using BLAST2Go, an all-in- one online tool for functional annotation of novel sequences (http://www.blast2go.org/; (Götz et al., 2008)). Transcripts were annotated by searching the NCBI nr-Database and the Arabidopsis TAIR9 database using the ‘blastx’ algorithm. The BLAST e-value cut-off was set to 10 -5. Using these settings, 5’251 out of 5’992 transcripts (87%) could be annotated when blasting against the nr-Database and 4’926 (82%) when blasting against the Arabidopsis database. If the best BLAST-hit from the nr-Database was considered for each transcript, then C. peltata transcripts were most similar to sequences from Vitis vinifera (35%), Ricinus communis (23.3%) and Populus trichocarpa (23.3%). A relevant number of BLAST-hits were also obtained against sequences from Glycine max (6.4%), Brassica napus (2.5%), Malus x (2.5%), Arabidopsis lyrata (0.7%), Oryza sativa (0.7%) and Arabidopsis thaliana (0.7%). BLAST2Go retrieves gene ontology (GO) terms associated with the BLAST hits. From the pool of GO- terms, the most specific GO-terms were selected by BLAST2Go to annotate transcripts with BLAST hits.

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 137 ______

5.3.5.2 GO Analysis of Transcripts Uniquely Expressed or Over-Expressed in Leaves Having annotated the C. peltata transcripts, comparisons between the whole set of annotated transcripts and subsets of transcripts uniquely expressed in either MBs or leaves or DE between the two tissues were made. Such GO enrichment analyses are possible directly in BLAST2Go, which has the GOSSIP package implemented for statistical assessment of annotation differences between 2 sets of sequences (http://gossip.gene-groups.net/). The GOSSIP package uses a ‘Fisher’s Exact Test’, looking for GO-terms which are significantly enriched in a given subset of transcripts as compared to all annotated transcripts. A maximum false discovery rate (FDR) of 5% was accepted for the identification of GO-terms over- represented in subsets. Table 5-3 summarizes the GO-terms over-represented in the set of transcripts uniquely expressed in leaves. Transcripts with GO-terms associated with the chloroplast thylakoid membrane, the photosynthetic electron transport or the transport of triosephosphate are over-represented among the transcripts uniquely expressed in leaves. This is not surprising, as MB plastids do not have a thylaokid membrane system. This finding shows that the GO enrichment analysis produces biologically meaningful results and that transcripts uniquely expressed in leaves reflect their photoautotrophic state as opposed to the heterotrophic nature of MBs. If the same analysis is applied to the transcripts over- expressed in leaves as compared to MBs, the differences are more detailed but in principle comparable. As the list of GO-terms significantly over-represented among the transcripts more highly expressed in leaves than in MBs is long (see Appendix III), GO-terms were visualized using REVIGO (http://revigo.irb.hr/, Supek et al. (2010)). REVIGO is a web server that can take long lists of GO-terms resulting from high-throughput experiments and summarizes the lists by visualizing the GO-terms in semantic similarity-based scatterplots (Fig 5-13).

As for the uniquely expressed transcripts, GO-terms associated with many aspects of photosynthesis are over-represented. These include for instance electron transport in photosystem (PS) I, light harvesting in PS I, chlorophyll biosynthesis, light harvesting complex II catabolism and protein-chromophore linkage. On the side of carbon fixation, the reductive pentosephosphate pathway (Calvin-cycle) is over- represented as well as the oxidative photosynthetic carbon pathway (photorespiration). In conclusion, the transcripts over-expressed in leaves closely reflect the leaf’s functions in photosynthesis and carbon fixation, neither of which are part of the MB metabolism.

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 138 ______

Table 5-3: GO-Terms Over-Represented Among Transcripts Uniquely Expressed in Leaves Number of Transcripts in GO-term Description of GO-term FDR Test Set Total Transcripts GO:0009535 chloroplast thylakoid membrane 16 155 6.2E-04 GO:0015717 triose phosphate transport 3 3 1.4E-02 GO:0008131 amine oxidase activity 3 3 1.4E-02 GO:0009670 triose-phosphate:phosphate antiporter activity 4 9 1.4E-02 GO:0009773 photosynthetic electron transport in photosystem I 3 3 1.4E-02 GO:0010598 NAD(P)H dehydrogenase complex (plastoquinone) 3 4 1.7E-02 GO:0015301 anion:anion antiporter activity 3 5 2.5E-02 ‘Fisher‘s Exact Test’ implem ented in the GOSSIP -package in BLAST2Go was used to test for over - representation of GO descriptions in the set of transcripts uniquely expressed in leaves as compared to the GO descriptions of the total set of transcripts. A false discovery rate (FDR) of 5% was used as cut-off. A description of the GO group which is overrepresented and the FDR are indicated. In addition, the numbers of transcripts present in the GO groups from the total transcripts and the numbers of transcripts uniquely expressed in leaves are indicated. Some transcripts may be shared between GO-terms.

Figure 5-13. Visualization of GO-Terms Over-Represented Among Transcripts Over-Expressed in Leaves GO descriptions from the set of transcripts, over-expressed in leaves as compared to MBs, were compared to the GO descriptions of the total transcripts using the ‘Fisher‘s Exact Test’ implemented in the GOSSIP-package in BLAST2Go. A false discovery rate (FDR) of 5% was used as cut-off. REVIGO was used to visualize the list of overrepresented GO-terms (see Appendix III for the full list). A scatterplot for the biological processes is shown. Some transcripts may be shared between terms but the size of the circles is proportional to the number of transcripts assigned to each GO-term. REVIGO visualizes the GO-terms in a two-dimensional space based on semantic similarity between GO-terms. The axes have no intrinsic meaning but semantically similar terms should cluster together.

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 139 ______

5.3.5.3 GO Analysis of Transcripts Uniquely Expressed or Over-Expressed in Müllerian Bodies When the groups of GO-terms over-represented among transcripts uniquely expressed in MBs are analyzed (Table 5-4), only few transcripts are present in these groups and their biological functions are difficult to explain with regard to MB metabolism. The GO-terms over-represented in this set of transcripts are placed in pathways of cholesterol biosynthesis (C-4 methylsterol oxidase activity) or ethylene biosynthesis (1-aminocyclopropane-1-carboxylate synthase activity). Metabolism in MBs is focused and specialized as compared to leaves. However, most of the metabolic processes active in MBs are likely also present in leaves. Thus, most of the transcripts expressed in MBs may also be expressed in leaves. It is therefore more useful to investigate transcripts over-expressed in MBs relative to leaves. The full list of enriched GO-terms is provided in Appendix IV and GO-terms are visualized using REVIGO in Fig 5-14. According to the GO-enrichment analysis a number of areas of metabolism are up-regulated in MBs as compared to leaves. Major areas include cell wall metabolism and modification, carbohydrate, polysaccharide and glucan metabolism but also lipid, long-chain fatty-acid and wax biosynthesis. Furthermore, cyanide metabolism and biosynthesis are overrepresented, like the response to fungi, hypoxia and to salicylic acid.

Table 5-4: GO-Terms Over-Represented Among Transcripts Uniquely Expressed in Müllerian Bodies Number of Transcripts in GO-term Description of GO-Group FDR Test Set Total Transcripts GO:0050390 valine decarboxylase activity 2 2 1.0E-02 GO:0000254 C-4 methylsterol oxidase activity 2 2 3.0E-02 GO:0016847 1-aminocyclopropane-1-carboxylate synthase activity 2 3 3.1E-02 ‘Fisher‘s Exact Test’ implemented in the GOSSIP-package in BLAST2Go was used to test for over- representation of GO descriptions in the set of transcripts uniquely expressed in MBs as compared to the GO descriptions of the total set of transcripts. A false discovery rate (FDR) of 5% was used as cut-off. Some transcripts may be shared between GO-terms.

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 140 ______

Figure 5-14. Visualization of GO-terms Over-Represented among Transcripts Over-Expressed in Leaves GO descriptions from the set of transcripts, over-expressed in MBs as compared to leaves, were compared to the GO descriptions of the total transcripts (see legend of Fig 5-13). A false discovery rate (FDR) of 5% was used as cut-off. REVIGO was used to visualize the GO-terms (see Appendix IV for the full list). A scatterplot for the biological processes is shown. Some transcripts may be shared between terms but the size of the circles is proportional to the number of transcripts assigned to each GO-term.

5.3.6 A Close-Up View on Starch Metabolic Genes in Müllerian Bodies

To address the question of how WSPs are synthesized instead of starch in MBs, expression changes must be investigated at an isoform level. Changes in the relative activities of SSSs, SBEs and DBEs could account for the changes in glucan structure and solubility observed in MBs. As these enzymes are present in multiple isoforms, with different specificities and substrate preferences, it can be imagined that isoform specific changes may also be important. The shorter internal chain segments observed in my β-CLD analyses (see Section 5.2.4) could result from increased branching, decreased debranching or altered synthase activities. To investigate the individual isoforms of starch biosynthetic genes, the C. peltata transcripts have to be annotated using a reference database, where proteins have been defined at the isoform level. Using the nr-Database from NCBI, the best hits for the C. peltata transcripts were found from different species with incompletely annotated genomes (see Section 5.3.5.1). However, almost as many transcripts could be annotated by BLAST searches against the Arabidopsis TAIR9 database (see Section 5.3.5.1), offering the required annotation of isoforms and descriptions of their activities. Thus, transcript annotation based on the Arabidopsis genome was used to look at transcription of starch

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 141 ______metabolic genes in C. peltata . First, transcripts with homology to genes encoding proteins involved in starch degradation were investigated and then transcripts likely to encode proteins involved in starch synthesis. To investigate the supply of hexoses to the MB plastids for WSP synthesis, the expression levels of transcripts matched to Arabidopsis genes encoding proteins involved in sugar phosphate metabolism were also addressed, as well as sucrose metabolism in the cytosol.

In some cases more than one transcript matching the same protein in the BLAST searches was found. This may result from the fact that the depth of our sequencing data was not sufficient to provide enough data to completely cover the transcriptome. The ‘transcripts’ obtained in the assembly process are quite long (see Section 5.2.3) but they remain contigs from an assembly process of a sequencing experiment. The full length mRNAs of some genes may not be entirely covered and thus may be represented by more than one contig (a ‘transcript’ as defined by the Oases/Velvet assembly) in the C. peltata dataset. Another reason why more than one transcript could be assigned to a gene might be the existence of paralogs - genes duplicated in the C. peltata genome but not duplicated in (or lost from) the Arabidopsis genome. In addition, differential splicing might also give rise to different transcripts, assigned to the same gene. As suggested by OASES, the program used for the assembly, differential splicing occurred in approximately 2.4% of all transcripts. I have verified that differential splicing does not apply to the genes which I have looked at in detail in the following sections but cannot exclude a small impact of differential splicing on the GO-enrichment analyses presented above. Whenever more than one transcript was found matching a gene of interest, the read counts as well as the DE analysis are shown for all transcripts. Generally, the log2FC values are comparable between different transcripts matching the same Arabidopsis gene.

5.3.6.1 Expression of Transcripts with Homology to Arabidopsis Genes Involved in Starch Degradation When the C. peltata transcriptome dataset was searched for genes encoding proteins involved in starch degradation, an incomplete set was found. However, a number of proteins with key functions in starch degradation were identified and are shown in Table 5-5. Levels of transcripts matching genes involved in transient starch phosphorylation ( GWD ) and dephosphorylation ( SEX4 ) is not significantly changed between MB and leaf tissue. Based on homology to Arabidopsis sequences, the expression of the BAM1 β-amylase is unaffected, while the expression of BAM3 and the BAM1-interacting LSF1 is clearly lower in MBs as compared to leaves. The expression of the ISA3 transcript is significantly higher in MBs as compared to leaves, while the expression of the LDA-type of DBE is unchanged. Although, I have not been able to identify homologous C. peltata sequences for all Arabidopsis genes known to be involved in starch or WSP degradation, the data available suggest that, besides ISA3 , their expression is unchanged or reduced in MBs as compared to leaves.

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 142 ______

Table 5-5: Expression Analysis of Transcripts Homologous to Genes Involved in Starch Degradation

BLAST Sum of Reads Sum ofReads AGI Code Gene Name Gene Description Log2FC p-Value FDR e-Value Leaf MB 5.77E-41 141 47 -1.79 2.07E-05 4.45E-05 AT1G10760 GWD1 Glucan, Water Dikinase 5.4E-37 795 402 -1.26 1.51E-04 2.99E-04 AT3G52180 SEX4 Phospho-Glucan Phosphatase 4.02E-23 1273 1005 -0.69 3.32E-02 4.86E-02 AT3G01510 LSF1 BAM1 Interacting Protein 1.54E-30 221 15 -4.30 5.30E-20 2.47E-19 AT3G23920 BAM1 β-Amylase 2.69E-37 1064 367 -1.88 2.58E-08 6.86E-08 AT4G17090 BAM3 β-Amylase 6.49E-82 6256 157 -5.71 3.98E-47 6.44E-46 AT4G09020 ISA3 Debranching Enzyme 1E-168 269 2318 2.77 1.74E-15 6.63E-15 AT5G04360 LDA1 Debranching Enzyme 2.15E-40 359 308 -0.54 1.10E-01 1.46E-01 AT5G64860 DPE1 Disproportionating Enzyme 7.6E-61 232 518 0.88 9.53E-03 1.52E-02 The C. peltata transcriptome dataset was searched for transcripts coding for peptides with homology to Arabidopsis proteins known to be involved in starch degra dation. All transcripts are shown with the respective BLAST e-value, the sum of raw read counts from leaf and MB tissue, as well as the log2 Fold Change (Log2FC), p-value and false discovery rate (FDR) from the differential expression analysis. If more tha n one transcript matching the same gene was found, results for all transcripts are shown. None of the transcripts in the table result from differential splicing. Log2FC, p-value and FDR are calculated based on normalized read counts by ‘edgeR’ and cannot be directly derived from the raw read counts in the table.

5.3.6.2 Expression of Transcripts with Homology to Arabidopsis Genes Involved in Starch Synthesis When the expression levels of transcripts with homology to Arabidopsis genes involved in the synthesis of starch were compared between leaves and MBs, large changes could be observed for some transcripts (Table 5-6). Based on the sequencing data, the composition of AGPase, the enzyme synthesizing the activated sugar for starch and WSP synthesis appears to be changed in MBs as compared to leaves (Table 5-6). AGPase is a heterotetrameric enzyme, which consists of two small and two large subunits. Different isoforms of the AGPase large subunits exist in Arabidopsis . While the expression of transcripts homologous to the large subunit 1 ( APL1 ) isoform is decreased to almost zero in MBs, the expression of a transcript matching to the large subunit 3 ( APL3 ) isoform is massively induced. Conflicting results are obtained for the expression of the AGPase small subunit ( APS1 ). Based on one transcript, the expression of APS1 is induced in MBs as compared to leaves, while the second transcript suggests down-regulation. I cannot exclude either one of the two transcripts, but based on the comparably high e-value of the down- regulated transcript it is possible that the gene annotation for this transcript is misleading and that there are in fact two separate genes. A clear statement about the expression of the C. peltata AtAPS1 homologue requires independent experimental verification.

Starch synthases, the enzymes elongating glucan chains, show interesting isoform-specific expression changes between leaves and MBs. It is striking that transcripts annotated by BLAST searches to encode

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 143 ______peptides homologous to GBSS, SSSIII and SSSIV show reduced expression levels in MBs as compared to leaves. In contrast, the expression of the SSSI homologue is suggested to be increased by almost 150- fold. GBSS is one of the genes to which several transcripts matched. They are not the result of differential splicing, but represent different parts of the GBSS full-length mRNA. Debranching of glucan chains is a process involved in the synthesis of semi-crystalline starch. Interestingly, the expression level of the ISA1 DBE homologous transcript is reduced in the MBs as compared to leaves. The expression of transcripts matching AtSBE2 and AtSBE3 , genes coding for enzymes creating the branch points, are not largely changed between tissues. The altered expression patterns of the starch synthetic genes suggest that WSP accumulation in MBs may not only be the result of altered expression of different classes of genes (SSSs, SBEs and DBEs) but might also result from isoform specific expression changes (SSSs).

Table 5-6: Expression Analysis of Transcripts Homologous to Genes Involved in Starch Synthesis

BLAST Sum of Reads Sum of Reads AGI Code Gene Name Gene Description Log2FC p-Value FDR e-Value Leaf MB AT4G24620 PGI1 Phosphoglucose Isomerase 1 7.69E-165 584 222 -1.79 2.28E-07 5.65E-07 AT5G51820 PGM1 Phosphoglucomutase 1 0.00E+00 1627 2193 0.11 7.36E-01 7.74E-01 ADP-Glucose 1.11E-58 595 0 -35.54 1.33E-42 1.68E-41 AT5G19220 APL1 Pyrophosphorylase Large Subunit 1 5.40E-149 595 1 -9.61 3.65E-43 4.73E-42 ADP-Glucose AT4G39210 APL3 Pyrophosphorylase Large 0.00E+00 263 27653 6.41 1.29E-56 3.07E-55 Subunit 3 ADP-Glucose 0.00E+00 4325 16178 1.58 1.04E-06 2.47E-06 AT5G48300 APS1 Pyrophosphorylase Small Subunit 1 5.34E-08 469 149 -2.05 5.26E-09 1.46E-08 1.52E-14 787 37 -4.67 3.62E-29 2.50E-28 AT1G32900 GBSS Granule Bound Starch Synthase 5.05E-67 461 31 -4.34 2.41E-25 1.41E-24 1.71E-29 430 24 -4.43 6.19E-24 3.40E-23 AT5G24300 SSSI Starch Synthase I 0.00E+00 58 8872 6.91 1.84E-58 4.75E-57 AT3G01180 SSSII Starch Synthase II 6.61E-35 353 138 -1.65 3.50E-06 8.00E-06 AT1G11720 SSSIII Starch Synthase III 7.19E-104 1456 109 -4.02 1.81E-26 1.11E-25 AT4G18240 SSSIV Starch Synthase IV 1.55E-46 460 118 -2.33 1.12E-10 3.44E-10 AT2G39930 ISA1 Debranching Enzyme 1.55E-17 251 47 -2.73 6.39E-12 2.08E-11 3.28E-140 1939 2108 -0.22 4.93E-01 5.53E-01 AT5G03650 SBE2 Starch Branching Enzyme 2 7.65E-33 88 65 -0.95 1.49E-02 2.31E-02 AT2G36390 SBE3 Starch Branching Enzyme 3 5.21E-18 117 167 0.23 5.20E-01 5.78E-01 The C. peltata transcriptome dataset was searched for transcripts coding for peptides with homology to Arabidopsis proteins known to be involved in starch synthesis (see legend of Table 5-6). The log2 fold change (Log2FC), the p-value and the false discovery rate (FDR) are calculated based o n normalized read counts by ‘edgeR’ and cannot be directly derived from the raw read counts shown in the table. None of the transcripts in the table result from differential splicing. Despite having an equal number of reads from leaves, the transcripts assigned to APL1 are not identical.

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 144 ______

5.3.6.3 Expression of Transcripts with Homology to Arabidopsis Genes Involved in Sucrose Metabolism and Hexose-Phosphate Import into Plastids With the MBs being a heterotrophic tissue, carbon for the synthesis of WSP is likely to be provided in the form of sucrose from the leaves. In the cytosol, sucrose can be converted to UDP-Glucose and fructose by sucrose synthases. Fructose can be phosphorylated to Fru-6P by either hexokinase or fructokinase. Fru-6P is then interconverted to Glc-6P by the cytosolic PGI (see Fig 1-1). UDP-Glc is converted to Glc-1P by UDP-Glc pyrophosphorylase and the Glc-1P is interconverted to Glc-6P by cytosolic PGM (see Fig 1-1).

The Glc-6P synthesized in the cytosol can then be translocated into the plastids by the Glc-6P/P i translocator, where it is used for ADPGlc synthesis and subsequently for WSP or starch synthesis. Transcripts annotated to encode these enzymes in the Arabidopsis database were investigated for their expression levels in leaves and MBs (Table 5-7). Among them, the up-regulation of a transcript encoding a sucrose synthase gene ( SUS4 ) in MBs is striking. Transcripts of another sucrose synthase gene ( SUS3 ) are unchanged between leaves and MBs. Expression of a fructokinase gene is higher in MBs than in leaves but the expression is not significantly increased. Expression levels of cytosolic PGI and PGM and of a hexokinase gene are unchanged between leaves and MBs. Another pathway to metabolize sucrose in the cytosol is via neutral invertases, which hydrolyze sucrose to glucose and fructose in the cytosol. Glucose and fructose can be used to synthesize Glc-6P by the enzymes described above. Two neutral invertases in the C. peltata transcriptome are both significantly up-regulated in MBs compared with leaves. Glc-6P which is synthesized from sucrose in the cytosol can be transported into plastids by the

Glc-6P/P i translocator. A transcript encoding a protein homologous to the GPT2 translocator from Arabidopsis is highly induced in MBs. In contrast, expression of transcripts likely coding for the structurally related triose-phosphate ⁄phosphate translocator, which is involved in the export of photosynthates from chloroplasts for sucrose synthesis, is undetectable in MBs (Table 5-8).

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 145 ______

Table 5-7: Expression Analysis of Transcripts Homologous to Genes Involved in Sucrose Metabolism

BLAST Sum of Reads Sum of Reads AGI Code Gene Name Gene Description Log2FC p-Value FDR e-Value Leaf MB

AT3G43190 SUS4 Sucrose Synthase 0.00E+00 1045 9941 2.89 2.76E-17 1.15E-16 1.29E-28 197 132 -0.97 6.14E-03 1.01E-02 AT4G02280 SUS3 Sucrose Synthase 3.84E-45 154 131 -0.61 9.79E-02 1.31E-01 UDP-Glucose AT5G17310 ATUGP2 0.00E+00 1269 2381 0.56 7.94E-02 1.08E-01 Pyrophosphorylase 2 AT3G06500 Invertase Neutral Invertase 0.00E+00 663 7661 3.23 1.58E-20 7.59E-20 AT4G09510 CINV5 Cytosolic Neutral Invertase 1.04E-87 213 1115 2.05 2.81E-09 7.92E-09 Fructokinas Carbohydrate Kinase Family 4.57E-68 679 2539 1.53 3.39E-06 7.76E-06 AT3G59480 e Protein 9.80E-52 386 1831 1.89 2.19E-08 5.86E-08 Cytosolic Phosphoglucomutase 0.00E+00 94 88 -0.91 1.29E-02 2.01E-02 AT1G23190 PGM3 1 0.00E+00 104 151 -0.03 8.75E-01 8.98E-01 Cytosolic Phosphoglucose AT5G42740 PGIC 0.00E+00 318 390 0.03 9.24E-01 9.40E-01 Isomerase 0.00E+00 805 668 -0.63 5.48E-02 7.71E-02 AT4G29130 HXK1 Hexokinase 6.38E-106 693 439 -1.07 1.27E-03 2.25E-03 The C. peltata transcriptome dataset was searched for transcripts coding for peptides with homology to Arabidopsis proteins known to be involved in sucrose metabolism (see legend of Table 5-6). The log2 fold change (Log2FC), the p-value and the false discovery rate (FDR) are calculated based on normalized read counts by ‘edgeR’ and cannot be directly derived from the raw read counts shown in the table. None of the transcripts in the table are predicted to result from differential splicing.

Table 5-8: Expression Analysis of Transcripts Homologous to Genes Involved in Sugar Transport Processes BLAST Sum of Reads Sum of Reads AGI Code Gene Name Gene Description Log2FC p-Value FDR e-Value Leaf MB

Plastidic Glucose- 1.54E-140 324 2006 2.21 9.55E-11 2.93E-10 AT1G61800 GPT2 6Phosphate/Phosphate Translocator 2 3.19E-38 123 2306 3.95 3.66E-26 2.22E-25 3.1E-39 375 410 -0.13 7.25E-01 7.65E-01 AT5G16150 GLT1 Plastidic Glucose Translocator 3E-34 166 173 -0.40 2.50E-01 3.04E-01 1.64E-147 539 a 0a -36.18 4.41E-50 8.40E-49 Plastidic Triose 8.43E-141 1013 a 0a -37.08 4.62E-61 1.35E-59 AT5G46110 TPT1 Phosphate/Phosphate a a Translocator 2.8E-147 10 0 -29.87 1.77E-04 3.49E-04

4.68E-32 1761 0 -37.07 5.43E-61 1.57E-59 The C. peltata transcriptome dataset was searched for transcripts coding for proteins known to be involved in sugar or sugar-phosphate transport processes (see legend of Table 5-6). a sequen cing errors in these transcripts have resulted in partially overlapping but not assembled transcripts. No reads from MB samples are matching any of those transcripts .

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 146 ______

5.3.7 Conclusions

5.3.7.1 NGS Provides an Overview of the C. peltata Transcriptome Using 454 pyrosequencing coupled with Illumina sequencing it has been possible to generate a large amount of sequence data on the C. peltata transcriptome. The assembly of short and long reads still represents a challenging task in this emerging field and it remains difficult to identify the optimal procedure. The procedure to combine pre-assembled contigs from the 454 sequencing with the short reads from the Illumina sequencing using the Velvet/Oases packages has resulted in a relatively small assembly with few (5992) transcripts. The focus in generating this assembly was on creating highly contiguous transcripts (Zerbino and Birney, 2008; Paszkiewicz and Studholme, 2010). This approach has been successful as it yields a high number of transcripts which could be annotated using BLAST searches (87%). Another indication that the assembly generated high quality transcripts is shown by the long average contig length (758bp). This means that even though some transcripts remained fragmented, we have been able to compute a highly contiguous assembly. This is beneficial for an RNA-seq analysis, but the reduction in number of transcripts/contigs may not be desirable for other approaches which need the broadest possible overview over the C. peltata transcriptome. To support such approaches, the Velvet/Oases assembly has been supplemented with the contigs generated from the 454 pyrosequencing data. This mixed assembly will prove useful in aiding a proteomic characterization of C. peltata leaf and MB proteomes, which is ongoing in our laboratories.

5.3.7.2 Transcriptomes of C. peltata Leaves and Müllerian Bodies are Highly Different The basic quality control steps presented in Section 5.2.4.1 revealed that, while biological replicas were highly comparable, large differences were present between the C. peltata leaf and MBs transcriptomes. Even when applying very strict criteria, more than 37% of all transcripts were DE. Judged from the outward appearance of leaves and MBs, such large differences are not surprising . The former is a complex photosynthetic organ, where major parts of a plant’s metabolism take place. The latter is a highly-specialized autotrophic tissue, produced specifically for the deposition of WSP, lipids and protein. This difference is also reflected in the structure of the plastids in MBs, which is very different from leaf chloroplasts. MBs are rapidly emerging and short lived in comparison to C. peltata leaves. In leaves, plants tend to remobilize nutrients from leaves before they are lost, for instance during senescence (Hörtensteiner and Feller, 2002). In contrast, MBs, at least in most natural habitats, are removed by ants and thus lost to the C. peltata trees. The fact that the transcriptomes are very different and that they reflect the different types of the tissues is not in itself a means to validate the methods used in my RNA-seq

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 147 ______approach. The biology of the MBs and the leaves has to be compared in more detail would need to studied further using different methods for such a validation.

5.3.7.3 Müllerian Bodies: A Rapidly Growing, Heterotrophic Tissue Transcripts expressed in leaves but not expressed in MBs were largely involved in photosynthetic processes or in the export of photosynthates from chloroplasts. This is not surprising; MBs are not green and have no thylakoid membranes or only remnants thereof as shown by TEM analysis. Another area where the transcriptome comparison produces results which can be corroborated by observation is the up- regulation of genes involved in cell wall metabolism in MBs as compared to leaves. An active cell wall metabolism certainly is required for the rapid growth of MBs. The emergence of MBs above the net of trichomes covering the trichilia is completed within minutes, and 30-40 MBs can be produced per trichilium per day. Therefore, I hypothesize that cell wall synthesis and modification is rapid in MBs. However, we do not yet fully understand the mechanisms of MB growth and I have not investigated the genes contributing to this process in detail. Energy supply in this fast growing tissue must be a major challenge. The fact that the GO-term (PDC) is over-represented (see Appendix IV) is interesting with regard to this problem. Similar to MB growth, pollen tube growth is a very rapid process which can reach up to 1cm h -1 in maize (Taylor and Hepler, 1997). Energy for growth of the pollen tube has been suggested to be supplied by ethanolic fermentation, a process requiring PDC (Gass et al., 2005). Thus, part of the energy supply in MBs might also result from fermentation rather than oxidative respiration.

GO-terms involved in lipid metabolism and fatty-acid synthesis are over-represented amongst transcripts over-expressed in MBs as compared to leaves. (Rickson, 1971, 1976a) observed that MBs contained high numbers of lipid droplets. I have confirmed this in my TEM pictures of MBs (Fig 5-2), although the actual lipid contents has not been directly compared between leaves and MBs in this work or elsewhere. Nevertheless, it is not surprising that this part of the metabolism is significantly over-expressed in MBs as compared to leaves. GO-terms related to wax biosynthesis and metabolism are also enriched among transcripts over-expressed in MBs. Once MBs have emerged above the trichomes, they have no direct connection to the vasculature and eventually dry-out. It may well be that surface waxes act as a moisture barrier protecting MBs from dessication, which would otherwise be rapid when exposed to intense sunlight and wind.

While over-represented GO-terms in one or the other tissue may give hints at alterations in the metabolism, clear statements about specific changes in metabolic pathways, requires examination on a transcript by transcript basis. Such an analysis of all pathways is beyond the scope of this work. However, it is possible to look at specific questions, such as how large amounts of WSPs are synthesized in MBs.

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 148 ______

5.3.7.4 Genes Encoding Proteins Involved in Sucrose Catabolism and Glc-6P Import into Plastids are Over-Expressed in Müllerian Bodies I looked in detail at the genes involved in sucrose metabolism, sugar transport and polyglucan synthesis and degradation. In addition to providing the energy supply and precursors for growth and lipid metabolism, sucrose catabolism will generate precursors for WSP synthesis. This flux will be large in MBs considering that WSP accounts for one quarter to one third of the fresh weight. High expression of genes encoding enzymes of sucrose catabolism will be important to provide a flux into the hexose- phosphate pool, which in turn supplies all the above pathways. The exact partitioning of carbon between the different processes is difficult to assess and cannot be determined by transcript analyses.

WSP synthesis in MBs occurs in plastids, as is obvious from the TEM images. Two genes coding for Glc-

6P/P i translocators, GPT1 and GPT2 , have been described in the Arabidopsis genome (Niewiadomski et al., 2005). In Arabidopsis , GPT2 is non-essential and generally expressed at lower levels than GPT1 . AtGPT1 is expressed in a broad range of tissues but strongest in petals. No transcript encoding a peptide sequence with high homology to the AtGPT1 protein was identified from C. peltata . A GPT2 homologue was identified, however, and shows high expression levels in MBs compared to leaves. The GPT2 gene in Arabidopsis is most highly expressed in sepals and senescing leaves and its loss has no obvious effects on plant growth (Niewiadomski et al., 2005). However, the expression of AtGPT2 has been shown recently to be increased in mutant plants where soluble sugar concentrations are high due to impaired starch synthesis (Bläsing et al., 2005; Kunz et al., 2010). Soluble sugar levels are high in MBs and it is possible that this causes the up-regulation of the GPT2 homologue I observe. Such up-regulation probably ensures high rates of Glc-6P import into MB plastids. TPT belongs to the subfamily of phosphate translocators as the GPT proteins. One gene coding for a TPT has been identified from the Arabidopsis genome (Schneider et al., 2002) and a C. peltata homologue was identified among the sequenced transcripts. The AtTPT is highly expressed in photoautotrophic tissue and is responsible for the export of triose- phosphates generated in the Calvin-cycle, sustaining daytime sucrose synthesis in the cytosol. The lack of Illumina reads from MB samples which map to the TPT homologue probably reflects the heterotrophic state of MBs. Triose-phosphates are not extensively produced in the MBs, plastids and therefore they do not need to be exported.

Differences in the expression levels of C. peltata homologues of the Arabidopsis AGPase large subunits have been observed between leaves and MBs. It is possible that the AGPase subunit composition is different in MBs and leaves. The C. peltata homologue of the AtAPL3 subunit is much more highly expressed in MBs than in leaves, while the AtAPL1 homologue is more highly expressed in leaves. In

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 149 ______

AGPase, the small subunits are catalytically active and the large subunits are considered to have regulatory functions (Tiessen et al., 2002; Hendriks et al., 2003; Crevillén et al., 2005). In Arabidopsis , APL3 is the large AGPase subunit most highly expressed in sink tissues like inflorescences and roots. In leaves, APL1 is more highly expressed than APL3 (Crevillén et al., 2005). In addition, Arabidopsis APL3 expression is induced by high sucrose levels (Crevillén et al., 2005). Investigating the heterologously expressed Arabidopsis AGPase large and small subunits, it has been suggested that an AGPase containing the APL1 subunits would show high sensitivity to allosteric effectors and have high affinity for substrates, while both would be lower for an AGPase containing APL3 subunits (Crevillén et al., 2003). The over-expression of the C. peltata APL3 homologue in MBs thus suggests, that AGPase composition in MBs is similar to the AGPase composition in source tissues and that this change in subunit composition could be due to increased APL3 expression in MBs as a result of the high sucrose levels.

The sum of the changes I have described above suggests that, in comparison to leaves, MBs have a) increased import of Glc-6P, b) decreased export of triose-phosphates and c) an altered AGPase composition, reflecting the sink status of MBs. All these changes are consistent with the metabolism a non-photosynthetic tissue where large amounts of polyglucans are synthesized. The observation that isoform-specific expression changes can be observed between leaves and MBs and that these changes are consistent with observations made in other species supports the suitability of an RNA-seq approach to compare C. peltata leaves and MBs.

5.3.7.5 WSP Accumulation in Müllerian Bodies; Triggered by Altered Starch Synthase Isoform Expression? Having detailed the steps leading to the synthesis of ADP-Glc, the precursor for starch synthesis, I investigated transcripts with homology to Arabidopsis genes coding for SSSs, SBEs and DBEs. While I did not observe significant changes in the expression levels of the SBE homologues between C. peltata leaves and MBs, the expression level of the ISA1 homologue, encoding a DBE, is significantly decreased in MBs as compared to leaves. It was already mentioned in Section 1.4.2.3 that in Arabidopsis , the lack of the ISA1/ISA2 debranching enzyme activity results in the accumulation of highly branched, soluble polyglucans, similar to the WSP accumulated in MBs. No transcript encoding a protein homologous to ISA2 was found in the C. peltata transcriptome. The presence of an ISA2-encoding gene is widely conserved among monocots and dicots (Hussain et al., 2003). Thus, it seems likely that the depth of the sequencing was not sufficient to recover reads from the ISA2 transcript. Of the ISA1 and ISA2 proteins in potato, ISA2 has been shown to be catalytically inactive and interacts with ISA1 in an enzyme complex (Hussain et al., 2003; Utsumi and Nakamura, 2006; Kubo et al., 2010) . Even though I do not have

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 150 ______information about the expression level of ISA2, the transcript data suggests that debranching of polyglucans in MBs could be reduced as a result of decreased ISA1 expression.

Among starch synthases, the transcriptome analysis reveals that the isoform expression pattern is markedly shifted between MBs and leaves. MBs have greatly induced levels of expression of the SSSI homologue, while the expression of GBSS, SSIV and SSSII homologues is significantly reduced. The SSS isoforms are believed to have unique functions in the synthesis of amylopectin (James et al., 2003)and alterations in their expression pattern are likely to impact on the structure of the polyglucan which accumulates in leaves or MBs. It has already been mentioned in Section 1.4.2.1 that SSSI in Arabidopsis is mainly involved in synthesizing short chains, while SSSII and SSIII preferentially act on and synthesize longer chains (Delvalle et al., 2005; Zhang et al., 2008). Based on the analysis of different SSS mutants in maize, it has been suggested that SSSI primarily synthesizes the shortest chains (DP ≤10) and that further extension of chains by SSSII and SSSIII is required for the synthesis of chains which can span clusters of branch points (see Section 1.3.3). The altered expression pattern of the SSS isoforms in MBs as compared to C. peltata leaves thus suggests that the capacity to synthesize very short chains might be significantly increased, while the capacity to elongate chains and synthesize longer chains might be reduced.

In conclusion, the transcriptome comparison of C. peltata leaves and MBs suggests that a) the expression of SBE encoding genes is not altered between tissues; b) the expression of a DBE involved in polyglucan synthesis is reduced in MBs as compared to leaves and c) an altered expression pattern of SSSs favors the synthesis of very short glucan chains in MBs. A combination of the latter two points could theoretically result in the accumulation of a polyglucan which is structurally different from amylopectin and soluble. Despite the power of transcriptome analyses in providing an overview of the genes expressed in different tissues, it is unclear whether transcriptional changes are reflected in protein levels or even enzyme activities. Complementary comparisons of tissues at metabolite, protein and enzymatic levels would provide the best dataset with which to explain the mechanisms of WSP accumulation in MBs.

5.4 Synthesis of Part I and Part II

The major goal of the investigation of the MB s of C. peltata has been to find out how large amounts of soluble WSP can be synthesized in the MB plastids. The biochemical analysis of the WSP has shown that the CLD and β-CLD of this WSP are highly unusual, revealing a massive over-representation of internal chain segments of DP4 and DP5. Such drastic differences have not been observed before, even in the DBE mutants, which also accumulate WSP (Zeeman et al., 1998c; Delatte et al., 2005; Wattebled et al.,

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 151 ______

2005; Streb et al., 2008). It is plausible that short internal segment length (i.e.: the short distances between branch points) is the basic structural reason preventing the polyglucan in MBs from crystallizing (O'Sullivan and Perez, 1999; Perez and Bertoft, 2010) . The dense arrangement of branch points might not allow for a sufficient number of glucan chains to form double helices. Double helices that are formed (if any) might not cluster together in alternating regions of high and low branch point frequency.

Enzymatic analyses have suggested that SBE activity is approximately 5-fold higher in MBs as compared to leaves. A high degree of branching enzyme activity likely contributes to the short internal segments observed in the β-CLD of the MB WSP. However, the comparison of SBE transcript levels between MBs and leaves has not revealed an increased expression of these genes in MBs. Apart from the fact that I might not have recovered transcripts encoding another SBE, up-regulated in MBs, SBE activity in MBs could be regulated by post-transcriptional or post-translational mechanisms. Based on my analyses, it is not possible to discriminate between the three possibilities for SBE regulation. Future studies involving proteomic analyses of the proteins present in MBs and leaves will help to answer this question. Increased SBE activity is not the only factor which can contribute to the occurrence of short internal chain segments of the MB WSP. It could be imagined that if chain branching was increased in parallel with increased chain elongation and chain debranching, a crystallization competent polyglucan might still be synthesized.

I have observed a general increase in the level of SSS enzyme activity in MBs as compared to leaves using native gels, but I have not quantified the increase or assigned it to a specific SSS isoform. Transcriptional analyses have revealed that the expression levels of SSSs, creating long chains are reduced in MBs, while the expression of a short chain-synthesizing SSS (SSSI) is strongly increased. The increased SSS enzyme activity observed in MBs might therefore be due to increased SSSI activity. I suggest that the short internal segments in the MB WSP are not only caused by increased branching but also by decreased synthesis of long chain segments. Again, proteomic and further enzymatic analyses will help corroborate the transcriptional changes and potentially assign the increased SSS enzyme activity to a specific isoform. The alterations in SBE and SSSs activity/expression I have described might be sufficient to synthesize polyglucans incapable of crystallization. It remains possible that a high level of DBE activity might ‘correct’ the polyglucan structure, remove ‘excess’ branch points rendering it crystallization competent once more.

When I tried to analyze DBE activity using glycogen-containing and amylopectin-containing native gels (see Section 5.2.6), an activity possibly attributable to the ISA1/ISA2 activity could be detected. This activity appeared to be reduced in MBs. However, it was not possible to reliably quantify the activity or confirm its identity as the ISA1/ISA2 enzyme. The transcriptomic analysis has revealed that the

5. Cecropia peltata – Glycogen Synthesis in a Vascular Plant 152 ______expression level of the ISA1 homologue in C. peltata is in fact reduced in MBs as compared to leaves. Thus, I suggest that a reduction in debranching does contribute to the accumulation of soluble polyglucans in MBs.

I have now considered all the enzymes generally associated with starch synthesis. However, it has recently been suggested that even enzymes considered to be involved in starch degradation might influence the structure of a newly synthesized polyglucan (Streb et al., 2008). A rapid shortening of glucan chains, concurrent with glucan biosynthesis by either exo-amylases ( β-amylases) or endo- amylases ( α-amylases) could result in short chains and thus also contribute to make a polyglucan less crystallization competent. Neither the native gel analyses nor the transcript analyses suggest increased activities or expression levels of amylases in MBs as compared to leaves. I therefore don’t consider amylases as a major factor contributing to WSP synthesis in MBs.

If I consider the data and conclusions above, I can put forward the following model for WSP synthesis in MBs: 1) the supply of ADP-Glc as precursor for WSP synthesis is unrestricted, 2) short glucan chains of DP ≤10 are efficiently and rapidly synthesized but elongation stalls 3) short chains are branched rapidly and at high frequency, 4) branches are not removed at a rate high enough to prevent the accumulation of a polyglucan with very short internal chain segments; the WSP of MBs.

6. General Conclusions 153 ______

6. General Conclusions

6.1 Proteins Involved in Starch Degradation are Present in High Molecular Weight Complexes

No comprehensive analysis of the NMW of the proteins involved in starch degradation has so far been conducted. Thus, even though the investigations of NMWs presented in Chapter 3 are preliminary, they provide a starting point for further research. I have already pointed out in Chapter 3 that large scale proteomics analyses have corroborated the presence of GWD and SEX4 in HMW complexes (Peltier et al., 2006; Olinares et al., 2010). The transient phosphorylation of starch granules is believed to be one of the first steps in starch degradation, whereby the packing of glucan chains is disrupted, permitting amylolytic degradation. It thus appears sensible that GWD and SEX4 could be key regulatory points in starch degradation and that protein-protein interactions mediate or modulate such a regulatory control. Interestingly, GWD is one of the starch degrading enzymes which, when heterologously expressed, is reductively activated in vitro (Mikkelsen et al., 2005; Kötting et al., 2010). Such a regulation would suggest that the enzyme had highest activity during the day in vivo , as chloroplasts are thought to be a more reducing environment during the day than at night. This is counterintuitive as transient starch phosphorylation is increased during starch degradation. A modulation of GWD redox regulation by interacting proteins could thus be conceived.

The de-phosphorylation of phospho-glucans is catalyzed by SEX4. Using phosphorylated crystalline maltodextrins and recombinant Arabidopsis SEX4 it has been shown that the enzyme can be significantly inhibited by low levels of maltoheptaose or soluble maltodextrins (Hejazi et al., 2010). The mechanism of this regulation is unclear so far and it could be mediated by protein-protein interactions. Furthermore, the most likely in vivo substrate of SEX4 is the surface of starch granules (Sokolov et al., 2006; Comparot- Moss et al., 2010; Hejazi et al., 2010). Mechanisms regulating the targeting of SEX4 to the starch granule surface might be involved in controlling starch de-phosphorylation and might be mediated by protein- protein interactions. With GWD and SEX4, glucan phosphorylating and de-phosphorylating enzymes are active on the starch granule surface and control mechanisms are probably required which avoid interference between them. SEX4 could counteract the GWD mediated solubilization of the granule surface (Hejazi et al., 2010). Mechanisms, potentially involving protein-protein interactions, should thus prevent SEX4 action before glucan chains have been degraded.

Like GWD and SEX4, BAM1 has been found to be redox regulated by means of heterologous expression and activity testing (Sparla et al., 2006). Again, the reductive activation of an enzyme hypothesized to be

6. General Conclusions 154 ______mostly active during the night is intriguing. It has been described in this work, that BAM1 interacts with LSF1. However, it has not been formally tested in this work whether the BAM1-LSF1 interaction directly influences the redox sensitivity of BAM1. The investigation of recombinant BAM1 enzyme properties in presence or absence of recombinant LSF1 might help to test for such effects in future work.

The GFC analysis presented in Chapter 3 of my work is incomplete. The resolution for the separation of proteins according to their NMW which was obtained using the GFC approach was insufficient to assign precise NMWs and it was not possible to detect elution peaks corresponding to different oligomeric states, for instance for BAM1. An increased precision of the GFC would help to assign sharper NMW ranges for proteins of interest. Another limitation of the GFC analysis results from the fact that for some proteins involved in starch degradation, no antibodies or activity assays were available. BAM4 is one of those proteins. It certainly is a most intriguing protein, which should be investigated in more detail. Its apparent non-catalytic nature, together with its starch binding capacity and the sex phenotype of the Atbam4 mutant (Fulton et al., 2008; Li et al., 2009) make it a fine candidate for a regulatory protein. The TAP-tagged BAM4 protein runs in two distinct bands on an amylopectin containing native gel (see Fig 4- 7). This might be due to the BAM4 starch binding capacity or partial protein degradation but might also reflect its presence in a HMW complex. The MS/MS based identification of proteins co-purifying with BAM4-TAP in the two-step affinity purification procedure, has not yet revealed any proteins known to be or suspected to be involved in starch metabolism. Therefore, the function of BAM4 and the question whether it is part of HMW complexes await further investigations.

6.2 BAM1, LSF1 and p-NAD-MDH: The ‘When’ ‘Where’ and ‘How Much’ of Protein-Protein Interactions

In this work, I have shown that the two major β-amylases in the chloroplast form complexes with the starch binding protein, LSF1. Such a detailed investigation of interactions between proteins involved in starch degradation has not been conducted previously. I have proposed two models for the impact of these interactions on starch degradation. The first suggests a direct involvement of the β-amylases, which are targeted to the granule surface via the interaction with LSF1. The second model suggests that the activity of p-NAD-MDH interacting with BAM1 (BAM3) and LSF1 is the major factor influencing starch degradation. This model envisages that for continuous ATP supply for starch degradation, reducing equivalents must be removed from the chloroplast in a dark version of the malate valve. Even though the first model does not explain the presence of p-NAD-MDH in the complex, I have provided genetic evidence suggesting that there is a direct impact of the LSF1 interaction with BAMs on starch

6. General Conclusions 155 ______degradation. The fact that no increase in the sex phenotype of the double bam1/bam3 mutant could be observed, when LSF1 was missing in addition (the bam1/bam3/lsf1 triple mutant) suggests that LSF1 acts via the BAMs. However, I cannot exclude that a reduction in NAD-MDH activity in lsf1 mutants influences starch degradation to some extent. Further investigations are crucial, especially understanding the ‘Where?’ and the ‘When?’ and ‘How much?’ of the protein interactions may help to discriminate between the models.

For the BAM1-LSF1-MDH complex identified in this work, preliminary experiments have been conducted to investigate its stability. Interestingly, my data suggest an influence of redox state on BAM1- LSF1 complex formation. A preliminary attempt to look at the complex dynamics over the diurnal cycle using low amounts of DTT in the extraction procedure failed to identify any changes. However, it might be desirable to use no DTT at all, as the added DTT equalizes the redox conditions in all extracts regardless of the time points. The change in redox potential during protein extractions might be sufficient to alter the i n vivo ratios of complexed versus monomeric BAM1 or LSF1 and complicate investigations of the dynamics of BAM1-LSF interaction over the diurnal cycle. A rapid crosslinking of proteins during the extraction of proteins from leaves, harvested at different time points in the diurnal cycle, might be suitable to address the question of ‘When?’ BAM1 and LSF1 interact.

The question of ‘Where?’ could be addressed by FRET (Förster Resonance Energy Transfer) or BiFC based approaches. Both of which are non-invasive microscopy-based methods and allow monitoring of protein-protein interactions in vivo and potentially also in real time (Bhat et al., 2006; Citovsky et al., 2006; Vogel et al., 2006). A preliminary BiFC investigation, showing the interaction of BAM1 and LSF1 in the chloroplasts of tobacco epidermal cells has been done by Dr A. Graf whose results were presented in Section 4.2.5.1. While the microscopic equipment for BiFC is relatively simple, it is more complex for FRET. However, in contrast to BiFC, FRET would allow for the detection of the interaction partners alone and would also allow the dynamics of the interaction to be monitored. Theoretically, an evaluation of the in vivo interaction dynamics of BAM1 (BAM3) and LSF1 using FRET might reveal whether the complex formation is dependent on certain conditions in the leaf. Even though these microscopy-based methods are non-invasive, the experimental system would have to be carefully evaluated. Transient co- expression of tagged proteins using for instance Agrobacterium tumefaciens mediated transfer might not reveal the in vivo situation in a starch degrading leaf. Despite certain limitations of the BiFC and FRET methods, it might be possible to obtain information as to where in the chloroplast BAM1 and LSF1 interact. As LSF1 has been shown to bind to starch granules, a localization of the BAM1-LSF1 complex on the granule surface would not be surprising and add additional weight to the model favoring a direct impact of LSF1 on BAM activity.

6. General Conclusions 156 ______

The question of how many of the LSF1 proteins actually are in monomeric state, in complex with BAM1 and p-NAD-MDH, in complex with BAM3 or in complex with p-NAD-MDH is crucial to understand the function(s) of LSF1. The TAP-tagged LSF1 and BAM1 proteins I have used to establish and confirm the BAM1-LSF1 interaction were expressed under the control of a double 35S promoter. This might distort the endogenous relationships of proteins in monomeric or oligomeric forms. These experiments are thus not useful to estimate the relative portion of LSF1 in different complexes. A GFC protocol with improved resolution which allows to separate LSF1 in its different oligomeric states would allow investigating the relative amounts of LSF1 protein in the different complexes and potentially also verifying the importance of the LSF1-pNAD-MDH interaction.

6.3 Local Redox Modulation by p-NAD-MDH

I have mentioned that a number of enzymes active in starch degradation are reductively activated and that is in some ways a surprising finding. However, observed redox control in vitro need not necessarily equate to redox regulation in vivo . Different mechanisms may allow redox activated enzymes to be active on the granule surface during night-time starch degradation. To my knowledge, no investigations so far have indicated the redox status to be uniform across the chloroplast. Local redox environments might exist. With the p-NAD-MDH, I have identified an enzyme, interacting with starch degrading enzymes, which has the potential to alter the redox balance. By oxidizing malate to OAA and the concurrent generation of reducing equivalents, p-NAD-MDH could theoretically influence the local redox balance. It is tempting to speculate that p-NAD-MDH, by interaction with a starch binding protein, could render the vicinity of the starch granule a more reducing environment than the rest of the chloroplast. However, for several reasons I consider such a function of p-NAD-MDH unlikely. Cvetic et al. (2008) have investigated the properties of NAD-dependent MDH isoforms from spinach leaves, identifying one tightly associated with chloroplast envelope membranes and one that was stroma localized. The latter is likely to correspond to the p-NAD-MDH isoform described by (Berkemeyer et al., 1998), as their pIs were similar. Both, stromal and envelope localized forms of NAD-MDH, showed almost no reaction with malate but readily reduced OAA. Even though the significance of this finding is of course influenced by the in vivo concentrations of these metabolites, it appears unlikely to me that large amounts of NADH are generated by p-NAD-MDH in vicinity of the starch granule surface. Furthermore, even if NADH were generated by p-NAD-MDH, there is no reported mechanism by which the transfer or reducing equivalents to thioredoxins which in turn could reduce the target enzymes of starch degradation could occur. The thioredoxin reductase localized to the chloroplast (NTRC, Serrato et al. (2004)) shows high affinity for

6. General Conclusions 157 ______

NADPH and has not been shown to efficiently use NADH. The reducing power potentially created by p- NAD-MDH on the granule surface could therefore not be directly transferred to enzymes.

For these reasons, a function of p-NAD-MDH in altering the redox potential at the granule surface seems rather unlikely. Our lack of knowledge about the in vivo redox state of chloroplasts and its dynamics limits our understanding of the redox regulation of starch degrading enzymes.

6.4 Transcript Profiling of Non-Model Species Using RNA-Seq

Using an RNA-seq approach I have characterized of the transcriptome of a non-model species: Cecropia peltata . The wealth of sequence information generated by the NGS technologies presents special challenges ranging from the simple transfer and storage of large amounts of data to questions of sequence assembly and validation. I was able to create a high quality assembly, which could be annotated at high rate and with high confidence. The results generated using our RNA-seq approach can be interpreted in a biological context suggesting that my procedures were sound. I doubt however that I have made the most of the sequence data. For instance, more than 30% of all Illumina reads were lost due to the fact that they did not match any of the transcripts derived from the assembly procedures. In addition, the number of transcripts generated using Velvet and Oases is comparably small. Both are factors which could be improved. It might then be possible, to detect genes from specific pathways which are of interest which are currently not represented. Computing the best possible assembly, especially for a mixed-source data- set like mine, could easily extend beyond the scope of this work. The toolbox to work with sequence data is continuously evolving (Birzele et al., 2010; Miller et al., 2010; Surget-Groba and Montoya-Burgos, 2010). Choosing the optimal tools and using them optimally in terms of computing power requires close collaboration of biologists with bioinformaticians.

The current state of data analysis with the C. peltata RNA-seq approach allows solid statements about the transcriptional landscape in MBs as compared to leaves. I can say that expression of genes involved in cell wall metabolism is high, reflecting their rapid growth and emergence from trichilia. I can hypothesize about the energy supply in MBs, deriving from fermentation, rather than, or in addition to, oxidative respiration. I reached the primary goal and have been able to demonstrate that there are specific changes in the expression of starch synthetic genes, all supporting the shift from starch synthesis to the accumulation of WSPs in MBs.

The analysis of the RNAseq data I presented was limited and focused specifically on and starch metabolism. As with most –omics approaches, a lot more information about MBs awaits discovery in this data set. As mentioned in Chapter 1, C. peltata is an important medicinal plant and active research is

6. General Conclusions 158 ______ongoing to characterize the C . peltata compounds with anti-malarial or anti-diabetes activity. The sequence information we have generated will represent a resource for this research, even if only as sequence information that allows close inspection of genes coding for enzymes in interesting pathways.

6.5 The Limits of Transcriptomics – The Next Steps

Some of the observations I made when comparing my enzymatic and transcriptomics approaches illustrate the limitations of transcript-based expression analyses. While the SBE activity in MBs was increased as compared to leaves, related transcript levels were unchanged. It is important to remember that many steps are required until mRNA turns into enzymatic activity. (Gibon et al., 2004b) investigated the diurnal changes of activity levels and transcripts of a set of genes involved in primary and found hardly any correlation between activity and transcript levels. Later, Gibon et al. (2006), extended their analyses to metabolite measurements, concluding that changes in enzyme activities are delayed as compared to changes in transcript levels. They propose that series of recurring changes in transcript levels can be ‘integrated’ to change enzyme activities over time. The ‘reprogramming of metabolism’ in MBs as compared to C. peltata leaves likely requires fundamental alterations in the transcriptional landscape. As I was not looking for rapid transcriptional changes, my transcriptomic analyses seem valid for general investigations of the MB metabolism. To make definite conclusions, protein levels and, where possible, enzymatic activities should be used to corroborate the transcript data. I suggest to do this on two different levels. It can of course be corroborated whether changes of transcripts of genes in a given pathway (i.e. starch synthesis) are reflected by changes in protein levels by western blotting and immunodetection of the specific proteins. This approach is straight forward, providing a suitable antibody exists, and could be used to evaluate the hypothesis about the altered SSS expression pattern. However, this approach is currently limited by our ability to separate and detect the different SSS isoforms of C. peltata . Another approach is to investigate the proteome of MBs and C. peltata leaves by mass spectrometry. However, due to database constraints, it could be that spectra reported for the peptides from C. peltata might not be assigned to proteins. Methods for improving the detection rate of proteins from unsequenced organisms have however been suggested (Grossmann et al., 2007). In addition, sequence databases which have been obtained by NGS approaches have proven to be very useful to improve databases for MS/MS peptide searches in proteomic approaches including non-model species (Bräutigam et al., 2008; Delmotte et al., 2009). The sequence information I have generated for C. peltata will be used to improve databases. Thus, a proteomic approach has a high chance of success and will help complement and, hopefully, corroborate the transcriptomics data we already have generated.

6. General Conclusions 159 ______

6.6 Outlook

I have been able to add substantial knowledge to the processes of both starch degradation and synthesis My major findings (the BAM1-LSF1-MDH interaction and the reprogramming of starch biosynthetic enzymes for WSP accumulation in MBs of C. peltata ), provide important new insights in the field.

The results which I presented in Chapters 3 and 4 will help to disentangle the regulation and coordination of starch degradation in Arabidopsis . They suggest that protein-protein interactins contribute to the regulation of starch degrading enzymes and may offer a starting point into new research. It would be interesting to investigate the relative importance of the BAM-LSF1 interaction in different species and different tissues. It is unclear so far, to what extent the insights into these protein-protein interactions are transferrable to other species and tissues.

The research presented in Chapter 5 offers new hypotheses as to how starch structures could be modified in plants. The semi-crystalline nature of the starch granule represents an obstacle to its industrial processing. If large amounts of WSP could be synthesized in crop plants, industrial processing might be less laborious and energy consuming. The observations and conclusions from the work in C. peltata will have to be tested by varying expression levels of starch synthetic genes in Arabidopsis . Most importantly, it remains to be tested whether the observations made in C. peltata are transferrable to crop species in particular.

The tools to work on non-model species, some of which have been applied in this work, are being developed at high speed. NGS technologies will become cheaper and easier to use. With this development, the current limitations one faces when working with non-model species will certainly be overcome, offering more chances to study fascinating biological phenomena soon.

7. References 160 ______

7. References

Abad, M.C., Binderup, K., Rios-Steiner, J., Arni, R.K., Preiss, J., and Geiger, J.H. (2002). The X-ray crystallographic structure of Escherichia coli branching enzyme. Journal of Biological Chemistry 277, 42164-42170. Alexander, R.D., and Morris, P.C. (2006). A proteomic analysis of 14-3-3 binding proteins from developing barley grains. Proteomics 6, 1886-1896. Alonso, M.D., Lomako, J., Lomako, W.M., and Whelan, W.J. (1995). A new look at the biogenesis of glycogen. FASEB Journal 9, 1126-1137. Andrade-Cetto, A., and Wiedenfeld, H. (2001). Hypoglycemic effect of Cecropia obtusifolia on streptozotocin diabetic rats. Journal of Ethnopharmacology 78, 145-149. Andrade-Cetto, A., and Heinrich, M. (2005). Mexican plants with hypoglycaemic effect used in the treatment of diabetes. Journal of Ethnopharmacology 99, 325-348. Andrade-Cetto, A., and Vazquez, R.C. (2010). Gluconeogenesis, inhibition and phytochemical composition of two Cecropia species. Journal of Ethnopharmacology 130, 93-97. Aragão, D.M.O., Guarize, L., Lanini, J., da Costa, J.C., Garcia, R.M.G., and Scio, E. (2010). Hypoglycemic effects of Cecropia pachystachya in normal and alloxan-induced diabetic rats. Journal of Ethnopharmacology 128, 629-633. Archibald, A., Wright, A., Fleming, I.D., Liddle, A.M., Manners, D.J., and Mercer, G.A. (1961). Alpha-1,4-Glucosans - Absorption spectra of glycogen and amylopectine-iodine complexes. Journal of the Chemical Society , 55, 1183-1185. Ayers, J.E., and Creech, R.G. (1969). Genetic control of phytoglycogen accumulation in maize ( Zea- mays L. ). Crop Science 9, 739-745. Backhausen, J.E., Vetter, S., Baalmann, E., Kitzmann, C., and Scheibe, R. (1998). NAD-dependent malate dehydrogenase and glyceraldehyde 3-phosphate dehydrogenase isoenzymes play an important role in dark metabolism of various plastid types. Planta 205, 359-366. Ball, A.B.S. (1991). Regional Chemotherapy for Colorectal Hepatic Metastases Using Degradable Starch Microspheres - a Review. Acta Oncologica 30, 309-313. Ball, S., Guan, H.P., James, M., Myers, A., Keeling, P., Mouille, G., Buleon, A., Colonna, P., and Preiss, J. (1996). From glycogen to amylopectin: a model for the biogenesis of the plant starch granule. Cell 86, 349-352. Ball, S.G., and Morell, M.K. (2003). From bacterial glycogen to starch: understanding the biogenesis of the plant starch granule. Annual Review of Plant Biology 54, 207-233. Ball, S.G., van de Wal, M., and Visser, R.G.F. (1998). Progress in understanding the biosynthesis of amylose. Trends in Plant Science 3, 462-467. Ballicora, M.A., Frueauf, J.B., Fu, Y.B., Schurmann, P., and Preiss, J. (2000). Activation of the potato tuber ADP-glucose pyrophosphorylase by thioredoxin. Journal of Biological Chemistry 275, 1315-1320. Beatty, M.K., Rahman, A., Cao, H.P., Woodman, W., Lee, M., Myers, A.M., and James, M.G. (1999). Purification and molecular genetic characterization of ZPU1, a pullulanase-type starch- debranching enzyme from maize. Plant Physiology 119, 255-266. Beck, E., and Ziegler, P. (1989). Biosynthesis and degradation of starch in higher-plants. Annual Review of Plant Physiology and Plant Molecular Biology 40, 95-117. Beldade, P., Rudd, S., Gruber, J.D., and Long, A.D. (2006). A wing expressed sequence tag resource for Bicyclus anynana butterflies, an evo-devo model. BMC Genomics 7,130-134. Berkemeyer, M., Scheibe, R., and Ocheretina, O. (1998). A novel, non-redox-regulated NAD- dependent malate dehydrogenase from chloroplasts of Arabidopsis thaliana L . Journal of Biological Chemistry 273, 27927-27933.

7. References 161 ______

Bhat, R., Lahaye, T., and Panstruga, R. (2006). The visible touch : in planta visualization of protein- protein interactions by fluorophore-based methods. Plant Methods 2, 12-14. Bhattacharyya, M.K., Smith, A.M., Ellis, T.H.N., Hedley, C., and Martin, C. (1990). The wrinkled- seed character of pea described by Mendel is caused by a transposon-like insertion in a gene encoding starch-branching enzyme. Cell 60, 115-122. Birzele, F., Schaub, J., Rust, W., Clemens, C., Baum, P., Kaufmann, H., Weith, A., Schulz, T.W., and Hildebrandt, T. (2010). Into the unknown: expression profiling without genome sequence information in CHO by next generation sequencing. Nucleic Acids Research 38, 3999-4010. Black, R.C., Loerch, J.D., McArdle, F.J., and Creech, R.G. (1966). Genetic interactions affecting maize phytoglycogen and phytoglycogen-forming branching enzyme. Genetics 53, 661-&. Bläsing, O.E., Gibon, Y., Gunther, M., Hohne, M., Morcuende, R., Osuna, D., Thimm, O., Usadel, B., Scheible, W.R., and Stitt, M. (2005). Sugars and circadian regulation make major contributions to the global regulation of diurnal gene expression in Arabidopsis . Plant Cell 17, 3257-3281. Blauth, S.L., Kim, K.N., Klucinec, J., Shannon, J.C., Thompson, D., and Guiltinan, M. (2002). Identification of Mutator insertional mutants of starch-branching enzyme 1 ( sbe1 ) in Zea mays L . Plant Molecular Biology 48, 287-297. Blennow, A., and Engelsen, S.B. (2010). Helix-breaking news: fighting crystalline starch energy deposits in the cell. Trends in Plant Science 15, 236-240. Blennow, A., Engelsen, S.B., Munck, L., and Moller, B.L. (2000a). Starch molecular structure and phosphorylation investigated by a combined chromatographic and chemometric approach. Carbohydrate Polymers 41, 163-174. Blennow, A., Bay-Smidt, A.M., Olsen, C.E., and Moller, B.L. (2000b). The distribution of covalently bound phosphate in the starch granule in relation to starch crystallinity. International Journal of Biological Macromolecules 27, 211-218. Blennow, A., Engelsen, S.B., Nielsen, T.H., Baunsgaard, L., and Mikkelsen, R. (2002). Starch phosphorylation: a new front line in starch research. Trends in Plant Science 7, 445-450. Blüthgen, N., Kielbasa, S.M., and Herzel, H. (2005). Inferring combinatorial regulation of transcription in silico. Nucleic Acids Research 33, 272-279. Boren, M., Larsson, H., Falk, A., and Jansson, C. (2004). The barley starch granule proteome - internalized granule polypeptides of the mature endosperm. Plant Science 166, 617-626. Bräutigam, A., and Gowik, U. (2010). What can next generation sequencing do for you? Next generation sequencing as a valuable tool in plant research. Plant Biology 12, 831-841. Bräutigam, A., Shrestha, R.P., Whitten, D., Wilkerson, C.G., Carr, K.M., Fröhlich, J.E., and Weber, A.P.M. (2008). Low-coverage massively parallel pyrosequencing of cDNAs enables proteomics in non-model species: Comparison of a species-specific database generated by pyrosequencing with databases from related species for proteome analysis of pea chloroplast envelopes. Journal of Biotechnology 136, 44-53. Buléon, A., Colonna, P., Planchot, V., and Ball, S. (1998). Starch granules: structure and biosynthesis. International Journal of Biological Macromolecules 23, 85-112. Burton, R.A., Jenner, H., Carrangis, L., Fahy, B., Fincher, G.B., Hylton, C., Laurie, D.A., Parker, M., Waite, D., van Wegen, S., Verhoeven, T., and Denyer, K. (2002). Starch granule initiation and growth are altered in barley mutants that lack isoamylase activity. Plant Journal 31, 97-112. Bustos, R., Fahy, B., Hylton, C.M., Seale, R., Nebane, N.M., Edwards, A., Martin, C., and Smith, A.M. (2004). Starch granule initiation is controlled by a heteromultimeric isoamylase in potato tubers. Proceedings of the National Academy of Sciences of the United States of America 101, 2215-2220. Carr, P.D., Verger, D., Ashton, A.R., and Ollis, D.L. (1999). Chloroplast NADP-malate dehydrogenase: structural basis of light-dependent regulation of activity by thiol oxidation and reduction. Structure 7, 461-475.

7. References 162 ______

Caspar, T., Lin, T.P., Monroe, J., Bernhard, W., Spilatro, S., Preiss, J., and Somerville, C. (1989). Altered regulation of beta-amylase activity in mutants of Arabidopsis with lesions in starch metabolism. Proceedings of the National Academy of Sciences of the United States of America 86, 5830-5833. Chang, I.F., Curran, A., Woolsey, R., Quilici, D., Cushman, J.C., Mittler, R., Harmon, A., and Harper, J.F. (2009). Proteomic profiling of tandem affinity purified 14-3-3 protein complexes in Arabidopsis thaliana . Proteomics 9, 2967-2985. Cheng, C., Mu, J., Farkas, I., Huang, D.Q., Goebl, M.G., and Roach, P.J. (1995). Requirement of the slef-glucosylating initiator proteins GLG1P and GLG2P for glycogen accumulation in Saccharomyces cerevisiae Molecular and Cellular Biology 15, 6632-6640. Chia, T., Thorneycroft, D., Chapple, A., Messerli, G., Chen, J., Zeeman, S.C., Smith, S.M., and Smith, A.M. (2004). A cytosolic glucosyltransferase is required for conversion of starch to sucrose in Arabidopsis leaves at night. Plant Journal 37, 853-863. Citovsky, V., Lee, L.-Y., Vyas, S., Glick, E., Chen, M.-H., Vainstein, A., Gafni, Y., Gelvin, S.B., and Tzfira, T. (2006). Subcellular localization of interacting proteins by bimolecular fluorescence complementation in clanta. Journal of Molecular Biology 362, 1120-1131. Clough, S.J., and Bent, A.F. (1998). Floral dip: a simplified method for Agrobacterium-mediated transformation of Arabidopsis thaliana . Plant Journal 16, 735-743. Comparot-Moss, S., Kötting, O., Stettler, M., Edner, C., Graf, A., Weise, S.E., Streb, S., Lue, W.L., MacLean, D., Mahlow, S., Ritte, G., Steup, M., Chen, J., Zeeman, S.C., and Smith, A.M. (2010). A putative phosphatase, LSF1, is required for normal starch turnover in Arabidopsis leaves. Plant Physiology 152, 685-697. Conesa, A., Götz, S., Garcia-Gomez, J.M., Terol, J., Talon, M., and Robles, M. (2005). Blast2GO: a universal tool for annotation, visualization and analysis in functional genomics research. Bioinformatics 21, 3674-3676. Coppin, A., Varre, J.S., Lienard, L., Dauvillee, D., Guerardel, Y., Soyer-Gobillard, M.O., Buléon, A., Ball, S., and Tomavo, S. (2005). Evolution of plant-like crystalline storage polysaccharide in the protozoan parasite Toxoplasma gondii argues for a red alga ancestry. Journal of Molecular Evolution 60, 257-267. Craig, J., Lloyd, J.R., Tomlinson, K., Barber, L., Edwards, A., Wang, T.L., Martin, C., Hedley, C.L., and Smith, A.M. (1998). Mutations in the gene encoding starch synthase II profoundly alter amylopectin structure in pea embryos. Plant Cell 10, 413-426. Crevillén, P., Ballicora, M.A., Mérida, Á., Preiss, J., and Romero, J.M. (2003). The different large subunit isoforms of Arabidopsis thaliana ADP-glucose pyrophosphorylase confer distinct kinetic and regulatory properties to the heterotetrameric enzyme. Journal of Biological Chemistry 278, 28508-28515. Crevillén, P., Ventriglia, T., Pinto, F., Orea, A., Mérida, Á., and Romero, J.M. (2005). Differential pattern of expression and sugar regulation of Arabidopsis thaliana ADP-glucose pyrophosphorylase-encoding genes. Journal of Biological Chemistry 280, 8143-8149. Critchley, J.H., Zeeman, S.C., Takaha, T., Smith, A.M., and Smith, S.M. (2001). A critical role for disproportionating enzyme in starch breakdown is revealed by a knock-out mutation in Arabidopsis . Plant Journal 26, 89-100. Cunha, M.R.H., Davel, A.P.C., Padilha, A.S., Mauad, H., and Cabral, A.M. (2004). Evaluation of the hypotensive effect induced by aqueous fraction of ethanolic extract of the Cecropia peltata (Embauba) in rats. Journal of Hypertension 22, 139-139. Cvetic, T., Veljovic-Jovanovic, S., and Vucinic, Z. (2008). Characterization of NAD-dependent malate dehydrogenases from spinach leaves. Protoplasma 232, 247-253. Dauvillée, D., Colleoni, C., Mouille, G., Buléon, A., Gallant, D.J., Bouchet, B., Morell, M.K., d'Hulst, C., Myers, A.M., and Ball, S.G. (2001a). Two loci control phytoglycogen production in the monocellular green alga Chlamydomonas reinhardtii . Plant Physiology 125, 1710-1722.

7. References 163 ______

Dauvillée, D., Colleoni, C., Shaw, E., Mouille, G., D'Hulst, C., Morell, M., Samuel, M.S., Bouchet, B., Gallant, D.J., Sinskey, A., and Ball, S. (1999). Novel, starch-like polysaccharides are synthesized by an unbound form of granule-bound starch synthase in glycogen-accumulating mutants of Chlamydomonas reinhardtii . Plant Physiology 119, 321-329. Dauvillée, D., Colleoni, C., Mouille, G., Morell, M.K., d'Hulst, C., Wattebled, F., Lienard, L., Delvalle, D., Ral, J.P., Myers, A.M., and Ball, S.G. (2001). Biochemical characterization of wild-type and mutant isoamylases of Chlamydomonas reinhardtii supports a function of the multimeric enzyme organization in amylopectin maturation. Plant Physiology 125, 1723-1731. Davis, J.P., Supatcharee, N., Khandelwal, R.L., and Chibbar, R.N. (2003). Synthesis of novel starches in planta : opportunities and challenges. Starch-Stärke 55, 107-120. Delatte, T., Trevisan, M., Parker, M.L., and Zeeman, S.C. (2005). Arabidopsis mutants Atisa1 and Atisa2 have identical phenotypes and lack the same multimeric isoamylase, which influences the branch point distribution of amylopectin during starch synthesis. Plant Journal 41, 815-830. Delatte, T., Umhang, M., Trevisan, M., Eicke, S., Thorneycroft, D., Smith, S.M., and Zeeman, S.C. (2006). Evidence for distinct mechanisms of starch granule breakdown in plants. Journal of Biological Chemistry 281, 12050-12059. Delmotte, N., Knief, C., Chaffron, S., Innerebner, G., Roschitzki, B., Schlapbach, R., von Mering, C., and Vorholt, J.A. (2009). Community proteogenomics reveals insights into the physiology of phyllosphere bacteria. Proceedings of the National Academy of Sciences 106, 16428-16433. Delvalle, D., Dumez, S., Wattebled, F., Roldan, I., Planchot, V., Berbezy, P., Colonna, P., Vyas, D., Chatterjee, M., Ball, S., Merida, A., and D'Hulst, C. (2005). Soluble starch synthase I: a major determinant for the synthesis of amylopectin in Arabidopsis thaliana leaves. Plant J 43, 398-412. Deschamps, P., Guillebeault, D., Devassine, J., Dauvillee, D., Haebel, S., Steup, M., Buleon, A., Putaux, J.L., Slomianny, M.C., Colleoni, C., Devin, A., Plancke, C., Tomavo, S., Derelle, E., Moreau, H., and Ball, S. (2008). The heterotrophic dinoflagellate Crypthecodinium cohnii defines a model genetic system to investigate cytoplasmic starch synthesis. Eukaryotic Cell 7, 872-880. Dinges, J.R., Colleoni, C., James, M.G., and Myers, A.M. (2003). Mutational analysis of the pullulanase-type debranching enzyme of maize indicates multiple functions in starch metabolism. Plant Cell 15, 666-680. Dumez, S., Wattwbled, F., Dauville, D., Delvalle, D., Planchot, V., Ball, S., and D'Hulst, C. (2006). Mutants of Arabidopsis lacking starch branching enzyme II substitute plastidial starch synthesis by cytoplasmic maltose accumulation. Plant Cell 18 , 2694-2709 Durek, P., Schmidt, R., Heazlewood, J.L., Jones, A., MacLean, D., Nagel, A., Kersten, B., and Schulze, W.X. (2010). PhosPhAt: the Arabidopsis thaliana phosphorylation site database. An update. Nucleic Acids Research 38, 828-834. Edner, C., Li, J., Albrecht, T., Mahlow, S., Hejazi, M., Hussain, H., Kaplan, F., Guy, C., Smith, S.M., Steup, M., and Ritte, G. (2007). Glucan, water dikinase activity stimulates breakdown of starch granules by plastidial beta-amylases. Plant Physiology 145, 17-28. Edwards, A., Fulton, D.C., Hylton, C.M., Jobling, S.A., Gidley, M., Rossner, U., Martin, C., and Smith, A.M. (1999). A combined reduction in activity of starch synthases II and III of potato has novel effects on the starch of tubers. Plant Journal 17, 251-261. Elkins, J.M., Gileadi, C., Shrestha, L., Phillips, C., Wang, J., Muniz, J.R.C., and Doyle, D.A. (2010). Unusual binding interactions in PDZ domain crystal structures help explain binding mechanisms. Protein Science 19, 731-741. Faveri, S.B., and Vasconcelos, H.L. (2004). The Azteca-Cecropia Association: Are ants always necessary for their host plants? Biotropica 36, 641-646. Fernbach, A. (1904). Quelques observations sur la composition de l’amidon de pommes de terre. C. R. Acad. Sci. Paris, Life Sciences 138, 428–430.

7. References 164 ______

Fettke, J., Eckermann, N., Poeste, S., Pauly, M., and Steup, M. (2004). The glycan substrate of the cytosolic ( Pho 2 ) phosphorylase isozyme from Pisum sativum L .: identification, linkage analysis and subcellular localization. Plant Journal 39, 933-946. Fettke, J., Malinova, I., Albrecht, T., Hejazi, M., and Steup, M. (2010a). Glucose 1-phosphate transport into protoplasts and chloroplasts from leaves of Arabidopsis thaliana . Plant Physiology , in press. Fettke, J., Hejazi, M., Smirnova, J., Hochel, E., Stage, M., and Steup, M. (2009). Eukaryotic starch degradation: integration of plastidial and cytosolic pathways. Journal of Experimental Botany 60, 2907-2922. Fettke, J., Albrecht, T., Hejazi, M., Mahlow, S., Nakamura, Y., and Steup, M. (2010b). Glucose 1- phosphate is efficiently taken up by potato ( Solanum tuberosum ) tuber parenchyma cells and converted to reserve starch granules. New Phytologist 185, 663-675. Fisher, D.K., Gao, M., Kim, K.N., Boyer, C.D., and Guiltinan, M.J. (1996). Two closely related cDNAs encoding starch branching enzyme from Arabidopsis thaliana . Plant Molecular Biology 30, 97-108. Fleming, T.H., and Williams, C.F. (1990). Phenology, seed dispersal, and recruitment in Cecropia- peltata (Moraceae ) in Costa Rican tropical dry forest. Journal of Tropical Ecology 6, 163-178. Folgarait, P.J., Johnson, H.L., and Davidson, D.W. (1994). Responses of Cecropia to experimental removal of Müllerian bodies. Functional Ecology 8, 22-28. Fu, Y.B., Ballicora, M.A., Leykam, J.F., and Preiss, J. (1998). Mechanism of reductive activation of potato tuber ADP-glucose pyrophosphorylase. Journal of Biological Chemistry 273, 25045- 25052. Fujita, N., Kubo, A., Francisco, P.B., Nakakita, M., Harada, K., Minaka, N., and Nakamura, Y. (1999). Purification, characterization, and cDNA structure of isoamylase from developing endosperm of rice. Planta 208, 283-293. Fulton, D.C., Stettler, M., Mettler, T., Vaughan, C.K., Li, J., Francisco, P., Gil, D., Reinhold, H., Eicke, S., Messerli, G., Dorken, G., Halliday, K., Smith, A.M., Smith, S.M., and Zeeman, S.C. (2008). BETA-AMYLASE 4, a noncatalytic protein required for starch breakdown, acts upstream of three active beta-amylases in Arabidopsis chloroplasts. Plant Cell 20, 1040-1058. Gallant, D.J., Bouchet, B., and Baldwin, P.M. (1997). Microscopy of starch: Evidence of a new level of granule organization. Carbohydrate Polymers 32, 177-191. Gao, M., Fisher, D.K., Kim, K.N., Shannon, J.C., and Guiltinan, M.J. (1997). Independent genetic control of maize starch-branching enzymes IIa and IIb - Isolation and characterization of a Sbe2a cDNA. Plant Physiology 114, 69-78. Gass, N., Glagotskaia, T., Mellema, S., Stuurman, J., Barone, M., Mandel, T., Roessner-Tunali, U., and Kuhlemeier, C. (2005). Pyruvate decarboxylase provides growing pollen tubes with a competitive advantage in petunia. Plant Cell 17, 2355-2368. Geigenberger, P., Kolbe, A., and Tiessen, A. (2005). Redox regulation of carbon storage and partitioning in response to light and sugars. Journal of Experimental Botany 56, 1469-1479. Gentleman, R., Carey, V., Bates, D., Bolstad, B., Dettling, M., Dudoit, S., Ellis, B., Gautier, L., Ge, Y., Gentry, J., Hornik, K., Hothorn, T., Huber, W., Iacus, S., Irizarry, R., Leisch, F., Li, C., Maechler, M., Rossini, A., Sawitzki, G., Smith, C., Smyth, G., Tierney, L., Yang, J., and Zhang, J. (2004). Bioconductor: open software development for computational biology and bioinformatics. Genome Biology 5, R80. Gibon, Y., Usadel, B., Bläsing, O.E., Kamlage, B., Höhne, M., Trethewey, R., and Stitt, M. (2006). Integration of metabolite with transcript and enzyme activity profiling during diurnal cycles in Arabidopsis rosettes. Genome Biology 7(8): R76

7. References 165 ______

Gibon, Y., Pyl, E.T., Sulpice, R., Lunn, J.E., Höhne, M., Gunther, M., and Stitt, M. (2009). Adjustment of growth, starch turnover, protein content and central metabolism to a decrease of the carbon supply when Arabidopsis is grown in very short photoperiods. Plant Cell and Environment 32, 859-874. Gibon, Y., Bläsing, O.E., Palacios-Rojas, N., Pankovic, D., Hendriks, J.H.M., Fisahn, J., Hohne, M., Gunther, M., and Stitt, M. (2004a). Adjustment of diurnal starch turnover to short days: depletion of sugar during the night leads to a temporary inhibition of carbohydrate utilization, accumulation of sugars and post-translational activation of ADP-glucose pyrophosphorylase in the following light period. Plant Journal 39, 847-862. Gibon, Y., Bläsing, O.E., Hannemann, J., Carillo, P., Hohne, M., Hendriks, J.H.M., Palacios, N., Cross, J., Selbig, J., and Stitt, M. (2004b). A robot-based platform to measure multiple enzyme activities in Arabidopsis using a set of cycling assays: comparison of changes of enzyme activities and transcript levels during diurnal cycles and in prolonged darkness. Plant Cell 16, 3304-3325. Götz, S., Garcia-Gomez, J.M., Terol, J., Williams, T.D., Nagaraj, S.H., Nueda, M.J., Robles, M., Talon, M., Dopazo, J., and Conesa, A. (2008). High-throughput functional annotation and data mining with the Blast2GO suite. Nucleic Acids Research 36, 3420-3435. Graf, A., Schlereth, A., Stitt, M., and Smith, A.M. (2010). Circadian control of carbohydrate availability for growth in Arabidopsis plants at night. Proceedings of the National Academy of Sciences of the United States of America 107, 9458-9463. Grimaud, F., Rogniaux, H., James, M.G., Myers, A.M., and Planchot, V. (2008). Proteome and phosphoproteome analysis of starch granule-associated proteins from normal maize and mutants affected in starch biosynthesis. Journal of Experimental Botany 59, 3395-3406. Grossmann, J., Fischer, B., Baerenfaller, K., Owiti, J., Buhmann, J.M., Gruissem, W., and Baginsky, S. (2007). A workflow to increase the detection rate of proteins from unsequenced organisms in high-throughput proteomics experiments. Proteomics 7, 4245-4254. Haebel, S., Hejazi, M., Frohberg, C., Heydenreich, M., and Ritte, G. (2008). Mass spectrometric quantification of the relative amounts of C6 and C3 position phosphorylated glucosyl residues in starch. Analytical Biochemistry 379, 73-79. Hancock, J.T., Henson, D., Nyirenda, M., Desikan, R., Harrison, J., Lewis, M., Hughes, J., and Neill, S.J. (2005). Proteomic identification of glyceraldehyde 3-phosphate dehydrogenase as an inhibitory target of hydrogen peroxide in Arabidopsis . Plant Physiology and Biochemistry 43, 828-835. Harn, C., Knight, M., Ramakrishnan, A., Guan, H.P., Keeling, P.L., and Wasserman, B.P. (1998). Isolation and characterization of the zSSIIa and zSSIIb starch synthase cDNA clones from maize endosperm. Plant Molecular Biology 37, 639-649. Heazlewood, J.L., Durek, P., Hummel, J., Selbig, J., Weckwerth, W., Walther, D., and Schulze, W.X. (2008). PhosPhAt: a database of phosphorylation sites in Arabidopsis thaliana and a plant- specific phosphorylation site predictor. Nucleic Acids Research 36, D1015-D1021. Hejazi, M., Fettke, J., Paris, O., and Steup, M. (2009). The Two Plastidial Starch-Related Dikinases Sequentially Phosphorylate Glucosyl Residues at the Surface of Both the A- and B-Type Allomorphs of Crystallized Maltodextrins But the Mode of Action Differs. Plant Physiology 150, 962-976. Hejazi, M., Fettke, J., Kötting, O., Zeeman, S.C., and Steup, M. (2010). The Laforin-Like Dual- Specificity Phosphatase SEX4 from Arabidopsis Hydrolyzes Both C6-and C3-Phosphate Esters Introduced by Starch-Related Dikinases and Thereby Affects Phase Transition of alpha-Glucans. Plant Physiology 152, 711-722. Hejazi, M., Fettke, J., Haebel, S., Edner, C., Paris, O., Frohberg, C., Steup, M., and Ritte, G. (2008). Glucan, water dikinase phosphorylates crystalline maltodextrins and thereby initiates solubilization. Plant Journal 55, 323-334.

7. References 166 ______

Hendriks, J.H.M., Kolbe, A., Gibon, Y., Stitt, M., and Geigenberger, P. (2003). ADP-glucose pyrophosphorylase is activated by posttranslational redox-modification in response to light and to sugars in leaves of Arabidopsis and other plant species. Plant Physiology 133, 838-849. Hennen-Bierwagen, T.A., Lin, Q., Grimaud, F., Planchot, V., Keeling, P.L., James, M.G., and Myers, A.M. (2009). Proteins from multiple metabolic pathways associate with starch biosynthetic enzymes in high molecular weight complexes: A model for regulation of carbon allocation in maize amyloplasts. Plant Physiology 149, 1541-1559. Hennen-Bierwagen, T.A., Liu, F., Marsh, R.S., Kim, S., Gan, Q., Tetlow, I.J., Emes, M.J., James, M.G., and Myers, A.M. (2008). Starch biosynthetic enzymes from developing maize endosperm associate in multisubunit complexes. Plant Physiology 146, 1892-1908. Hill, L.M., and Smith, A.M. (1991). Evidence that glucose-6-phosphate is imported as the substrate for starch synthesis by the plastids of developing pea embryos. Planta 185, 91-96. Hill, L.M., and Smith, A.M. (1995). Coupled movements of glucose-6-phosphate and triose phosphate through the envelopes of plastids from developing embryos of pea ( Pisum-Sativum L ). Journal of Plant Physiology 146, 411-417. Hörtensteiner, S., and Feller, U. (2002). Nitrogen metabolism and remobilization during senescence. Journal of Experimental Botany 53, 927-937. Hu, C.-D., Chinenov, Y., and Kerppola, T.K. (2002). Visualization of interactions among bZIP and Rel family proteins in living cells using bimolecular fluorescence complementation. Molecular Cell 9, 789-798. Hussain, H., Mant, A., Seale, R., Zeeman, S., Hinchliffe, E., Edwards, A., Hylton, C., Bornemann, S., Smith, A.M., Martin, C., and Bustos, R. (2003). Three isoforms of isoamylase contribute different catalytic properties for the debranching of potato glucans. Plant Cell 15, 133-149. Imsande, J., Berkemeyer, M., Scheibe, R., Schumann, U., Gietl, C., and Palmer, R.G. (2001). A soybean plastid-targeted NADH-malate dehydrogenase: Cloning and expression analyses. American Journal of Botany 88, 2136-2142. James, M.G., Robertson, D.S., and Myers, A.M. (1995). Characterization of the Maize Gene Sugary1 , a Determinant of Starch Composition in Kernels. Plant Cell 7, 417-429. James, M.G., Denyer, K., and Myers, A.M. (2003). Starch synthesis in the cereal endosperm. Current Opinion in Plant Biology 6, 215 - 222. Jane, J.L., Kasemsuwan, T., Leas, S., Zobel, H., and Robyt, J.F. (1994). Anthology of starch granule morphology by scanning electron-microscopy. Starch-Stärke 46, 121-129. Janzen, D.H. (1969). Allelopathy by myrmecophytes – ant azteca as an allelopathhic agent of Cecropia . Ecology 50, 147-&. Jenkins, J.P.J., Cameron, R.E., and Donald, A.M. (1993). A universal feature in the structure of starch granules from different botanical sources. Starch-Stärke 45, 417-420. Jobling, S. (2004). Improving starch for food and industrial applications. Current Opinion in Plant Biology 7, 210-218. Jobling, S.A., Schwall, G.P., Westcott, R.J., Sidebottom, C.M., Debet, M., Gidley, M.J., Jeffcoat, R., and Safford, R. (1999). A minor form of starch branching enzyme in potato ( Solanum tuberosum L.) tubers has a major effect on starch structure: cloning and characterisation of multiple forms of SBE A. Plant Journal 18, 163-171. Keller, A., Nesvizhskii, A.I., Kolker, E., and Aebersold, R. (2002). Empirical statistical model to estimate the accuracy of peptide identifications made by MS/MS and database search. Analytical Chemistry 74, 5383-5392. Kempa, S., Krasensky, J., Dal Santo, S., Kopka, J., and Jonak, C. (2008). A Central Role of Abscisic Acid in Stress-Regulated Carbohydrate Metabolism. PLoS One 3, e3935. Kerk, D., Conley, T.R., Rodriguez, F.A., Tran, H.T., Nimick, M., Muench, D.G., and Moorhead, G.B.G. (2006). A chloroplast-localized dual-specificity protein phosphatase in Arabidopsis contains a phylogenetically dispersed and ancient carbohydrate-binding domain, which binds the polysaccharide starch. Plant Journal 46, 400-413.

7. References 167 ______

Koizumi, K., Fukuda, M., and Hizukuri, S. (1991). Estimation of the distributions of chain-length of amylopectins by high-performance liquid-chromatography with pulsed amperometric detection. Journal of Chromatography 585, 233-238. Kolbe, A., Tiessen, A., Schluepmann, H., Paul, M., Ulrich, S., and Geigenberger, P. (2005). Trehalose 6-phosphate regulates starch synthesis via posttranslational redox activation of ADP- glucose pyrophosphorylase. Proceedings of the National Academy of Sciences of the United States of America 102, 11118-11123. Koshiba, T., and Minamikawa, T. (1981). Purification by affinity-chromatography of alpha-amylase - a main amylase in cotyledons of germinating Vigna mungo seeds. Plant and Cell Physiology 22, 979-987. Kötting, O., Kossmann, J., Zeeman, S.C., and Lloyd, J.R. (2010). Regulation of starch metabolism: the age of enlightenment? Current Opinion in Plant Biology 13, 321-329. Kötting, O., Pusch, K., Tiessen, A., Geigenberger, P., Steup, M., and Ritte, G. (2005). Identification of a novel enzyme required for starch metabolism in Arabidopsis leaves. The phosphoglucan, water dikinase. Plant Physiology 137, 242-252. Kötting, O., Santelia, D., Edner, C., Eicke, S., Marthaler, T., Gentry, M.S., Comparot-Moss, S., Chen, J., Smith, A.M., Steup, M., Ritte, G., and Zeeman, S.C. (2009). STARCH EXCESS4 is a laforin-like phosphoglucan phosphatase required for starch degradation in Arabidopsis thaliana . Plant Cell 21, 334-346. Kruger, N.J., and Ap Rees, T. (1983). Maltose metabolism by pea-chloroplasts. Planta 158, 179-184. Kubo, A., Fujita, N., Harada, K., Matsuda, T., Satoh, H., and Nakamura, Y. (1999). The starch- debranching enzymes isoamylase and pullulanase are both involved in amylopectin biosynthesis in rice endosperm. Plant Physiology 121, 399-409. Kubo, A., Colleoni, C., Dinges, J.R., Lin, Q., Lappe, R.R., Rivenbark, J.G., Meyer, A.J., Ball, S.G., James, M.G., Hennen-Bierwagen, T.A., and Myers, A.M. (2010). Functions of heteromeric and homomeric isoamylase-type starch-debranching enzymes in developing maize endosperm. Plant Physiology 153, 956-969. Kunz, H.H., Hausler, R.E., Fettke, J., Herbst, K., Niewiadomski, P., Gierth, M., Bell, K., Steup, M., Flugge, U.I., and Schneider, A. (2010). The role of plastidial glucose-6-phosphate/phosphate translocators in vegetative tissues of Arabidopsis thalian a mutants impaired in starch biosynthesis. Plant Biology 12, 115-128. Kurtz, S., Phillippy, A., Delcher, A., Smoot, M., Shumway, M., Antonescu, C., and Salzberg, S. (2004). Versatile and open software for comparing large genomes. Genome Biology 5, R12. Laby, R.J., Kim, D., and Gibson, S.I. (2001). The ram1 mutant of Arabidopsis exhibits severely decreased beta-amylase activity. Plant Physiology 127, 1798-1807. Lao, N.T., Schoneveld, O., Mould, R.M., Hibberd, J.M., Gray, J.C., and Kavanagh, T.A. (1999). An Arabidopsis gene encoding a chloroplast-targeted beta-amylase. Plant Journal 20, 519-527. Lawson, T. (2009). Guard cell photosynthesis and stomatal function. New Phytologist 181, 13-34. Li, B., Servaites, J.C., and Geiger, D.R. (1992). Characterization and subcellular-localization of debranching enzyme and endoamylase from leaves of sugar-beet. Plant Physiology 98, 1277- 1284. Li, J., Francisco, P., Zhou, W.X., Edner, C., Steup, M., Ritte, G., Bond, C.S., and Smith, S.M. (2009). Catalytically-inactive beta-amylase BAM4 required for starch breakdown in Arabidopsis leaves is a starch-binding-protein. Archives of Biochemistry and Biophysics 489, 92-98. Lindeboom, N., Chang, P.R., and Tyler, R.T. (2004). Analytical, biochemical and physicochemical aspects of starch granule size, with emphasis on small granule starches: A review. Starch-Stärke 56, 89-99. Lister, R., O'Malley, R.C., Tonti-Filippini, J., Gregory, B.D., Berry, C.C., Millar, A.H., and Ecker, J.R. (2008). Highly integrated single-base resolution maps of the epigenome in Arabidopsis . Cell 133, 523-536.

7. References 168 ______

Ljerka Kunst. (1998). Preparation of Physiologically Active Chloroplasts from Arabidopsis . In Methods in Molecular Biology, M.-Z.a.J. Salinas, ed (Totowa: Humana Press) 82 , 43-48. Lloyd, J.R., Blennow, A., Burhenne, K., and Kossmann, J. (2004). Repression of a novel isoform of disproportionating enzyme (stDPE2) in potato leads to inhibition of starch degradation in leaves but not tubers stored at low temperature. Plant Physiology 134, 1347-1354. Lohrig, K., Muller, B., Davydova, J., Leister, D., and Wolters, D.A. (2009). Phosphorylation site mapping of soluble proteins: bioinformatical filtering reveals potential plastidic phosphoproteins in Arabidopsis thaliana . Planta 229, 1123-1134. Lorberth, R., Ritte, G., Willmitzer, L., and Kossmann, J. (1998). Inhibition of a starch-granule-bound protein leads to modified starch and repression of cold sweetening. Nature Biotechnology 16, 473-477. Luan, S. (2003). Protein phosphatases in plants. Annual Review of Plant Biology 54, 63-92. Manners, D.J. (1991). Recent developments in our understanding of glycogen structure. Carbohydrate Polymers 16, 37-82. Margulies, M., Egholm, M., Altman, W.E., Attiya, S., Bader, J.S., Bemben, L.A., Berka, J., Braverman, M.S., Chen, Y.J., Chen, Z.T., Dewell, S.B., Du, L., Fierro, J.M., Gomes, X.V., Godwin, B.C., He, W., Helgesen, S., Ho, C.H., Irzyk, G.P., Jando, S.C., Alenquer, M.L.I., Jarvie, T.P., Jirage, K.B., Kim, J.B., Knight, J.R., Lanza, J.R., Leamon, J.H., Lefkowitz, S.M., Lei, M., Li, J., Lohman, K.L., Lu, H., Makhijani, V.B., McDade, K.E., McKenna, M.P., Myers, E.W., Nickerson, E., Nobile, J.R., Plant, R., Puc, B.P., Ronan, M.T., Roth, G.T., Sarkis, G.J., Simons, J.F., Simpson, J.W., Srinivasan, M., Tartaro, K.R., Tomasz, A., Vogt, K.A., Volkmer, G.A., Wang, S.H., Wang, Y., Weiner, M.P., Yu, P.G., Begley, R.F., and Rothberg, J.M. (2005). Genome sequencing in microfabricated high-density picolitre reactors. Nature 437, 376-380. Marioni, J.C., Mason, C.E., Mane, S.M., Stephens, M., and Gilad, Y. (2008). RNA-seq: an assessment of technical reproducibility and comparison with gene expression arrays. Genome Research 18, 1509-1517. Marshall, J.J., and Rickson, F.R. (1973). Characterization of alpha-D-glucan from plastids of Cecropia peltata as a glycogen-zype polysaccharide. Carbohydrate Research 28, 31-37. Meléndez, R., Meléndez-Hevia, E., Mas, F., Mach, J., and Cascante, M. (1998). Physical constraints in the synthesis of glycogen that influence its structural homogeneity: a two-dimensional approach. Biophysical Journal 75, 106-114. Mettler, T. (2006). Distinguishing the roles of beta-amylases in transitory starch breakdown in Arabidopsis thaliana leaves. Diplomarbeit der philosophisch-naturwissenschaftlichen Fakultät der Universität Bern. Metzker, M.L. (2010). Applications of next generation sequencing. Sequencing technologies - the next generation. Nature Reviews Genetics 11, 31-46. Mikkelsen, R., Mutenda, K.E., Mant, A., Schurmann, P., and Blennow, A. (2005). Alpha-glucan, water dikinase (GWD): a plastidic enzyme with redox-regulated and coordinated catalytic activity and binding affinity. Proceedings of the National Academy of Sciences of the United States of America U S A 102, 1785-1790. Miller, J.R., Koren, S., and Sutton, G. (2010). Assembly algorithms for next-generation sequencing data. Genomics 95, 315-327. Monroe, J.D., and Preiss, J. (1990). Purification of a beta-amylase that accumulates in Arabidopsis- thaliana mutants defective in starch metabolism. Plant Physiology 94, 1033-1039. Mortazavi, A., Williams, B.A., Mccue, K., Schaeffer, L., and Wold, B. (2008). Mapping and quantifying mammalian transcriptomes by RNA-Seq. Nature Methods 5, 621-628. Müller, F. (1876). Über das Haarkissen am Blattstiel der Imbauba ( Cecropia ), das Gemüsebeet der Imbauba-Ameise. Jenaische Zeitschrift für Medizin und Naturwissenschaft 10, 281-286. Nagalakshmi, U., Waern, K., and Snyder, M. (2010). RNA-Seq: a method for comprehensive transcriptome analysis. Current Protocols in Molecular Biology 4, 11-13.

7. References 169 ______

Nakamura, Y., Takahashi, J., Sakurai, A., Inaba, Y., Suzuki, E., Nihei, S., Fujiwara, S., Tsuzuki, M., Miyashita, H., Ikemoto, H., Kawachi, M., Sekiguchi, H., and Kurano, N. (2005). Some cyanobacteria synthesize semi-amylopectin type alpha-polyglucans instead of glycogen. Plant and Cell Physiology 46, 539-545. Nesvizhskii, A.I., Keller, A., Kolker, E., and Aebersold, R. (2003). A statistical model for identifying proteins by tandem mass spectrometry. Analytical Chemistry 75, 4646-4658. Neuhaus, H.E., and Schulte, N. (1996). Starch degradation in chloroplasts isolated from C-3 or CAM (crassulacean acid metabolism)-induced Mesembryanthemum crystallinum L . Biochemical Journal 318, 945-953. Nicasio, P., Aguilar-Santamaria, L., Aranda, E., Ortiz, S., and Gonzalez, M. (2005). Hypoglycemic effect and chlorogenic acid content in two Cecropia species. Phytotherapy Research 19, 661-664. Niewiadomski, P., Knappe, S., Geimer, S., Fischer, K., Schulz, B., Unte, U.S., Rosso, M.G., Ache, P., Flugge, U.-I., and Schneider, A. (2005). The Arabidopsis plastidic glucose 6- phosphate/phosphate translocator GPT1 is essential for pollen maturation and embryo sac development. Plant Cell 17, 760-775. Niittylä, T., Messerli, G., Trevisan, M., Chen, J., Smith, A.M., and Zeeman, S.C. (2004). A previously unknown maltose transporter essential for starch degradation in leaves. Science 303, 87-89. Niittylä, T., Comparot-Moss, S., Lue, W.L., Messerli, G., Trevisan, M., Seymour, M.D.J., Gatehouse, J.A., Villadsen, D., Smith, S.M., Chen, J.C., Zeeman, S.C., and Alison, M.S. (2006). Similar protein phosphatases control starch metabolism in plants and glycogen metabolism in mammals. Journal of Biological Chemistry 281, 11815-11818. Noda, T., Kimura, T., Otani, M., Ideta, O., Shimada, T., Saito, A., and Suda, I. (2002). Physicochemical properties of amylose-free starch from transgenic sweet potato. Carbohydrate Polymers 49, 253-260. O'Sullivan, A.C., and Perez, S. (1999). The relationship between internal chain length of amylopectin and crystallinity in starch. Biopolymers 50, 381-390. Ocheretina, O., and Scheibe, R. (1997). Cloning and sequence analysis of cDNAs encoding plant cytosolic malate dehydrogenase. Gene 199, 145-148. Oecking, C., and Jaspert, N. (2009). Plant 14-3-3 proteins catch up with their mammalian orthologs. Current Opinion in Plant Biology 12, 760-765. Olinares, P.D.B., Ponnala, L., and van Wijk, K.J. (2010). Megadalton complexes in the chloroplast stroma of Arabidopsis thaliana characterized by size exclusion chromatography, mass spectrometry, and hierarchical clustering. Molecular & Cellular Proteomics 9, 1594-1615. Ottmann, C., Yasmin, L., Weyand, M., Veesenmeyer, J.L., Diaz, M.H., Palmer, R.H., Francis, M.S., Hauser, A.R., Wittinghofer, A., and Hallberg, B. (2007). Phosphorylation-independent interaction between 14-3-3 and exoenzyme S: from structure to pathogenesis. EMBO Journal 26, 902-913. Pan, D., and Nelson, O.E. (1984). A debranching enzyme deficiency in endosperms of the sugary-1 mutants of maize. Plant Physiology 74, 324-328. Paszkiewicz, K., and Studholme, D.J. (2010). De novo assembly of short sequence reads. Briefings in Bioinformatics 11, 457-472. Peltier, J.B., Cai, Y., Sun, Q., Zabrouskov, V., Giacomelli, L., Rudella, A., Ytterberg, A.J., Rutschow, H., and van Wijk, K.J. (2006). The oligomeric stromal proteome of Arabidopsis thaliana chloroplasts. Molecular & Cellular Proteomics 5, 114-133. Perez-Guerrero, C., Herrera, M.D., Ortiz, R., de Sotomayor, M.A., and Fernandez, M.A. (2001). A pharmacological study of Cecropia obtusifolia Bertol aqueous extract. Journal of Ethnopharmacology 76, 279-284. Perez, S., and Bertoft, E. (2010). The molecular structures of starch components and their contribution to the architecture of starch granules: A comprehensive review. Starch-Stärke 62, 389-420.

7. References 170 ______

Ponting, C.P. (1997). Evidence for PDZ domains in bacteria, yeast, and plants. Protein Science 6, 464- 468. Preiss, J. (1982). Regulation of the biosynthesis and degradation of starch. Annual Review of Plant Physiology and Plant Molecular Biology 33, 431-454. Ral, J.P., Colleoni, C., Wattebled, F., Dauvillee, D., Nempont, C., Deschamps, P., Li, Z.Y., Morell, M.K., Chibbar, R., Purton, S., d'Hulst, C., and Ball, S.G. (2006). Circadian clock regulation of starch metabolism establishes GBSSI as a major contributor to amylopectin synthesis in Chlamydomonas reinhardtii . Plant Physiology 142, 305-317. Reiland, S., Messerli, G., Bärenfaller, K., Gerrits, B., Endler, A., Grossmann, J., Gruissem, W., and Baginsky, S. (2009). Large-scale Arabidopsis phosphoproteome profiling reveals novel chloroplast kinase substrates and phosphorylation networks. Plant Physiology 150, 889-903. Rickson, F.R. (1971). Glycogen plastids in Müllerian body cells of Cecropia peltata – a higher green plant. Science 173, 344-347. Rickson, F.R. (1976a). Anatomical development of leaf trichilium and Müllerian bodies of Cecropia peltata L . American Journal of Botany 63, 1266-1271. Rickson, F.R. (1976b). Ultrastructural differentiation of Müllerian body glycogen plastid of Cecropia Peltata L . American Journal of Botany 63, 1272-1279. Ritte, G., Scharf, A., Eckermann, N., Haebel, S., and Steup, M. (2004). Phosphorylation of transitory starch is increased during degradation. Plant Physiology 135, 2068-2077. Ritte, G., Lloyd, J.R., Eckermann, N., Rottmann, A., Kossmann, J., and Steup, M. (2002). The starch-related R1 protein is an alpha-glucan, water dikinase. Proceedings of the National Academy of Sciences of the United States of America 99, 7166-7171. Ritte, G., Heydenreich, M., Mahlow, S., Haebel, S., Kötting, O., and Steup, M. (2006). Phosphorylation of C6- and C3-positions of glucosyl residues in starch is catalysed by distinct dikinases. FEBS Letters 580, 4872-4876. Robinson, M.D., and Oshlack, A. (2010). A scaling normalization method for differential expression analysis of RNA-seq data. Genome Biology 11 :R25-. Robinson, M.D., McCarthy, D.J., and Smyth, G.K. (2010). EdgeR: a bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26, 139-140. Roldan, I., Wattebled, F., Lucas, M.M., Delvalle, D., Planchot, V., Jimenez, S., Perez, R., Ball, S., D'Hulst, C., and Merida, A. (2007). The phenotype of soluble starch synthase IV defective mutants of Arabidopsis thaliana suggests a novel function of elongation enzymes in the control of starch granule formation. Plant Journal 49, 492 - 504. Ronaghi, M., Uhlen, M., and Nyren, P. (1998). A sequencing method based on real-time pyrophosphate. Science 281, 363-365. Roper, H. (2002). Renewable raw materials in Europe - Industrial utilisation of starch and sugar. Starch- Stärke 54, 89-99. Rosenquist, M., Alsterfjord, M., Larsson, C., and Sommarin, M. (2001). Data mining the Arabidopsis genome reveals fifteen 14-3-3 genes. Expression is demonstrated for two out of five novel genes. Plant Physiology 127, 142-149. Rubio, V., Shen, Y.P., Saijo, Y., Liu, Y.L., Gusmaroli, G., Dinesh-Kumar, S.P., and Deng, X.W. (2005). An alternative tandem affinity purification strategy applied to Arabidopsis protein complex isolation. Plant Journal 41, 767-778. Rydberg, U., Andersson, L., Andersson, R., Aman, P., and Larsson, H. (2001). Comparison of starch branching enzyme I and II from potato. European Journal of Biochemistry 268, 6140-6145. Satoh, H., Shibahara, K., Tokunaga, T., Nishi, A., Tasaki, M., Hwang, S.K., Okita, T.W., Kaneko, N., Fujita, N., Yoshida, M., Hosaka, Y., Sato, A., Utsumi, Y., Ohdan, T., and Nakamura, Y. (2008). Mutation of the plastidial alpha-glucan phosphorylase gene in rice affects the synthesis and structure of starch in the endosperm. Plant Cell 20, 1833-1849. Scheibe, R. (2004). Malate valves to balance cellular energy supply. Physiologia Plantarum 120, 21-26.

7. References 171 ______

Schneider, A., Hausler, R.E., Kolukisaoglu, U., Kunze, R., van der Graaff, E., Schwacke, R., Catoni, E., Desimone, M., and Flügge, U.I. (2002). An Arabidopsis thaliana knock-out mutant of the chloroplast triose phosphate/phosphate translocator is severely compromised only when starch synthesis, but not starch mobilisation is abolished. Plant Journal 32, 685-699. Sehnke, P.C., Chung, H.J., Wu, K., and Ferl, R.J. (2001). Regulation of starch accumulation by granule-associated plant 14-3-3 proteins. Proceedings of the National Academy of Sciences of the United States of America 98, 765-770. Seo, B.S., Kim, S., Scott, M.P., Singletary, G.W., Wong, K.S., James, M.G., and Myers, A.M. (2002). Functional interactions between heterologously expressed starch-branching enzymes of maize and the glycogen Synthases of brewer's yeast. Plant Physiology 128, 1189-1199. Serino, G., and Deng, X.-W. (2007). Protein Coimmunoprecipitation in Arabidopsis . Cold Spring Harbour Protocols 2007, pdb.prot4683-4687. Serrato, A.J., Pérez-Ruiz, J.M., Spínola, M.C., and Cejudo, F.J. (2004). A novel NADPH thioredoxin reductase, localized in the chloroplast, which deficiency causes hypersensitivity to abiotic stress in Arabidopsis thaliana . Journal of Biological Chemistry 279, 43821-43827. Sheng, M., and Sala, C. (2001). PDZ domains and the organization of supramolecular complexes. Annual Review of Neuroscience 24, 1-29. Shure, M., Wessler, S., and Fedoroff, N. (1983). Molecular-identification and isolation of the waxy locus in maize. Cell 35, 225-233. Smith, A.M. (2008). Prospects for increasing starch and sucrose yields for bioethanol production. The Plant Journal 54, 546-558. Smith, A.M., and Stitt, M. (2007). Coordination of carbon supply and plant growth. Plant Cell and Environment 30, 1126-1149. Smith, S.M., Fulton, D.C., Chia, T., Thorneycroft, D., Chapple, A., Dunstan, H., Hylton, C., Zeeman, S.C., and Smith, A.M. (2004). Diurnal changes in the transcriptome encoding enzymes of starch metabolism provide evidence for both transcriptional and posttranscriptional regulation of starch metabolism in Arabidopsis leaves. Plant Physiology 136, 2687-2699. Sokolov, L.N., Dominguez-Solis, J.R., Allary, A.L., Buchanan, B.B., and Luan, S. (2006). A redox- regulated chloroplast protein phosphatase binds to starch diurnally and functions in its accumulation. Proceedings of the National Academy of Sciences of the United States of America 103, 9732-9737. Sonnewald, U., Basner, A., Greve, B., and Steup, M. (1995). A 2nd L-type isozyme of potato glucan phosphorylase - cloning, antisense inhibition and expression analysis. Plant Molecular Biology 27, 567-576. Sparla, F., Costa, A., Lo Schiavo, F., Pupillo, P., and Trost, P. (2006). Redox regulation of a novel plastid-targeted beta-amylase of Arabidopsis . Plant Physiology 141, 840-850. Stanley, D., Fitzgerald, A.M., Farnden, K.J.F., and MacRae, E.A. (2002). Characterisation of putative alpha-amylases from apple ( Malus domestica ) and Arabidopsis thaliana . Biologia 57, 137-148. Steichen, J.M., Petty, R.V., and Sharkey, T.D. (2008). Domain characterization of a 4-alpha- glucanotransferase essential for maltose metabolism in photosynthetic leaves. Journal of Biological Chemistry 283, 20797-20804. Stitt, M., and Quick, W.P. (1989). Photosynthetic carbon partitioning: its regulation and possibilities for manipulation. Physiologia Plantarum 77, 633-641. Streb, S., Delatte, T., Umhang, M., Eicke, S., Schorderet, M., Reinhardt, D., and Zeeman, S.C. (2008). Starch granule biosynthesis in Arabidopsis is abolished by removal of all debranching enzymes but restored by the subsequent removal of an endoamylase. Plant Cell 20, 3448-3466. Sullivan, M.A., Vilaplana, F., Cave, R.A., Stapleton, D., Gray-Weale, A.A., and Gilbert, R.G. (2010). Nature of α and β particles in glycogen using molecular size distributions. Biomacromolecules 11, 1094-1100.

7. References 172 ______

Sulpice, R., Pyl, E.T., Ishihara, H., Trenkamp, S., Steinfath, M., Witucka-Wall, H., Gibon, Y., Usadel, B., Poree, F., Piques, M.C., Von Korff, M., Steinhauser, M.C., Keurentjes, J.J.B., Guenther, M., Hoehne, M., Selbig, J., Fernie, A.R., Altmann, T., and Stitt, M. (2009). Starch as a major integrator in the regulation of plant growth. Proceedings of the National Academy of Sciences of the United States of America 106, 10348-10353. Supek, F., Škunca, N., Repar, J., Vlahovi ček, K., and Šmuc, T. (2010). Translational selection is ubiquitous in prokaryotes. PLoS Genetics 6, e1001004. Surget-Groba, Y., and Montoya-Burgos, J.I. (2010). Optimization of de novo transcriptome assembly from next-generation sequencing data. Genome Research 20, 1432-1440. Swarbreck, D., Wilks, C., Lamesch, P., Berardini, T.Z., Garcia-Hernandez, M., Foerster, H., Li, D., Meyer, T., Muller, R., Ploetz, L., Radenbaugh, A., Singh, S., Swing, V., Tissier, C., Zhang, P., and Huala, E. (2008). The Arabidopsis Information Resource (TAIR): gene structure and function annotation. Nucleic Acids Research 36, 1009-1014. Szydlowski, N., Ragel, P., Raynaud, S., Lucas, M.M., Roldan, I., Montero, M., Munoz, F.J., Ovecka, M., Bahaji, A., Planchot, V., Pozueta-Romero, J., D'Hulst, C., and Merida, A. (2009). Starch granule initiation in Arabidopsis requires the presence of either class IV or class III starch synthases. Plant Cell 21, 2443-2457. Takashima, Y., Senoura, T., Yoshizaki, T., Hamada, S., Ito, H., and Matsui, H. (2007). Differential chain-length specificities of two isoamylase-type starch-debranching enzymes from developing seeds of kidney bean. Bioscience, Biotechnology, and Biochemistry 71, 2308-2312. Takeda, Y., and Hizukuri, S. (1981). Studies on starch phosphate – reexamination of the action of sweet-potato beta-amylase on phosphorylated (1-4)-alpha-D-glucan. Carbohydrate Research 89, 174-178. Takeda, Y., Guan, H.P., and Preiss, J. (1993). Branching of amylose by the branching isoenzymes of maize endosperm. Carbohydrate Research 240, 253-263. Tanaka, N., Fujita, N., Nishi, A., Satoh, H., Hosaka, Y., Ugaki, M., Kawasaki, S., and Nakamura, Y. (2004). The structure of starch can be manipulated by changing the expression levels of starch branching enzyme llb in rice endosperm. Plant Biotechnology Journal 2, 507-516. Taniguchi, M., Taniguchi, Y., Kawasaki, M., Takeda, S., Kato, T., Sato, S., Tahata, S., Miyake, H., and Sugiyama, T. (2002). Identifying and characterizing plastidic 2-oxoglutarate/malate and dicarboxylate transporters in Arabidopsis thaliana . Plant and Cell Physiology 43, 706-717. Taylor, L.P., and Hepler, P.K. (1997). Pollen germination and tube growth. Annual Review of Plant Physiology and Plant Molecular Biology 48, 461-491. Tetlow, I.J., Morell, M.K., and Emes, M.J. (2004a). Recent developments in understanding the regulation of starch metabolism in higher plants. Journal of Experimental Botany 55, 2131-2145. Tetlow, I.J., Wait, R., Lu, Z.X., Akkasaeng, R., Bowsher, C.G., Esposito, S., Kosar-Hashemi, B., Morell, M.K., and Emes, M.J. (2004b). Protein phosphorylation in amyloplasts regulates starch branching enzyme activity and protein-protein interactions. Plant Cell 16, 694-708. Tetlow, I.J., Beisel, K.G., Cameron, S., Makhmoudova, A., Liu, F., Bresolin, N.S., Wait, R., Morell, M.K., and Emes, M.J. (2008). Analysis of protein complexes in wheat amyloplasts reveals functional interactions among starch biosynthetic enzymes. Plant Physiology 146, 1878-1891. Tiessen, A., Hendriks, J.H.M., Stitt, M., Branscheid, A., Gibon, Y., Farre, E.M., and Geigenberger, P. (2002). Starch synthesis in potato tubers is regulated by post-translational redox modification of ADP-glucose pyrophosphorylase: A novel regulatory mechanism linking starch synthesis to the sucrose supply. Plant Cell 14, 2191-2213. Tiessen, A., Prescha, K., Branscheid, A., Palacios, N., McKibbin, R., Halford, N.G., and Geigenberger, P. (2003). Evidence that SNF1-related kinase and hexokinase are involved in separate sugar-signalling pathways modulating post-translational redox activation of ADP- glucose pyrophosphorylase in potato tubers. Plant Journal 35, 490-500.

7. References 173 ______

Tomlinson, K., and Denyer, K. (2003). Starch synthesis in cereal grains. Advances in Botanical Research, Vol 40, pp. 1-61. Uchoa, V.T., de Paula, R.C., Krettli, L.G., Santana, A.E.G., and Krettli, A.U. (2010). Antimalarial Activity of Compounds and Mixed Fractions of Cecropia pachystachya . Drug Development Research 71, 82-91. Utsumi, Y., and Nakamura, Y. (2006). Structural and enzymatic characterization of the isoamylase1 homo-oligomer and the isoamylase1-isoamylase2 hetero-oligomer from rice endosperm. Planta 225, 75-87. Valerio, C., Costa, A., Marri, L., Issakidis-Bourguet, E., Pupillo, P., Trost, P., and Sparla, F. (2010). Thioredoxin-regulated beta-amylase (BAM1) triggers diurnal starch degradation in guard cells, and in mesophyll cells under osmotic stress. Journal of Experimental Botany . Vogel, S.S., Thaler, C., and Koushik, S.V. (2006). Fanciful FRET. Science Signalling – The Signalling Transduction Knowledge Environment 2006, re2. Wall, P.K., Leebens-Mack, J., Chanderbali, A.S., Barakat, A., Wolcott, E., Liang, H.Y., Landherr, L., Tomsho, L.P., Hu, Y., Carlson, J.E., Ma, H., Schuster, S.C., Soltis, D.E., Soltis, P.S., Altman, N., and dePamphilis, C.W. (2009). Comparison of next generation sequencing technologies for transcriptome characterization. BMC Genomics 10, 19. Wang, Q., Monroe, J., and Sjolund, D. (1995). Identification and characterization of a phloem-specific beta-amylase. Plant Physiology 109, 743-750. Wang, Z., Gerstein, M., and Snyder, M. (2009). RNA-Seq: a revolutionary tool for transcriptomics. Nature Reviews Genetics 10, 57-63. Wattebled, F., Buleon, A., Bouchet, B., Ral, J.P., Lienard, L., Delvalle, D., Binderup, K., Dauvillee, D., Ball, S., and D'Hulst, C. (2002). Granule-bound starch synthase I - A major enzyme involved in the biogenesis of B-crystallites in starch granules. European Journal of Biochemistry 269, 3810-3820. Wattebled, F., Dong, Y., Dumez, S., Delvalle, D., Planchot, R., Berbezy, P., Vyas, D., Colonna, P., Chatterjee, M., Ball, S., and D'Hulst, C. (2005). Mutants of Arabidopsis lacking a chloroplastic isoamylase accumulate phytoglycogen and an abnormal form of amylopectin. Plant Physiology 138, 184-195. Weber, A., Servaites, J.C., Geiger, D.R., Kofler, H., Hille, D., Groner, F., Hebbeker, U., and Flugge, U.I. (2000). Identification, purification, and molecular cloning of a putative plastidic glucose translocator. Plant Cell 12, 787-801. Weiner, H., Stitt, M., and Heldt, H.W. (1987). Subcellular compartmentation of pyrophosphate and alkaline pyrophosphatase in leaves. Biochimica et Biophysica Acta 893, 13-21. Weise, S.E., Weber, A.P.M., and Sharkey, T.D. (2004). Maltose is the major form of carbon exported from the chloroplast at night. Planta 218, 474-482. Weise, S.E., Kim, K.S., Stewart, R.P., and Sharkey, T.D. (2005). Beta-maltose is the metabolically active anomer of maltose during transitory starch degradation. Plant Physiology 137, 756-761. Wu, C.Y., Colleoni, C., Myers, A.M., and James, M.G. (2002). Enzymatic properties and regulation of ZPU1, the maize pullulanase-type starch debranching enzyme. Archives of Biochemistry and Biophysics 406, 21-32. Yoo, S.-H., and Jane, J.-l. (2002). Molecular weights and gyration radii of amylopectins determined by high-performance size-exclusion chromatography equipped with multi-angle laser-light scattering and refractive index detectors. Carbohydrate Polymers 49, 307-314. Young, A.L., Abaan, H.O., Zerbino, D., Mullikin, J.C., Birney, E., and Margulies, E.H. (2010). A new strategy for genome assembly using short sequence reads and reduced representation libraries. Genome Research 20, 249-256. Yu, T.S., Kofler, H., Hausler, R.E., Hille, D., Flugge, U.I., Zeeman, S.C., Smith, A.M., Kossmann, J., Lloyd, J., Ritte, G., Steup, M., Lue, W.L., Chen, J.C., and Weber, A. (2001). The Arabidopsis sex1 mutant is defective in the R1 protein, a general regulator of starch degradation in plants, and not in the chloroplast hexose transporter. Plant Cell 13, 1907-1918.

7. References 174 ______

Yu, T.S., Zeeman, S.C., Thorneycroft, D., Fulton, D.C., Dunstan, H., Lue, W.L., Hegemann, B., Tung, S.Y., Umemoto, T., Chapple, A., Tsai, D.L., Wang, S.M., Smith, A.M., Chen, J., and Smith, S.M. (2005). Alpha-amylase is not required for breakdown of transitory starch in Arabidopsis leaves. Journal of Biological Chemistry 280, 9773-9779. Zeeman, S.C., and Ap Rees, T. (1999). Changes in carbohydrate metabolism and assimilate export in starch-excess mutants of Arabidopsis . Plant Cell and Environment 22, 1445-1453. Zeeman, S.C., Smith, S.M., and Smith, A.M. (2004a). The breakdown of starch in leaves. New Phytologist 163, 247-261. Zeeman, S.C., Smith, S.M., and Smith, A.M. (2006). The diurnal metabolism of leaf starch. Biochemical Journal 401 , 13-28. Zeeman, S.C., Kossmann, J., and Smith, A.M. (2010). Starch: its metabolism, evolution, and biotechnological modification in plants. Annual Review of Plant Biology 61 , 209-234. Zeeman, S.C., Northrop, F., Smith, A.M., and Rees, T. (1998a). A starch-accumulating mutant of Arabidopsis thaliana deficient in a chloroplastic starch-hydrolysing enzyme. Plant Journal 15, 357 - 365. Zeeman, S.C., Northrop, F., Smith, A.M., and ap Rees, T. (1998b). A starch-accumulating mutant of Arabidopsis thaliana deficient in a chloroplastic starch-hydrolysing enzyme. Plant Journal 15, 357-365. Zeeman, S.C., Tiessen, A., Pilling, E., Kato, K.L., Donald, A.M., and Smith, A.M. (2002). Starch synthesis in Arabidopsis . Granule synthesis, composition, and structure. Plant Physiology 129, 516-529. Zeeman, S.C., Umemoto, T., Lue, W.L., Au-Yeung, P., Martin, C., Smith, A.M., and Chen, J. (1998c). A mutant of Arabidopsis lacking a chloroplastic isoamylase accumulates both starch and phytoglycogen. Plant Cell 10, 1699-1711. Zeeman, S.C., Thorneycroft, D., Schupp, N., Chapple, A., Weck, M., Dunstan, H., Haldimann, P., Bechtold, N., Smith, A.M., and Smith, S.M. (2004b). Plastidial alpha-glucan phosphorylase is not required for starch degradation in Arabidopsis leaves but has a role in the tolerance of abiotic stress. Plant Physiology 135, 849-858. Zerbino, D.R., and Birney, E. (2008). Velvet: Algorithms for de novo short read assembly using de Bruijn graphs. Genome Research 18, 821-829. Zhang, X., Szydlowski, N., Delvalle, D., D'Hulst, C., James, M., and Myers, A. (2008). Overlapping functions of the starch synthases SSII and SSIII in amylopectin biosynthesis in Arabidopsis . BMC Plant Biology 8, 96. Zhang, Z., Schwartz, S., Wagner, L., and Miller, W. (2000). A greedy algorithm for aligning DNA sequences. Journal of Computational Biology 7, 203-214. Zhang, Z.Y., and Dixon, J.E. (1993). Actie-site labelling of the Yersinia protein-tyrosine-phosphatase – the determinations of the pKa of the acitve site cysteine and the function of the conserved histidine-402 Biochemistry 32, 9340-9345. Zhulidov, P.A., Bogdanova, E.A., Shcheglov, A.S., Vagner, L.L., Khaspekov, G.L., Kozhemyako, V.B., Matz, M.V., Meleshkevitch, E., Moroz, L.L., Lukyanov, S.A., and Shagin, D.A. (2004). Simple cDNA normalization using kamchatka crab duplex-specific nuclease. Nucleic Acids Research 32 .

8. Appendix 175 ______

8. Appendix

APPENDIX I

AtS1b-YC AtS1b-YC BAM1-YC LSF1-YC LSF1-YN BAM1-YN AtS1b-NC AtS1b-NC YFP Chlorophyll Overlay

Bimolecular Fluorescence Complementation Controls The N- and C-terminal parts of YFP were fused in frame to the c-terminal ends of LSF1 and BAM1 proteins as well as to the small subunit of RuBisCo (AtS1b). Constructs under the control of a 2x 35S promoter were transiently expressed in N. benthamiana epidermal cells by Agrobacterium mediated co- infiltration. Tests for self-assembly of the YFP protein were negative (Courtesy of Alex Graf).

8. Appendix 176 ______

APPENDIX II

Comparison of Non-Normalized Read Counts between Biological Replicas The reads for a given transcript from either two ‘Leaf’ replicates (‘L1 vs L2’ and ‘L2 vs L3’) or two ‘Müllerian Body’ replicates (‘MB1 vs MB2’) were log2 transformed and plotted against each other. The general correlation between biological replicas was good. The distribution of points follows a line with a slope similar to the blue line which indicates equal expression in both conditions. Unequal total number of reads reported for the different biological samples results in a bias which shifts the distribution of points in parallel to the blue line.

8. Appendix 177 ______

APPENDIX III

Suppl. Table: GO-Terms Overrepresented Among Transcripts Overexpressed in Leaves

GO Term Name # in test group FDR GO:0016168 chlorophyll binding 24 1.40E-09 GO:0009570 chloroplast stroma 117 1.40E-09 GO:0018298 protein-chromophore linkage 18 1.40E-09 GO:0009055 electron carrier activity 68 4.10E-09 GO:0009941 chloroplast envelope 56 7.00E-09 GO:0010287 plastoglobule 24 3.10E-08 GO:0009773 photosynthetic electron transport in photosystem I 13 3.70E-08 GO:0030076 light-harvesting complex 13 3.70E-08 GO:0048046 apoplast 60 1.10E-07 GO:0009543 chloroplast thylakoid lumen 23 2.40E -07 GO:0009654 oxygen evolving complex 11 6.80E-06 GO:0008266 poly(U) RNA binding 9 1.50E-05 GO:0008187 poly-pyrimidine tract binding 9 1.50E-05 GO:0019253 reductive pentose-phosphate cycle 12 3.70E-05 GO:0042742 defense response to bacterium 31 4.90E-05 GO:0015995 chlorophyll biosynthetic process 17 6.40E-05 GO:0009658 chloroplast organization 17 1.30E-04 GO:0046246 terpene biosynthetic process 7 2.80E-04 GO:0009772 photosynthetic electron transport in photosystem II 7 2.80E-04 GO:0020037 heme binding 28 5.40E-04 GO:0045038 protein import into chloroplast thylakoid membrane 6 1.00E-03 GO:0005960 glycine cleavage complex 6 1.00E-03 GO:0010598 NAD(P)H dehydrogenase complex (plastoquinone) 6 1.00E-03 GO:0009538 photosystem I reaction center 6 1.00E-03 PSII associated light-harvesting complex II catabolic GO:0010304 6 process 1.00E-03 GO:0018130 heterocycle biosynthetic process 28 1.20E-03 GO:0048038 quinone binding 8 1.60E -03 GO:0009409 response to cold 38 3.20E-03 GO:0009854 oxidative photosynthetic carbon pathway 5 4.90E-03 GO:0009416 response to light stimulus 42 6.10E-03 GO:0016851 magnesium chelatase activity 6 6.90E-03 GO:0030095 chloroplast photosystem II 6 6.90E -03 GO:0004176 ATP-dependent peptidase activity 8 9.90E-03 GO:0009508 plastid chromosome 8 9.90E-03 GO:0045454 cell redox homeostasis 21 1.40E-02 GO:0006536 glutamate metabolic process 6 1.60E-02 GO:0015114 phosphate transmembrane transporter activity 6 1.60E-02 GO:0003973 (S) -2-hydroxy -acid oxidase activity 4 2.20E -02 GO:0009768 photosynthesis, light harvesting in photosystem I 4 2.20E-02 GO:0016120 carotene biosynthetic process 4 2.20E-02 GO:0004375 glycine dehydrogenase (decarboxylating) activity 4 2.20E-02 GO:0042132 fructose 1,6-bisphosphate 1-phosphatase activity 4 2.20E-02 GO:0030091 protein repair 5 2.40E-02 GO:0010206 photosystem II repair 5 2.40E-02 GO:0006546 glycine catabolic process 5 2.40E-02 GO:0010007 magnesium chelatase complex 5 2.40E-02 GO:0009813 flavonoid biosynthetic process 9 2.60E-02 GO:0016667 activity, acting on sulfur group of donors 16 2.70E-02

8. Appendix 178 ______

GO:0016209 antioxidant activity 16 2.70E-02 GO:0008237 metallopeptidase activity 17 3.20E-02 GO:0009414 response to water deprivat ion 17 3.90E -02 oxidoreductase activity, acting on paired donors, with incorporation or reduction of molecular oxygen, NADH or GO:0016709 7 NADPH as one donor, and incorporation of one atom of oxygen 4.30E-02 GO:0031408 oxylipin biosynthetic process 5 4.90E-02 GO:0042548 regulation of photosynthesis, light reaction 5 4.90E -02 Fisher‘s Exact Test implemented in the GOSSIP -package in BLAST2Go was used to test for over - representation of GO descriptions in the set of transcripts over-expressed in leaves as compared to the GO descriptions of the total set of transcripts. A false discovery rate (FDR) of 5% was used as cut-off. A description of the GO group which is overrepresented and the FDR are indicated. In addition, the numbers of transcripts present in the GO groups of transcripts over-expressed in leaves are indicated. Some transcripts may be shared between GO terms.

8. Appendix 179 ______

APPENDIX IV

Suppl. Table: GO-Terms Overrepresented Among Transcripts Over-Expressed in Müllerian Bodies

GO Term Name # in test group FDR GO:0004553 activity, hydrolyzing O-glycosyl compounds 41 5.20E-07 GO:0030312 external encapsulating structure 59 5.20E-07 GO:0005618 cell wall 58 6.10E-07 GO:0016798 hydrolase activity, acting on glycosyl bonds 41 1.00E-06 GO:0003824 catalytic activity 380 1.00E-06 GO:0005975 carbohydrate metabolic process 84 5.10E -05 GO:0016740 transferase activity 140 5.40E-05 GO:0006629 lipid metabolic process 55 6.00E-05 GO:0008202 steroid metabolic process 13 6.00E-05 GO:0009505 plant-type cell wall 30 1.60E-04 GO:0016762 xyloglucan:xyloglucosyl transferase activity 9 2.80E-04 GO:0016757 transferase activity, transferring glycosyl groups 41 2.90E-04 GO:0006633 fatty acid biosynthetic process 24 2.90E-04 GO:0030599 pectinesterase activity 11 3.90E-04 GO:0008610 lipid biosynthetic process 40 5.70E-04 GO:0006631 fatty acid metabolic process 26 6.50E-04 GO:0042545 cell wall modification 10 6.90E -04 GO:0006694 steroid biosynthetic process 11 7.00E-04 oxidoreductase activity, acting on paired donors, with GO:0016705 17 7.60E-04 incorporation or reduction of molecular oxygen GO:0016788 hydrolase activity, acting on ester bonds 46 8.80E-04 GO:0031225 anchored to membrane 16 8.80E-04 GO:0009620 response to fungus 12 1.50E-03 GO:0003872 6-phosphofructokinase activity 6 2.00E-03 GO:0005945 6-phosphofructokinase complex 6 2.00E-03 GO:0071555 cell wall organization 16 2.10E-03 GO:0004737 pyruvate decarboxylase activity 5 2.60E-03 GO:0016773 phosphotransferase activity, alcohol group as acceptor 50 3.00E-03 GO:0071554 cell wall organization or biogenesis 19 3.00E-03 GO:0005576 extracellular region 43 3.00E-03 GO:0031988 membrane-bounded vesicle 83 3.00E-03 GO:0016023 cytoplasmic membrane-bounded vesicle 83 3.00E-03 GO:0031410 cytoplasmic vesicle 83 3.00E-03 GO:0031982 vesicle 83 3.00E-03 GO:0016126 sterol biosynthetic process 8 3.40E-03 GO:0016125 sterol metabolic process 8 3.40E-03 GO:0004091 carboxylesterase activity 16 3.90E-03 GO:0070482 response to oxygen levels 6 4.20E-03 GO:0001666 response to hypoxia 6 4.20E-03 GO:0008443 phosphofructokinase activity 6 4.20E -03 GO:0009566 fertilization 6 4.20E-03 GO:0016758 transferase activity, transferring hexosyl groups 30 4.50E-03 GO:0030976 thiamin pyrophosphate binding 5 7.00E-03 GO:0009877 nodulation 5 7.00E-03 GO:0044403 symbiosis, encompassing mutualism through parasitism 5 7.00E-03 GO:0009969 xyloglucan biosynthetic process 4 1.10E -02 GO:0000254 C-4 methylsterol oxidase activity 4 1.10E-02 GO:0046202 cyanide biosynthetic process 4 1.10E-02 GO:0010025 wax biosynthetic process 6 1.20E-02

8. Appendix 180 ______

GO:0010166 wax metabolic process 6 1.20E-02 GO:0050662 coenzyme binding 35 1.20E-02 GO:0009751 response to salicylic acid stimulus 7 1.30E -02 GO:0050832 defense response to fungus 9 1.40E-02 GO:0044042 glucan metabolic process 20 1.40E-02 GO:0016830 carbon-carbon activity 18 1.40E-02 GO:0016831 carboxy-lyase activity 14 1.40E-02 GO:0006073 cellular glucan metabolic process 19 1.40E-02 GO:0032787 monocarboxylic acid metabolic process 32 1.60E -02 GO:0004312 fatty-acid synthase activity 6 2.10E-02 GO:0006066 alcohol metabolic process 36 2.50E-02 GO:0009698 phenylpropanoid metabolic process 11 2.70E-02 GO:0000038 very-long-chain fatty acid metabolic process 4 3.20E-02 GO:0019299 rhamnose metabolic process 4 3.20E-02 GO:0048235 pollen sperm cell differentiation 4 3.20E-02 GO:0005991 trehalose metabolic process 4 3.20E-02 GO:0019499 cyanide metabolic process 4 3.20E-02 GO:0004672 protein kinase activity 37 3.30E-02 GO:0005976 polysaccharide metabolic process 25 3.60E-02 GO:0016301 kinase activity 53 3.60E-02 GO:0044419 interspecies interaction between organisms 5 3.90E -02 GO:0044262 cellular carbohydrate metabolic process 48 4.10E-02 GO:0048037 binding 44 4.10E-02 Fisher‘s Exact Test implemented in the GOSSIP-package in BLAST2Go was used to test for over- representation of GO descriptions in the set of transcripts over-expressed in Müllerian bodies as compared to the GO descriptions of the total set of transcripts. A false discovery rate (FDR) of 5% was used as cut-off. A description of the GO group which is overrepresented and the FDR are indicated. In addition, the numbers of transcripts present in the GO groups of transcripts over-expressed in MBs are indicated. Some transcripts may be shared between GO terms.

9. Curriculum vitae 181 ______

9. Curriculum vitae

MARTIN UMHANG

Home Address: Greyerzstrasse 38, 3013 Bern, Switzerland Tel.: +41 79 733 22 40 Email: [email protected] University Address: Swiss Federal Institute of Technology Department of Biology, Institute of Plant Sciences Universitätsstr. 2 8092 Zürich, Switzerland Tel.: +41 44 632 38 39 Marital Status: Single Date of Birth: 20.07.1980 Place of Birth: Thun, Switzerland

EDUCATION

Feb 2007 - present PhD in Plant Biochemistry , ETH Zürich. Introduction and validation of a Tandem-Affinity- Purification method for protein complexes. Establishment of gel permeation chromatography techniques to screen for protein complexes. Responsible for FPLC protein purification unit. Supervisor: Prof. Dr S. Zeeman Nov 2004 - Dec 2005 MSc in Biology, University of Bern and ETH Zürich. Heterologous expression, purification and characterization of enzymes involved in starch degradation Supervisors: Prof. Dr S. Zeeman and Prof. Dr D. Rentsch Aug 2001 - Nov 2004 Studies in Biology , University of Bern Aug 1997 - July 2001 Gymnasium Thun - Schadau, Typus DERu (modern languages: English, French, Russian)

WORK EXPERIENCE

Jan 2006 - Jan 2007 Research assistant at the ETH Zürich, in the groups of Plant Physiology and Plant Biochemistry (Prof. Dr N. Amrhein and Prof. Dr S. Zeeman)

TEACHING and SUPERVISING EXPERIENCE

Lectures in ‘Primary Plant Metabolism’ for 5 th semester students at the ETH Zürich

9. Curriculum vitae 182 ______

Supervision of two semester students (3 months) and partial supervision of a master student (6 months) Supervision of a trainee technician (3 rd year)

PUBLICATIONS and PRESENTATIONS

Umhang, M. , Mueller, A., Koetting, O., Stettler, M., Chen, J., and Zeeman, S.C. (2010) Complex Formation between Enzymes Involved in Starch Degradation. ASPB Meeting, Montréal, Canada, July 31 st - August 4 th American Society of Plant Biology (oral presentation and poster ) Streb, S., Delatte, T., Umhang, M. , Eicke, S., Schorderet, M., Reinhardt, D., and Zeeman, S.C. (2008) Starch Granule Biosynthesis in Arabidopsis Is Abolished by Removal of All Debranching Enzymes but Restored by the Subsequent Removal of an Endoamylase. The Plant Cell, 20, 3448–3466 Zeeman, S.C., Delatte, T., Messerli, G., Umhang, M ., Stettler, M., Mettler, T., Streb, S., Reinhold, H., and Kötting, O. (2007) Starch breakdown: recent discoveries suggest distinct pathways and novel mechanisms. Functional Plant Biology 34, 465–473 Delatte, T., Umhang, M. , Trevisan, M., Eicke, S., Thorneycroft, T., Smith, S.M. and Zeeman, S.C. (2006) Evidence for distinct mechanisms of starch granule breakdown in plants. Journal of Biological Chemistry , 281, 12050-12059

FUNDING and FELLOWSHIPS

Jan 2007 - Jan 2008 Successful application for funding from the ‘Roche Research Foundation’ Feb 2008 - present Applied for and received the ‘Heinz Imhof-Fellowship’ from Syngenta

LANGUAGE SKILLS

German: mother tongue English: proficient, written and spoken French: fluent, written and spoken Russian: basic reading, writing and understanding