Advanced bioanalysis of light-controlled S. cerevisiae production strains - Dissertation - zur Erlangung des Grades eines Doktor der Naturwissenschaften (Dr. rer. nat.) des Fachbereichs Chemie der Philipps-Universität Marburg Vorgelegt von Filipp Bezold aus Wetzlar Marburg, 2023 Die vorliegende Arbeit wurde unter der Betreuung von Herrn Prof. Dr. Lars-Oliver Essen von September 2017 bis Februar 2023 am Fachbereich Chemie der Philipps-Universität Marburg angefertigt. Vom Fachbereich Chemie der Philipps-Universität Marburg (Hochschulkennziffer 1180) als Dissertation angenommen am: Erstgutachter: Prof. Dr. Lars-Oliver Essen Zweitgutachter: Priv. Doz. Dr. Christof Taxis Tag der Disputation: Eidesstattliche Erklärung Ich versichere, dass ich meine Dissertation mit dem Titel: ŤAdvanced bioanalysis of light-controlled S. cerevisiae production strainsŤ selbstständig, ohne unerlaubte Hilfe angefertigt und mich dabei keiner anderen als der von mir ausdrücklich bezeichneten Quellen und Hilfen bedient habe. Diese Dissertation wurde in der jetzigen oder einer ähnlichen Form noch bei keiner anderen Hochschule eingereicht und hat noch keinen sonstigen Prüfungszwecken gedient. Ort, Datum Filipp Bezold Publications Maestre-Reyna M, Wang P-H, Nango E, Hosokawa Y, Saft M, Furrer A, Yang C-H, Gusti Ngurah Putu E P, Wu W-J, Emmerich H-J, Franz-Badur S, Engilberge S, Caramello N, Wranik M, Glover H L, Weinert T, Wu H-Y, Lee C-C, Huang W-C, Huang K-F, Chang Y-K, Liao J-H, Weng J-H, Gad W, Chang C-W, Pang A H, Gashi D, Beale E, Ozerov D, Nass K, Knopp G, Johnson P J M, Cirelli C, Milne C, Bacellar C, Sugahara M, Owada S, Joti Y, Yamashita A, Tanaka R, Tanaka T, Luo F, Tono K, Müller P, Bezold F, Fuchs V, Gnau P, Kiontke S, Korf L, Reithofer V, Rosner C J, Werel L, Spadaccini R, Royant A, Yamamoto J, Iwata S, Standfuss J, Bessho Y, Essen L-O and Tsai M-D, ŤWatching the entire DNA repair process by a photolyase at atomic resolution in real time Ť, manuscript in preparation Bezold F, Scheffer J, Wendering P, Razaghi-Moghadam Z, Trauth J, Pook B, NuSShär H, Hasenjäger S, Nikoloski Z, Essen L-O and Taxis C ŤOptogenetic control of Cdc48 for dynamic metabolic engineering in yeastŤ, manuscript submitted Essen L-O, Taxis C, Bezold F and Scheffer J (2020) ŤVerfahren zur Herstellung eines Planzeninhaltsstoffes durch optogenetische Kontrolle von Zellzyklusregulatoren Ť, patent: EP4006164A1 Gramazio S, Trauth J, Bezold F, Essen L-O, Taxis C, Spadaccini R (2022) ŤLight- induced fermenter production of derivatives of the sweet protein monellin is maximized in prestationary Saccharomyces cerevisiae culturesŤ, Biotechnology Journal Hepp S, Trauth J, Hasenjäger S, Bezold F, Essen L-O and Taxis C (2020) ŤAn Opto- genetic Tool for Induced Protein Stabilization Based on the Phaeodactylum tricornutum Aureochrome 1a Light-Oxygen-Voltage Domain Ť, Journal of Molecular Biology Linne U, Bezold F and Bamberger J (2019) ŤCoupling Methods in Mass SpectrometryŤ, GIT Laboratory Journal Wenn Sie die Art und Weise ändern, wie Sie die Dinge betrachten, ändern sich die Dinge, die Sie betrachten. Ű Max Planck Abstract Vast progress in the Ąeld of biotechnology enabled metabolic engineering to become a powerful tool for the production of numerous biogenic substances in the last decade. Alongside methods for editing biological pathways, bioanalytical techniques play a cen- tral role in metabolic engineering approaches, as they allow for monitoring of pathway modiĄcations, identiĄcation of bottlenecks and quantiĄcation of target substances. The collaborative project Ťmetabolic engineering with light controlled modulesŤ (MELICOMO) aimed to harness optogenetic tools for the production of valuable sec- ondary metabolites in Saccharomyces cerevisiae and to implement a light-inducible cell cycle arrest as the desired cellular production state. Within the present work a broad spectrum of bioanalytical methods was applied for the characterization of light-controlled S. cerevisiae cell cycle mutants (CCMs) and for the quantiĄcation of target molecules in the context of metabolic engineering. The Ąrst part of this thesis summarizes results of biomass composition for cell cycle arrested Clb2ΔDB-psd3 cells and the corresponding wild type (WT) strain. Here, growth- restricted Clb2ΔDB-psd3 cells were found to contain more protein and RNA and less carbohydrates in comparison to WT cells. Moreover, results of a proteomics experiment performed for the light-controlled CCM strains Cdc48-psd3, Clb2ΔDB-psd3 and bPAC were analyzed. Thereby, relative protein abundances comparing the restrictive growth conditions to the permissive growth con- ditions and to the WT strain were examined. Obtained results were processed using heatmaps, volcano plots, Venn diagrams and Gene Ontology enrichment analyses. Here, for each of the growth-restricted Cdc48-psd3 and bPAC strains an acetyl-CoA synthetase isoform was found to be high abundant in comparison to the permissive growth condition and to the WT strain. This Ąnding might explain increased β-carotene yields observed for both growth-restricted strains and indicates beneĄcial production conditions for other isoprenoid-derived products generated from precursors provided by the mevalonate path- way. Furthermore, the phosphatase Pho8 was found to be present at low levels in growth-restricted Cdc48-psd3 cells compared with the WT strain, which may explain the observed increased cordycepin production of the growth-restricted CCM strain, since Pho8 is responsible for the degradation of the cordycepin precursor 3ŠAMP. In addi- tion, enzymes of the respiratory chain and the tricarboxylic acid cycle were found to be upregulated for growth-restricted bPAC cells, which could imply an increased energy availability for this condition. Finally a selection of LC-MS and HPLC analyses is shown, which helped to establish proof of principle 3ŠAMP, cordycepin and GA4 production strains. This thesis provides a large toolbox of state of the art bioanalytical methods, which helped to improve and rationalize the metabolic engineering approaches established within the MELICOMO project. I II Zusammenfassung Dank enormer Fortschritte im Bereich der Biotechnoligie hat sich metabolic engineering im letzten Jahrzehnt zu einer leistungsfähigen Methode entwickelt um die Herstellung biogener Substanzen zu ermöglichen. Neben molekularbiologischen Techniken spielen beim metabolic engineering bioanalytische Methoden eine zentrale Rolle. Ziel des Projektes Ťmetabolic engineering with light controlled modulesŤ (MELICOMO) war es, optogenetische Werkzeuge für die Produktion wertvoller Sekundärmetabolite in Saccharomyces cerevisiae zu nutzen und einen lichtinduzierbaren Zellzykluss-Arrest als zellulären Produktionszustand zu implementieren. Dabei wurde im Rahmen der vor- liegenden Arbeit ein breites Spektrum bioanalytischer Methoden für die Charakterisierung der Produktionsstämme und die QuantiĄzierung von Zielmolekülen etabliert. Im ersten Teil der vorliegenden Arbeit werden die Ergebnisse einer Biomasse-Analyse von zellzyklusarretierten Clb2ΔDB-psd3-Zellen und dem entsprechenden Wildtyp-Stamm zusammengefasst. Hier wurde festgestellt, dass zellzyklusarretierte Clb2ΔDB-psd3-Zellen im Vergleich zum Wildtyp-Stamm mehr Protein und RNA und weniger Kohlenhydrate enthalten. Darüber hinaus wurden Proteomics-Experimente zu den lichtgesteuerten Zellzyklus- Mutanten Cdc48-psd3, Clb2ΔDB-psd3 und bPAC durchgeführt und die Ergebnisse anhand von heatmaps, vulcano plots, Venn-Diagrammen und Gene Ontology Anreicherungs- Analysen diskutiert. Dabei wurde für den zellzyklusarretierten Cdc48-psd3 und bPAC- Stamm im Vergleich zum permissiven Zustand und zum Wildtyp-Stamm jeweils eine hochabundante Acetyl-CoA-Synthetase-Isoform gefunden. Dies könnte die erhöhten Ausbeute an β-Carotin erklären, welche bei den beiden zellzyklusarretierten Stämmen festgestellt wurde und deutet auf generell günstige Produktionsbedingungen für Iso- prenoide hin, die aus Produkten des Mevalonatwegs erzeugt werden. Desweiteren wurde festgestellt, dass die Phosphatase Pho8 in zellzyklusarretierten Cdc48-psd3-Zellen im Vergleich zum Wildtyp-Stamm niedrig abundant ist, was die beobachtete erhöhte Cordycepin-Produktion des zellzyklusarretierten Cdc48-psd3-Stamms erklären könnte, da Pho8 für den Abbau des Cordycepin-Vorläufers 3ŠAMP verantwortlich ist. AuSSer- dem lagen beim zellzyklusarretierten bPAC-Stamm viele Enzyme der Atmungskette und des Citratzyklus hochabundant vor, was auf eine erhöhte Energieverfügbarkeit in diesem Zustand hindeuten könnte. Im letzten Teil der Arbeit wird schlieSSlich eine Auswahl von LC-MS- und HPLC-Analysen gezeigt, mit deren Hilfe die Etablierung von 3ŠAMP, Cordycepin und GA4 Produktions- stämmen nachgewiesen werden konnte. Damit stellt diese Arbeit eine umfangreiche Toolbox modernster bioanalytischer Meth- oden bereit, die dazu beigetragen haben, die im Rahmen des MELICOMO-Projekts etablierten metabolic engineering Ansätze zu verbessern und zu rationalisieren. III IV Abbreviations Abbreviation Term AC adenylate cyclase ACN acetonitrile BLUF blue light using Ćavin BSA bovine serum albumin CCM cell cycle mutant DAD diode array detector DAP differentially abundant protein DB destruction box DBAA dibutylammonium acetate DNA deoxyribonucleic acid EC end-capped EIC extracted ion chromatogram ER endoplasmic reticulum ERAD ER-associated protein degradation FA formic acid FBA Ćux ballance analysis FDR false discovery rate FMN Ćavinmononucleotid FPP farnesyl pyrophosphate FT-ICR fourier transform ion cyclotron resonance GA4 gibberellin A4 GB gigabyte GGPP geranylgerarnyl pyrophosphate GO Gene Ontology ddH2O double-distilled water HAc acetic acid HPLC high performance liquid chromatography IAA iodoacetamide ID identiĄcation IEC Ion exchange column V Abbreviation Term LC liquid chromatography LCM light controlled mutant LFM low Ćuorescence media LFQ label free quantiĄcation LOD limit of detection LOQ limit of quantiĄcation LOV ight oxygen voltage MELICOMO metabolic engineering with light controlled modules MPI Max Planck Institute MRM multiple reaction monitoring MS mass spectrometer MWD multiple wavelength detector NAC N-acetylcysteine NC not calculated ND not detected ONC overnight culture O/N overnight PAS Per-Arnt-Sim PKA protein kinase A psd photosensitive degron RNA ribonucleic acid RP reversed phase RT room temperature / retention time SC synthetic complete SGD Saccharomyces genome database TCA tricarboxylic acid cycle TCEP tris(2-carboxyethyl)phosphine TC Ćask tissue culture Ćask TFA triĆuoroacetic acid (TFA) TIC total ion current UV ultra-violet VI Contents 1 Introduction 1 1.1 The Eukaryotic Model Organism S. cerevisiae . . . . . . . . . . . . . . . 1 1.1.1 Isoprenoid Production in Yeast . . . . . . . . . . . . . . . . . . 3 1.2 Optogenetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 1.3 Cell Cycle of S. cerevisiae . . . . . . . . . . . . . . . . . . . . . . . . . 8 1.4 Scope of the Project . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2 Results 11 2.1 Biomass Analysis of Light-Controlled S. cerevisiae . . . . . . . . . . . . 11 2.1.1 Cell Cultivation . . . . . . . . . . . . . . . . . . . . . . . . . . 11 2.1.2 Protein QuantiĄcation . . . . . . . . . . . . . . . . . . . . . . 12 2.1.3 Amino-Acid Composition . . . . . . . . . . . . . . . . . . . . . 13 2.1.4 Carbohydrate QuantiĄcation . . . . . . . . . . . . . . . . . . . . 14 2.1.5 RNA and DNA QuantiĄcation . . . . . . . . . . . . . . . . . . . 14 2.1.6 Lipid QuantiĄcation . . . . . . . . . . . . . . . . . . . . . . . . 16 2.1.7 Summary of Biomass Analysis . . . . . . . . . . . . . . . . . . . 18 2.2 Proteomics of Light-Controlled S. cerevisiae . . . . . . . . . . . . . . . 20 2.2.1 Heatmaps of Relative Protein Abundance . . . . . . . . . . . . 21 2.2.2 Volcano Plots and Differentially Abundant Proteins . . . . . . . 26 2.2.3 Gene Ontology (GO) Enrichment Analysis . . . . . . . . . . . . 41 2.2.4 Proteomics of Light-Controlled S. cerevisiae Ű Summary . . . . . 46 2.3 Bioanalysis for Production Strain Establishment . . . . . . . . . . . . . 48 2.3.1 β-Carotene Production . . . . . . . . . . . . . . . . . . . . . . . 48 2.3.2 3ŠAMP and Cordycepin Production . . . . . . . . . . . . . . . . 49 2.3.3 Gibberellin A4 Production . . . . . . . . . . . . . . . . . . . . . 57 3 Discussion 61 3.1 Biomass Analysis of Light-Controlled S. cerevisiae . . . . . . . . . . . . 61 3.2 Proteomics of Light-Controlled S. cerevisiae . . . . . . . . . . . . . . . 63 3.3 Bioanalysis for Production Strain Establishment . . . . . . . . . . . . . 73 3.4 Final Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 VII 4 Methods and Materials 79 4.1 Biomass Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 4.1.1 Cell Cultivation for Biomass Analysis . . . . . . . . . . . . . . . 79 4.1.2 Protein QuantiĄcation (Biuret assay) . . . . . . . . . . . . . . . 79 4.1.3 Protein QuantiĄcation (Bradford assay) . . . . . . . . . . . . . . 80 4.1.4 Determination of Amino-acid Composition . . . . . . . . . . . . 80 4.1.5 Carbohydrate QuantiĄcation . . . . . . . . . . . . . . . . . . . . 81 4.1.6 DNA and RNA QuantiĄcation . . . . . . . . . . . . . . . . . . . 82 4.1.7 Lipid QuantiĄcation . . . . . . . . . . . . . . . . . . . . . . . . 83 4.2 Proteomics of S. cerevisiae . . . . . . . . . . . . . . . . . . . . . . . . 84 4.2.1 Cell Production for Proteomics Experiments . . . . . . . . . . . 84 4.2.2 Sample Preparation for Proteomics Experiments . . . . . . . . . 85 4.2.3 timsTOF Data Acquisition . . . . . . . . . . . . . . . . . . . . . 86 4.2.4 FASTA Database Preparation . . . . . . . . . . . . . . . . . . . 86 4.2.5 PEAKS Studio Xpro Data Analysis WorkĆow . . . . . . . . . . 87 4.2.6 Selection of Proteins from Protein Abundance Exports . . . . . 88 4.2.7 Protein Heatmaps . . . . . . . . . . . . . . . . . . . . . . . . . 88 4.2.8 GO-term analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 88 4.3 Product QuantiĄcation . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 4.3.1 β-Carotene QuantiĄcation from Cell Extracts by HPLC . . . . . . 89 4.3.2 Cordycepin QuantiĄcation by LC-MS/MS . . . . . . . . . . . . . 89 4.3.3 QuantiĄcation of GA4 from Cell and Media Extracts . . . . . . . 90 4.3.4 QuantiĄcation of 3ŠAMP from Cell Extracts . . . . . . . . . . . 91 4.4 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92 4.4.1 Software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 4.4.2 Databases and Tools . . . . . . . . . . . . . . . . . . . . . . . . 96 5 Appendix 111 5.0.1 Biomass Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 111 5.0.2 Proteomics of S. cerevisiae . . . . . . . . . . . . . . . . . . . . 117 5.0.3 Product QuantiĄcation . . . . . . . . . . . . . . . . . . . . . . 128 5.0.4 Plasmids and Strains . . . . . . . . . . . . . . . . . . . . . . . . 135 5.0.5 Protein Sequences . . . . . . . . . . . . . . . . . . . . . . . . . 137 5.0.6 Data Deposition . . . . . . . . . . . . . . . . . . . . . . . . . . 142 VIII 1 Chapter 1 Introduction 1.1 S. cerevisiae, Eukaryotic Model Organism and Versatile Host for Metabolic Engineering In 1996, the budding yeast S. cerevisiae became the Ąrst eukaryotic organism to have its genome completely sequenced and has been studied extensively as a eukaryotic model organism because of its ease of cultivation and genetic manipulation. [36] S. cerevisiae can exist either haploid (mating type a and α) or diploid and two haploid cells of the opposite mating types readily mate to form diploids. Under nutrient-poor conditions, meiosis of diploid cells is induced, forming four haploid spores from diploid cells. In the haploid form of S. cerevisiae, gene mutations can be readily studied and a wide range of speciĄc/conditional knockout mutants for haploid and diploid strain backgrounds were made commercially available to facilitate high throughput screening. [34] The genome of S. cerevisiae harbors ≈ 6000 open reading frames (ORFs), of which ≈ 5000 have a known function and 18 % of the total ORFs were found to be essential for growth on rich glu- cose medium. [28] [35] Since many fundamental processes and biochemical pathways are conserved among yeast and higher eukaryotes, S. cerevisiae has been intensively studied to gain a better understanding of DNA repair, mRNA translation/degradation, and the cell cycle. [34] Moreover ≈ 40 % of the S. cerevisiae proteins have signiĄcant sequence similarity to at least one human protein (BLAST E-value < 10 -10), with hundreds of the human orthologs known to be involved in disease. [85] Consequently, S. cerevisiae is used as a model organism to understand diseases, a prominent example being research by Yoshinori Ōsumi on autophagy for which he was awarded the Nobel Prize in 2016. Cultivation of S. cerevisiae for the production of fermented foods, wine, beer and sake has been practiced for millennia by societies around the globe and a vast number of spe- cialized strains are used today by food industry and biofuel companies in order to convert carbon sources into products. [66] Economically, the most important industrial applica- tion of S. cerevisiae is the production of ethanol. Here S. cerevisiae is pre-dominatly responsible for the 100 billion liters of ethanol produced annually worldwide. The two 2 CHAPTER 1. INTRODUCTION major ethanol producing countries are the United States of America, where 40 % of the corn harvest is used for the production of ethanol, and Brazil, where sugar cane is used for ethanol production. Each industry relies on a different specialized yeast strain, as the substrates and fermentation processes are fundamentally different. [30] A targeted approach for the development of cell factories to produce desired compounds is metabolic engineering. This involves the rational genetic manipulation of cellular pathways to enable the production of speciĄc targets or to increase the titer of an estab- lished production. As an example metabolic engineering is used to expand the substrate spectrum of the host organism or to improve the organismŠs stress tolerance towards the production conditions, with the goal of efficiently converting low-cost feedstock into valuable products. [67] In this context, metabolic engineering aims to create sustainable alternatives to conventional fossil fuel-based production and to the costly extraction of natural products from biological matrices. In metabolic engineering establishment of production strains is driven by iterative rounds of design/build/test/learn cycles. [87] This concept serves to integrate the results of a rationally designed modiĄcation into the next round of the genetic engineering process. Each step of the optimization cycle beneĄts greatly from scientiĄc advances of the past decades. Rational pathway design takes ad- vantage from a wealth of accessible information about biological systems, modern gene editing allows short turnover times for genetic modiĄcations, advanced analytical in- struments have become readily available to laboratories and data analysis has beneĄted greatly from software and hardware improvements. In addition, methods of systems biology such as metabolic Ćux modeling are already helping to streamline metabolic pathway design, and in perspective, machine learning might fundamentally revolutionize metabolic engineering approaches. [5] [65] S. cerevisiae is used as a platform organism for metabolic engineering approaches due to the in-depth understanding of its metabolism and because it is easy to culture and genet- ically manipulate Ű reasons that made S. cerevisiae a model organism in the Ąrst place. In addition, this yeast is generally recognized as safe (GRAS) which facilitates industrial application. [67] As an eukaryote, S. cerevisiae harbors multiple cellular compartments which are often necessary for pathway establishment. Furthermore, it is capable of many post-translational modiĄcations to produce functional recombinant eukaryotic enzymes. S. cerevisiae was engineered to develop industrial scale bioreĄneries for the production of platform chemicals including organic acids, sugar alcohols, furans, isoprene and glycerol derivatives. Current progress in this Ąeld has been described in detail by Baptista et al. [9] Moreover, the successful production of a large number of high-value Ąne chemicals, such as the cholesterol-lowering drug simvastatin acid or cannabinoids has been demonstrated in S. cerevisiae. [15] [72] 1.1. THE EUKARYOTIC MODEL ORGANISM S. CEREVISIAE 3 1.1.1 Isoprenoid Production in Yeast Isoprenoids or terpenoids are the largest class of natural compounds, with numerous biological functions and industrial applications. They derive from C5 building blocks provided by isopentenyl diphosphate (IPP) and dimethylallyl diphosphate (DMAPP) and comprise mono- (C10), sesqui- (C15), di- (C20), sester- (C25), tri- (C30), tetra- (C40) and polyterpenoids, which can be composed of thousands of repeating isoprene sub- units. [99] Isoprenoids are present in all living organisms and have diverse functions. They inĆuence membrane Ćuidity, attenuate environmental stress, function as photoprotec- tion or photosynthetic pigments, are substrates in the prenylation of proteins, are signal molecules to attract or repulse organisms and regulate growth in plants. [8] Furthermore isoprenoids play an important role in many industrial applications as pharmaceuticals, nutraceuticals, agricultural chemicals, Ćavors/fragrances, chemical feedstocks, colorants and as fuels or fuel additives in petroleum industry. [102] The isomeric isoprenoid precur- sors IPP and DMAPP are supplied by the cytosolic mevalonate pathway in archea, few eubacteria (probably through horizontal gene transfer events) and eukaryotes. In addi- tion, both molecules are produced via the methylerythritol phosphate pathway, which is present in most bacteria and in chloroplasts. [99] [59] In the past, isoprenoid precur- sors derived from the yeast mevalonate pathway were used as precursors in metabolic engineering approaches in order to produce isoprenoids such as the antimalarian drug artemisinin and the antibacterial molecule limonene at high titers. [83] [23] Yeast is of- ten the preferred host organism for isoprenoid production, as yeast readily expresses functional recombinant cytochrome P450 monooxygenases, which are obligatory for the synthesis of many modiĄed isoprenoids. [59] Within the MELICOMO project mainly two isoprenoids/isoprenoid-derivatives were pro- duced via recombinant pathways in S. cerevisiae: The plant hormone gibberellin A4 (GA4) and the terpene β-carotene. As visualized in Ągure 1.1, synthesis of both molecules branches off ergosterol biosynthesis from farnesyl pyrophosphate (FPP) via the common endogenous precursor geranylgeranyl pyrophosphate (GGPP). GA4 is a plant hormone applied in agriculture e.g. to prevent russeting on apple fruits and in the production of seedless fruit. On the commercial scale it is available from fermentation extracts of the fungus Fusarium (Gibberella) fujikuroi. Nevertheless, due to puriĄcation difficul- ties, GA4 is commercially available only contaminated with GA7. [88] This is problematic because different biological effects have been described for the two gibberellins. For instance, GA7 is reported to inhibit Ćowering in apple, while GA4 promotes Ćowering in apple. [20] In order to produce pure GA4, the recombinant GA4 production pathway established during the MELICOMO project aimed to mimic GA4 production in plants. Here GA4 is generated without GA7 as unwanted side product and GGPP is converted to ent-kaurenoic acid by catalytic activity of ent-copalyl diphosphate synthase (CPS), ent- kaurene synthase (KS) and the cytochrome P450 monooxygenase ent-kaurene oxidase 1.2. OPTOGENETICS 5 The second target molecule derived from intermediates of the mevalonate pathway was the red-orange pigment β-carotene. Among other things β-carotene plays a role in photo- protection and photosynthesis in plants and as provitamin A in human nutrition. Within the MELICOMO project β-carotene was used as a well established reporter molecule to monitor the production capacity for target molecules derived from the mevalonate pathway for different yeast strains. Corresponding strains expressed the bifunctional lycopene cyclase/phytoene synthase CrtYB and the phytoene desaturase CrtI from Xan- thophyllomyces dendrorhous in order to enable β-carotene production from GGPP. In addition the geranylgeranyl pyrophosphate synthase XdCrtE was expressed to increase the availability of GGPP for β-carotene production. [10] 1.2 Optogenetics Optogenetics describes the use of genetically encoded light-responsive proteins to control cellular activity. [80] For the generation of optogenetic tools a wide variety of photorecep- tors from different host organisms have been engineered in the past to enable photoin- ducible physiological responses. The optogenetic toolbox includes proteins responsive to the visible and non-visible regions of the light spectrum, including UV, violet, blue, cyan, green, red and far-red light. [60] Essential for an optogenetic tool is a chromophore, in most cases a cofactor, which absorbs light of a speciĄc wavelength and mediates the induced photoexcitation to an effector module. Depending on the chromophore and the surrounding protein scaffold, photoexcitation can lead to the formation or cleavage of covalent bonds, change in hydrogen-bonding, photoreduction, or photoisomerization of the chromophore. [80] Thus, excitation of the chromophore induces changes in the con- formation or the general physicochemical properties of the protein. This either directly affects the function/signaling of the protein or leads to the formation or dissociation of homomeric or heteromeric protein complexes as a result of changes in interaction surfaces. In a functional optogenetic tool, the light response is transferred to an effector domain, which subsequently changes its functionality. So far several receptor and effec- tor domains have been combined to create optogenetic tools with the desired excitation properties and physiological response. The potential applications for optogenetic tools are broad and optogenetics has been used in the past e.g. to trigger gene expression by light-dependent transcription factors, to induce protein degradation, or to control neural activity by light-sensitive ion channels. [90] [89] [16] Light as an inducer of cellular processes has many advantages because it is inexpensive, in general non-toxic, adjustable in in- tensity and can be precisely controlled in a temporospatial fashion. In addition, the activation of many optogenetic tools is reversible by incubation in the dark and/or by exposure to light of a speciĄc wavelength. 8 CHAPTER 1. INTRODUCTION An optogenetic tool used in the present work containing a BLUF domain was the pho- toactivatable adenylate cyclase from Beggiatoa (bPAC). bPAC consists of a N-terminal BLUF domain and a C-terminal class III adenylate cyclase (AC) and is a popular optoge- netic tool due to its small size, low dark activity and large increase in activity upon light exposure. [68] bPAC forms a parallel dimer and dimerizes via an intermolecular coiled coil interaction of the two α3 helices in the BLUF domain regions as visualized by Ągure 1.3. When exposed to light, activation of the BLUF domains triggers rotation of the AC domains around a hinge region, opening the active site and repositioning amino acids for substrate binding. [68] Adenylate cyclases provide the cell with the secondary messenger cAMP, which activates the cAMP-dependent protein kinase (PKA). PKA activity regu- lates many cellular processes such as growth, proliferation, metabolism, stress resistance, aging, morphogenesis and signaling of nutrient availability. [18] Within the MELICOMO project Cyr1, the endogenous adenylate cyclase of S. cerevisiae, was substituted by bPAC in order to control cellular cAMP concentrations in a blue light dependent manner. Here experiments focused on growth restriction of corresponding S. cerevisiae strains in dark- ness as inactivation of the PKA leads to cell cylce arrest in the G1 phase of the cell cycle. [112] 1.3 Cell Cycle of S. cerevisiae The cell cycle is a highly conserved process among eukaryotes and consists of G1 (gap 1), S (synthesis), G2 (gap 2) and M (mitotic) phase. During the S. cerevisiae G1 phase, the cell grows and assesses at the ŤStartŤ checkpoint whether environmental and internal conditions are met to initiate cell cycle progression or to delay/arrest the cell cycle. [4] When all requirements are met, the cell prepares to enter the S phase, where DNA is replicated in preparation for mitosis. After DNA replication is complete, the cell transitions to G2 phase, continues to grow and repairs DNA defects before entering M phase. The mitotic phase is further divided into prophase (chromosome condensation, formation of the spindle apparatus), prometaphase (attachment of the spindle apparatus to the chromosomes, breakdown of the nuclear envelope), metaphase (alignment of the chromosmes at the metaphase plate), anaphase (separation of sister chromatids), telophase (formation of the nuclear envelope), followed by cytokinesis (separation of daughter cells), after which the two newly formed cells re-enter G1 phase. [105] 10 CHAPTER 1. INTRODUCTION The desired production state for S. cerevisiae cells in the MELICOMO project was cell cy- cle arrest, and several different cell cycle mutant (CCM) strains were generated during the project. Within the present work three different CCM strains were tested. As described in section 1.2, for bPAC cells G1 arrest is induced at ŤStartŤ by incubation in darkness due to PKA inactivation. [112] For the Clb2ΔDB-psd3 CCM strain an additional copy of CLB2 is expressed lacking the destruction box (DB) and fused to the 3rd generation photo- sensitive degron (psd). Deletion of the destruction box, disables APC-mediated targeted degradation of the protein during the cell cycle and brings Clb2ΔDB-psd3 degradation under blue light control. Under restrictive, dark growth conditions, Clb2ΔDB-psd3 accu- mulates in the cell and the cell cycle arrests at the transition from metaphase to anaphase, as Clb2ΔDB-psd3 prevents the cell from exiting M phase. [89] [13] For the Cdc48-psd3 strain endogenous Cdc48 was genetically fused to psd3. The essential AAA-ATPase Cdc48 par- ticipates in many cellular processes, including endoplasmic reticulum-associated degra- dation (ERAD), where it is central part of an adaptor complex to extract misfolded and ubiquitylated protein from the endoplasmic reticulum. [25] Conditional Cdc48 mutants show cell cycle arrest at the metaphase-anaphase transition, which can be induced by blue light in Cdc48-psd3. [89] During anaphase Cdc48 is reported to promote the nuclear localization of tyrpe I phosphatase to antagonize Aurora B kinase activity. [24] In addi- tion, Cdc48 and its adaptor Ubx4 affect proteasome distribution, leading to impaired degradation of B-cyclins, which prevents the cell from undergoing cytokinesis. [25] 1.4 Scope of the Project The project Ťmetabolic engineering with light controlled modulesŤ (MELICOMO) aimed to harness optogenetic tools for the production of valuable secondary metabolites in S. cerevisiae. Within the project PhD students supervised by Prof. Dr. L.-O. Essen and Dr. Christof Taxis at Philipps University Marburg collaborated with the working group of Prof. Dr. Zoran Nikoloski at the university of Potsdam. The team in Marburg contributed to the project by generating corresponding S. cerevisiae production strains and by ad- vancing respective optogenetic tools. The role of the working group of Prof. Dr. Zoran Nikoloski was to perform metabolic Ćux calculations adapted to the light-controlled yeast strains in order to identify bottlenecks for the production of the desired molecules. The MELICOMO project relied strongly on data for the evaluation of production capacities for different yeast strains and as a solid basis for Ćux modeling calculations. Hence, the present work aimend to establish reliable analytical workĆows, providing required experimental data on: 1. Biomass composition of cell cycle arrested S. cerevisiae for metabolic Ćux modeling. 2. Effects of growth restriction on the proteome for S. cerevisiae cell cycle mutants. 3. Analyte quantiĄcation for S. cerevisiae production strains. 11 Chapter 2 Results 2.1 Biomass Analysis of Light-Controlled S. cerevisiae Within this work a series of protocols for the quantiĄcation of S. cerevisiae biomass com- position was implemented from literature. Protocols were down-scaled for application on 96-well plates and modiĄed to suit the instrumentation provided by the university. The main goal was to investigate differences in biomass composition between cell cycle mutant (CCM) and wild type (WT) cells. Resulting protocols were used to provide a robust data basis for metabolic Ćux modeling performed by the group of Prof. Dr. Zo- ran Nikoloski at the university of Potsdam. Thereby a full analysis quantifying protein, carbohydrates, DNA, RNA and lipids was conducted for Clb2ΔDB-psd3 cells grown under restrictive conditions and the corresponding WT strain. Furthermore, protein content was quantiĄed for cells of the Clb2ΔDB-psd3, Cdc48-psd3 and bPAC CCM strains, grown under restrictive and permissive conditions, including the corresponding WT strains. 2.1.1 Cell Cultivation In order to investigate differences in biomass composition between CCM and WT cells Clb2ΔDB-psd3 was grown under restrictive conditions and corresponding WT cells were cultured as described in section 4.1.1. Obtained Clb2ΔDB-psd3 cells showed the distinct enlarged budding phenotype as visualized in Ągure 2.1. [89] In general, WT cells grew to a higher OD (e.g. OD WT (batch2) ≈ 6.1 vs. OD Clb2 (batch2) ≈ 3.3) but interestingly the differ- ence in dry mass was less prominent (e.g. m WT (batch2) ≈ 2.3 g/replicate vs. m Clb2(batch2) ≈ 1.5 g/replicate). Accordingly, normalized to OD the Clb2ΔDB-psd3 cells had 20 % more dry mass compared to WT cells. 2.1. BIOMASS ANALYSIS OF LIGHT-CONTROLLED S. CEREVISIAE 19 2.1.7.1 Protein QuantiĄcation of S. cerevisiae Strains for Flux Modeling In order to supply a robust data basis for the metabolic Ćux modeling performed by the group of Prof. Dr. Zoran Nikoloski, protein content was determined for Cdc48-psd3, Clb2ΔDB-psd3 and bPAC cells grown under permissive and restrictive condition and for corresponding WT strains. Cells were cultured as described in section 4.2.1 and freeze dried after the washing procedure. From the obtained cell pellets protein content was determined using the Biuret assay. Table 2.2 summarizes the results of the analysis. Table 2.2: Protein content of the examined CCM strains grown under blue light condi- tions or in darkness under cultivation conditions of the proteomics experiment. darkness blue light SK1 18.8 ± 1.3 % 18.8 ± 1.3 % bPAC 18.2 ± 0.1 % 20.9 ± 0.4 % ESM356 22.0 ± 0.5 % 22.4 ± 0.3 % Cdc48-psd3 20.0 ± 0.3 % 20.0 ± 0.6 % ESM356 Leu 19.8 ± 0.5 % 19.7 ± 0.4 % Clb2ΔDB-psd3 21.2 ± 0.5 % 20.0 ± 1.0 % Results for protein quantiĄcation for cells grown under the conditions of the proteomics experiment were signiĄcantly lower compared to the results summarized by section 2.1.2. This is probably due to major differences in cultivation conditions of the cells. For the results shown by table 2.2 cultures of 50 mL were grown in TC Ćasks for 12 h under permissive conditions followed by 12 h under restrictive conditions. The results of section 2.1.2 were obtained for cultures grown for 3 h under permissive conditions followed by 24 h under restrictive conditions in 2 L in glass Ćasks. 20 CHAPTER 2. RESULTS 2.2 Proteomics of Light-Controlled S. cerevisiae In previous experiments, cell cycle mutants (CCMs) showed an increased production of valuable target molecules in comparison to the corresponding wild type (WT). Here, re- markable differences in product yield between restrictive and permissive light conditions were observed. [92,100] The increased product yield of CCM cells grown under restrictive conditions could be assigned to an increased energy availability for the production of secondary metabolites due to the cell cycle arrest. Furthermore, shifts in the proteome of CCM cells towards enzymes involved in the production of target molecules could ex- plain the increased product yields. In order to investigate into reasons for the higher productivity of restricted CCM cells on the proteome level, a series of proteomics experiments was performed for the Cdc48-psd3, Clb2ΔDB-psd3 and bPAC strain and corresponding WT strains as described in section 4.2. Culture ODs and protein content determined from dry mass are summarized in tables 5.14 and 5.15 in the appendix section. This data was later used together with results of the proteomics experiment for modeling calculations, which were performed by the group of Prof. Dr. Zoran Nikoloski at the university of Potsdam. Within the proteomics experiment a total number of 24 LC-MS/MS measurements were performed for each of the three CCM strains. Here, each 2 h MS/MS measurement generated about 3 GB of raw data using the timsTOF Pro instrument and was analyzed using the PEAKS software. Exemplary speciĄed for the bPAC strain and the corre- sponding WT, the dataset contained ≈ 225 million MS spectra and ≈ 4 million MS/MS spectra, which were assigned to ≈ 2 million de novo peptide spectra in the de novo analysis step of the software. During PEAKS analysis ≈ 54.000 peptides were mapped to the database and were assigned to 3546 of the 5441 protein sequences, which were provided by the database. Subsequently, in the label-free quantiĄcation (LFQ) step of the analysis, peptides were Ąltered for quality and average area according to guidelines of the software manufacturer. [3] Finally LFQ results were exported for 1019 proteins meeting the LFQ Ąlter criteria listed in table 4.6 and plotted as heatmaps. Results for the analysis of the proteomics datasets for all examined CCM yeast strains are summarized by table 2.3. LFQ results were generated for the Cdc48-psd3 and Clb2ΔDB-psd3 strains with identical Ąlter parameters. For bPAC stricter Ąlter criteria were applied due to a higher data quality. 2.2. PROTEOMICS OF LIGHT-CONTROLLED S. CEREVISIAE 21 Table 2.3: Summary of proteomics data. Assigned proteins are denoted relative to the number of proteins provided by the corresponding FASTA database. analyzed CCM strain mapped peptides assigned proteins Ąltered proteins Cdc48-psd3 78.001 4388/5924 1962 Clb2ΔDB-psd3 83.332 4387/5923 2152 bPAC 54.314 3546/5441 1018 LFQ results were analyzed in a pairwise manner comparing two sample groups (e.g. CCM cultured under restricted growth condition with the corresponding WT) and volcano plots were generated for the resulting abundance ratios. From the comparisons, proteins were identiĄed showing a signiĄcant abundance difference (p-value of the StudentŠs t-test < 0.01 and mean abundance difference > 10 %). These differentially abundant proteins (DAPs) were further analyzed and used as input for GO-term enrichment analyses to reveal affected cellular pathways, functions, processes and components in the comparison of both sample groups. 2.2.1 Heatmaps of Relative Protein Abundance Within the analysis workĆow of proteomics data, large heatmaps were used to visualize protein abundance of biological and technical replicates in order to provide an overview of the overall outcome of the experiment. Heatmaps include abundance of all proteins meeting the quality parameters of the PEAKS analysis and were automatically generated in the LFQ step by the PEAKS software. Nevertheless heatmaps generated by PEAKS allow only the export to .png format, hindering customization and further analysis. In order to solve this problem, within this work R scripts were established to generate heatmaps from exported raw protein abundances. Heatmaps were generated for all three CCM and corresponding WT strains cultured under the respective restrictive and permissive light condition of the CCM strain. Each heatmap summarizes a total of 24 measurements (four sample groups, grown as bio- logical triplicates, measured in duplicate). In addition to the heatmaps visualizing all 24 data sets, condensed heatmaps were created from averaged abundances of technical duplicates. 2.2. PROTEOMICS OF LIGHT-CONTROLLED S. CEREVISIAE 23 The generated heatmaps illustrate well the inĆuence of blue light on the proteome of Cdc48-psd3 cells. Here, large sections of the heatmaps were impacted in the comparison of Cdc48-psd3 grown under restrictive (30 µmol m2 ·s blue light) and permissive light con- ditions (darkness), thereby exhibiting clear differences in protein abundance. Between Cdc48-psd3 grown under restricted and permissive condition, the abundance pattern of Cdc48-psd3 cultured under permissive growth conditions related stronger to the abun- dance pattern of the WT strain. Hence Cdc48-psd3 grown under permissive condition appeared to be more WT-like than restricted Cdc48-psd3 on the proteome level. This observation will be further analyzed in section 2.2.2.1. Lacking optogenetic modiĄcations the WT did not exhibit obvious differences between the two illumination conditions. No- ticeable in Ągure 2.11A is distinct pattern of abundance ratios effecting every duplicate measurement with varying impact on different proteins. This behavior is probably due to the very long sample list, which meant that each technical replicate was measured four days after the Ąrst replicate. In the meantime, apparently instrumental or sample conditions changed causing the pattern structure in the heatmap. These changes in abundance are remarkably similar within the biological replicates and the heatmap in Ągure 2.11B (which is based on mean abundance of the technical duplicates) shows a uniform abundance pattern for biological replicates. Since the mean values of techni- cal replicates are used for downstream analyses, the data generated for the Cdc48-psd3 strain should provide a solid basis. 2.2.1.2 Heatmaps Ű Clb2ΔDB-psd3 Heatmaps generated for the Clb2ΔDB-psd3 strain and the corresponding WT strain back- ground visualized in Ągure 2.12 are remarkably pale due to a overall similar protein abundance for most proteins over the entire analyzed sample set. Both heatmaps do not show an obvious visual pattern in order to distinguish between Clb2ΔDB-psd3 grown in darkness (restrictive growth condition) and under 30 µmol m2 ·s blue light (permissive growth condition). Furthermore, only slight differences on the proteome level are implied in the comparison between Clb2ΔDB-psd3 and the WT, visible in the top and bottom third of the heatmap. The distinct stripe pattern described for the proteomics experiment for Cdc48-psd3 is also present in the heatmap of Ągure 2.12A and appears to be a little more prominent due to missing colorful abundance clusters in the center of the heatmaps. Fi- nal conclusions about the outcome of the proteomics experiment for the Clb2ΔDB-psd3 strain will be drawn after the evaluation of protein abundances below. 26 CHAPTER 2. RESULTS 2.2.2 Volcano Plots and Differentially Abundant Proteins Volcano plots are scatter-plots used in proteomics to visualize fold changes and allow signiĄcance Ąltering for two related sample groups. Within this work volcano plots were generated both for the comparison of the growth-restricted CCM strain and the cor- responding WT strain as well as for the growth-restricted CCM and the CCM strain cultured under permissive conditions. For both comparisons a positive log2(ratio) refers to a higher protein abundance for the growth-restricted CCM. In the plots the thresh- old for differentially abundant proteins (DAPs) (p < 0.01, mean abundance difference > 10 %) is visualized. Highlighted in the plots are proteins involved in pathways for pre- cursor generation of gibberellin A4 (GA4), β-carotene and cordycepin, which are target molecules produced by the light-controlled production strains within the MELICOMO project. The synthesis of gibberellins and β-carotene branches off the ergosterol biosyn- thesis superpathway after the mevalonate pathway from farnesyl pyrophosphate (FPP) via geranyl geranyl pyrophosphate (GGPP), as visualized in Ągure 1.1 of the introduc- tion section. In the volcano plots, enzymes upstream of GGPP and proteins downstream of FPP involved in the biosynthesis of ergosterol are highlighted. Furthermore, the re- pressible vacuolar alkaline phosphatase Pho8 is highlighted in the volcano plots, whose depletion was found to increase the availability of the cordycepin precursor 3ŠAMP. [52] The full list of highlighted proteins and plotted data is summarized in tables 5.16 to 5.18 in the appendix section. Furthermore, obtained DAPs were used for Venn diagrams in order to visualize the over- lap of different comparisons. Here, DAPs were excluded from the cross section of the Venn diagram in case of an opposite abundance proĄle (e.g exclusion of DAP with higher abundance in the comparison to the WT and lower abundance in the comparison to the permissive condition). 2.2.2.1 Volcano Plots and DAP Analysis Ű Cdc48-psd3 For the Cdc48-psd3 proteomics analysis 1962 proteins were meeting the export criteria of the PEAKS software. Regarding the comparison of Cdc48-psd3 grown under restric- tive light conditions and the WT strain 504 DAPs were identiĄed. Here, remarkably the acetyl-CoA synthetase Acs2 exhibited twice the abundance for growth-restricted Cdc48-psd3 compared to the WT strain. Additionally, the proteins Erg13 and Hmg1 of the mevalonate pathway were higher abundant in Cdc48-psd3 cells. Furthermore, the proteins Erg9, Erg1, Erg11, Erg25 and Erg3 downstream of FPP in the ergosterol biosyn- thesis superpathway were among the DAPs with higher abundance for the Cdc48-psd3 strain. Noteworthy, these proteins appear in the region of the volcano plot with high -log10(p) values and log2(ratio). DAPs lower abundant in Cdc48-psd3 cells were Erg26, an enzyme downstream of FPP, and the protein Pho8 involved in the depletion of 3ŠAMP. In the comparison of Cdc48-psd3 grown under restrictive and permissive condition only 30 CHAPTER 2. RESULTS 2.2.2.3 Volcano Plots and DAP Analysis Ű bPAC For the proteomics analysis of the bPAC strain, protein abundances for 1018 proteins were exported using the PEAKS software. In the comparison of restricted bPAC and the WT 506 proteins were identiĄed as DAPs. Hence, abundances of about half of the ex- amined proteins were found to be signiĄcantly different between the two strains. For this comparison Acs1 and Erg13 upstream of GGPP were higher abundant for the restricted bPAC strain. Here, Acs1 stands out with a six-fold increased abundance (log2(6) ≈ 2.6), which was among the highest abundance differences observed within the experiment. Additionally, in the biosynthesis pathway downstream of FPP towards ergosterol biosyn- thesis Erg9 abundance was found to be increased for restricted bPAC compared to the WT strain. For the WT strain Erg20 and Erg26 were found to be higher abundant. In the comparison of the bPAC strain grown under restrictive and permissive conditions 527 DAPs were identiĄed. Highly abundant in the restricted bPAC strain was Acs1, which was also highly abundant in the comparison to WT cells as mentioned above. Furthermore, Pho8 involved in the precursor depletion of cordycepin was among the DAPs and of increased abundance for restricted bPAC cells. For the bPAC strain grown under permissive conditions Acs2, Erg19 and Erg20 upstream of FPP and Erg11, Erg26 and Erg2 downstream of FPP were found to be of higher abundance. Figure 2.18 shows volcano plots for the exported proteins of both comparisons. Data of highlighted proteins in both plots is summarized in table 5.18 in the appendix section. In the comparison between bPAC grown under permissive conditions and the WT strain 625 DAPs were identiĄed, thereby exceeding the number of DAPs found for the com- parison of bPAC grown under restrictive conditions and the WT strain. This observation underlines the impact of Cyr1 substitution by bPAC for this strain and the broad impact of light dependent cAMP production on the proteome. 32 CHAPTER 2. RESULTS dance for growth-restricted Cdc48-psd3 cells in the comparisons to the WT strain and to Cdc48-psd3 cells grown under permissive conditions. This observation is important because the biosynthetic pathways for GA4 and β-carotene branch off this pathway. For the Clb2ΔDB-psd3 strain a total of only two enzymes of the ergosterol biosynthesis super- pathway were identiĄed as DAPs. Here, one protein was higher abundant and the other was lower abundant for growth-restricted Clb2ΔDB-psd3 in comparison to Clb2ΔDB-psd3 grown under permissive condition. In regard to the bPAC strain, identiĄed DAPs of the superpathway were distributed evenly between high and low abundant for growth- restricted bPAC cells in the comparison of the growth-restricted bPAC strain to the WT strain. In the comparison of bPAC cultured under restrictive and permissive light condi- tion more of these DAPs were found to be low abundant than high abundant for bPAC grown under restrictive conditions. Additionally, Acs2 for the Cdc48-psd3 strain and Acs1 for the bPAC strain were found to be higher abundant for the restricted CCM both in the comparisons between restricted CCM and WT and in the comparisons of CCM grown under restrictive and permissive condition. The two acetyl-CoA synthetase isoforms pro- vide the ergosterol biosynthesis superpathway with the initial precursor acetyl-CoA. Furthermore, Pho8, involved in the depletion of the cordycepin precursor 3ŠAMP, was among the DAPs for the Cdc48-psd3 strain. In the comparison of Cdc48-psd3 cells grown under restrictive conditions to the WT strain, Pho8 was found to be less abundant for Cdc48-psd3 cells, thereby potentially increasing the availability of 3ŠAMP for cordycepin synthesis. However, in the comparison of the Cdc48-psd3 strain cultured under restric- tive and permissive growth conditions, Pho8 was higher abundant for growth-restricted Cdc48-psd3 cells. In the comparison of bPAC cells cultured under restrictive and permis- sive growth condition, Pho8 was found to be higher abundant for the growth-restricted bPAC strain. To examine the distribution of protein ratios in the volcano plots, the average fold changes of protein abundance were calculated. Here, reciprocal values were calculated for ratios less than 1 before averaging. The average fold changes of protein abundance for all identiĄed proteins and the total number of identiĄed DAPs for all CCM strains examined are summarized in table 2.4. 2.2. PROTEOMICS OF LIGHT-CONTROLLED S. CEREVISIAE 33 Table 2.4: IdentiĄed DAPs for different CCM strains relative to the restricted condition of the CCM strain and average fold change of protein abundance in pairwise comparisons. DAPs high low average comparison abundant abundant fold change Cdc48-psd3(restrictive) 271 233 1.25 WT Cdc48-psd3(restrictive) 138 56 1.18 Cdc48-psd3(permissive) Clb2ΔDB-psd3(restrictive) 84 23 1.12 WT Clb2ΔDB-psd3(restrictive) 13 38 1.10 Clb2ΔDB-psd3(permissive) bPAC (restrictive) 420 86 1.41 WT bPAC (restrictive) 237 290 1.37 bPAC (permissive) As shown in table 2.4, average fold changes of protein ratios were the highest for the bPAC strain, followed by the Cdc48-psd3 strain. For the Clb2ΔDB-psd3 strain, the lowest average fold changes were calculated. For each yeast strain, the average fold change in protein ratios was similar between the two comparisons made, but slightly higher for the comparison between restrictive CCM and WT strain than for the comparison between restrictive and permissive growth condition. In order to investigate into similarities between the different CCM strains Venn diagrams were generated comparing DAPs obtained in the comparison to the WT strain and between restrictive and permissive growth condition. 2.2. PROTEOMICS OF LIGHT-CONTROLLED S. CEREVISIAE 35 comparison of the respective CCM strain grown under restrictive conditions and the WT strain. Furthermore, for the Cdc48-psd3 and the bPAC strain the cyclin-dependent ki- nase Cdk1 and the fumarase Fum1 of the tricarboxylic acid (TCA) cycle were found to be higher abundant in comparison to the WT strain. Nevertheless, in the comparison of bPAC cells cultured under restrictive and permissive conditions, Cdk1 was found to be lower abundant for the growth-restricted bPAC strain. Overlapping DAPs between the Clb2ΔDB-psd3 and the bPAC strain in comparison to the WT strain were strongly related to translation processes and were found to be of higher abundance for the WT strain compared to the restricted CCM strains. The identiĄed DAPs for the analysis of CCM strains grown under restrictive and per- missive conditions showed only a small number of overlaps. As visualized by the Venn diagram in Ągure 2.20, no common DAPs between the analyses of all three CCM strains were identiĄed. Nevertheless, as already mentioned above, Erg3 was higher abundant under restrictive conditions for the Cdc48-psd3 and Clb2ΔDB-psd3 strain. In the com- parison of DAPs for the Cdc48-psd3 and bPAC strain Ubx1 was higher abundant for cells grown under restrictive conditions. This Ąnding matches the observation made for the comparison of restricted cells to the WT strain, where Ubx1 was found with higher abundance for both CCM strains cultured under restrictive growth conditions. Additionally, for both strains grown under restrictive conditions, Fum1 and Lpd1, two enzymes of the TCA cycle, were higher abundant in comparison to cells grown under permissive conditions. Whereas the catalytic subunit of the acetolactate synthase Ilv2, which is involved in fermentation processes, and the 6-phosphogluconate dehydroge- nase Gnd1 of the pentose phosphate pathway were higher abundant for Cdc48-psd3 and bPAC cells grown under permissive conditions. Ilv2 and Gnd1 were also found to be lower abundant for the bPAC strain grown under restrictive conditions in comparison to the WT strain. Furthermore, as mentioned earlier, Pho8 involved in the depletion of the cordycepin precursor 3ŠAMP was higher abundant for Cdc48-psd3 and bPAC cells grown under restrictive conditions in comparison to permissive growth conditions. In the comparison of DAPs for the Clb2ΔDB-psd3 and the bPAC strain, the NADH-cytochrome b5 reductase Mcr1 involved in ergosterol production was found to be higher abundant for both strains grown under restrictive growth conditions but was found to be lower abundant for growth-restricted Cdc48-psd3 cells in the comparison to the WT and to Cdc48-psd3 cells grown under permissive light conditions. [63] A detailed overview of the DAPs summarized in Ągure 2.20 is provided by tables 5.19 and 5.20 in the appendix section. In addition to the DAPs mentioned earlier, the two PKA subunits Tpk1 and Tpk2 were low abundant for Cdc48-psd3 cells cultured under restrictive light condition in comparison to the WT strain, whereas the third subunit was not found in the dataset. This Ąnding might imply a lower PKA activity for Cdc48-psd3 cells in comparison to the WT strain, which would link the Cdc48-psd3 and the bPAC strain, where PKA activity is controlled by blue light dependent cAMP production. 36 CHAPTER 2. RESULTS 2.2.2.5 Heatmaps for DAPs of Interest Heatmaps in Ągure 2.21 were generated to visualize relative protein for the DAPs of interest identiĄed above and show great homogeneity of abundance ratios among the biological triplicates for each sample group. This is reasonable because DAP criteria are demanding p < 0.01 for a StudentŠs t-test comparing the abundance of the triplicates of both sample groups. In terms of abundance difference Erg1, Erg3, Erg11 and Erg25 stand in the comparison of growth-restricted Cdc48-psd3 cells with the WT strain. In the comparison of Cdc48-psd3 cells grown under restrictive and permissive growth condition, Erg25 and Erg3 exhibit the highest abundance differences for the DAPs of interest. For the Clb2ΔDB-psd3 strain, only small abundance differences between the selected DAPs are visualized by the heatmaps. In regard to the bPAC strain Acs1 exhibits noticeably higher abundance for restricted bPAC cells in the comparisons to the WT strain and to the permissive growth condition. In addition, abundances for proteins involved in ergosterol production were mapped on the mevalonate pathway and the ergosterol biosynthesis superpathway. Corresponding heatmaps were generated for the Cdc48-psd3 and bPAC strains and are summarized by Ągure 3.1 in the appendix section in order to avoid redundance. These heatmaps also visualize relative protein abundance for non-DAP pathway proteins and will be discussed in section 3.2. 38 CHAPTER 2. RESULTS 2.2.2.6 DAPs Involved in Pathways of the Central Metabolism Metabolites and energy provided by the central metabolism are crucial for the metabolic capacity of the desired light-controlled production strains. Hence, the obtained pro- teomics data was analyzed for abundances of proteins involved in the TCA cycle, res- piratory chain, glycolysis/glucose fermentation and the pentose phosphate pathway. In a Ąrst step, proteins involved in the mentioned pathways were analyzed for overlaps with identiĄed DAPs for each CCM yeast strain and comparison. The corresponding protein targets of the central metabolism and overlaps with the respective DAPs are summarized in tables 5.21 to 5.25 in the appendix section. Obtained results were then used to identify pathways which were particularly impacted in speciĄc comparisons. For these comparisons, heatmaps were generated in order to visualize relative abundance for all proteins involved in the speciĄc pathway, thereby helping to understand the overall pathway regulation. For the Cdc48-psd3 strain, many proteins involved in glycolysis/glucose fermentation were found to be lower abundant for growth-restricted Cdc48-psd3 in comparison to the WT strain. Nevertheless, the alcohol dehydrogenases Adh1 and Adh3 responsible for ethanol synthesis were found to be higher abundant for growth-restricted Cdc48-psd3 in this comparison. As visualized by the heatmap in Ągure 2.22, moderate abundance differences were observed in the analysis. For the bPAC strain, signiĄcantly less DAPs involved glycolysis/glucose fermentation were identiĄed. Most of them were identiĄed in the comparison between the restrictive and the permissive growth condition. Here, the hexokinase II (Hxk2) and both subunits of the phosphofructokinase (Pfk1 and Pfk2) were found to be ≈ 20 % lower abundant for growth-restricted bPAC. Furthermore, the pyruvate decarboxylase Pdc5 was found to be less abundant, whereas the enolase Eno1 and the alcohol dehydrogenase Adh1 were found to be higher abundant for growth- restricted bPAC in comparison to the permissive condition. In the comparison of the growth-restricted Clb2ΔDB-psd3 strain to the permissive growth condition, many DAPs assigned to the TCA cycle and the respiratory chain were found to be low abundant for the growth-restricted Clb2ΔDB-psd3 strain. Even though these abundance trends were clearly visible on the corresponding heatmaps visualized in Ąg- ure 2.22, abundance differences between both light conditions were relatively small. However, this observation was not made comparing both light conditions of the WT strain (data not shown), which could have indicated a more general effect caused by the cultivation conditions. For the bPAC strain, rather strong abundance differences were observed for proteins involved in the TCA cycle and the respiratory chain. Here, most DAPs were high abun- dant in the comparisons of the growth-restricted bPAC strain to the WT strain and the permissive growth condition. Corresponding heatmaps visualizing protein abundances for both pathways are shown in Ągure 2.23. 2.2. PROTEOMICS OF LIGHT-CONTROLLED S. CEREVISIAE 41 2.2.3 Gene Ontology (GO) Enrichment Analysis The targeted DAP analysis revealed DAPs of interest for speciĄc metabolic pathways, nevertheless the overall picture of differences between sample groups remained unclear. Each comparison of protein abundances in section 2.2.2 for Clb2ΔDB-psd3 resulted in dozens and for Cdc48-psd3 and for the bPAC strain in hundreds of identiĄed DAPs. Corresponding lists of DAPs are almost impossible to curate manually and a Gene On- tology (GO) enrichment analysis was performed in order to statistically evaluate which processes, functions and components of the cell were affected in the comparison of both sample groups. Results of the GO enrichment analyses were plentiful and needed to be condensed. Here, representative, non redundant GO-terms were selected to visualize the overall outcome of the GO enrichment analysis. 2.2.3.1 GO-Terms Ű Cdc48-psd3 For the Cdc48-psd3 strain, a total of 109 statistically signiĄcant GO-terms were iden- tiĄed in the comparison of Cdc48-psd3 cells grown under restrictive conditions and the WT strain. Corresponding representative GO-terms are summarized in Ągure 2.24. For the comparison of the Cdc48-psd3 and the WT strain shifts in the abundance of proteins involved in α-amino acid metabolic processes (GO:1901605) were observed. Further- more, the term Ťoxidoreductase activity, acting on the CH-OH group of donors, NAD+ or NADP+ as acceptorŤ (GO:0016616) was obtained implying changes on catabolic and anabolic activities between both sample groups. For proteins involved in alcohol biosynthesis processes (GO:0046165), relative protein abundances were found to be in- creased for the growth-restricted Cdc48-psd3 strain in comparison to the WT strain. In terms of the allocation of DAPs to speciĄc cellular components, the ribonucleopro- tein granule (GO:0035770) exhibited a high enrichment factor. Proteins associated to RNA granules control the localization, stability, and translation of their RNA cargo and might imply a stress response of growth-restricted Cdc48-psd3 cells. Furthermore, the GO-term (GO:0044432) Ťendoplasmic reticulum partŤ was identiĄed (p-value: 0.006, enrichment: 1.80, data not shown). Here, 37 of the 53 associated DAPs were higher abundant for growth-restricted Cdc48-psd3 cells. In the comparison of the Cdc48-psd3 strain grown under restrictive and permissive con- ditions, a total of only 37 GO-terms were identiĄed. As in the comparison to the WT strain the term Ťα-amino acid metabolic processesŤ (GO:1901605) was identiĄed as GO- term for the list of provided DAPs. Here, for the most part protein abundances were higher for Cdc48-psd3 cells grown under restrictive conditions. Furthermore, the GO- term Ťoxidoreductase activityŤ (GO:0016491), a parental term of GO:0016616 (GO-term found in the comparison to the WT strain), was among the results of the GO-term en- richment analysis. In regard to the allocation of DAPs to cellular components, the GO enrichment analysis pointed towards the endoplasmic reticulum (GO:0005783). Shared 44 CHAPTER 2. RESULTS 2.2.3.3 GO-Terms Ű bPAC For the growth-restricted bPAC strain 203 GO-terms were identiĄed in the to the WT strain. Here, the GO-term Ťsmall molecule biosynthetic processŤ (GO:0044283) exhib- ited more high abundant DAPs for bPAC in comparison to the WT strain. As observed for the Cdc48-psd3 strain the term Ťoxidoreductase activity, acting on the CH-OH group of donors, NAD+ or NADP+ as acceptorŤ (GO:0016616) showed high enrichment, im- plying impacts on catabolism and anabolism in the comparison of both sample groups. Furthermore, the term Ťcytoplasmic translationŤ (GO:0002181) was obtained for the GO enrichment analysis. Here, all corresponding DAPs were higher abundant for growth- restricted bPAC cells in comparison to the WT strain. In regard to the cellular allocation of DAPs the term Ťribonucleoprotein granuleŤ (GO:0035770) was identiĄed. This Ąnd- ing is in accordance with observations made for the Cdc48-psd3 strain in comparison to the WT strain and may indicate a general stress response of growth-restricted CCM cells. The GO enrichment analysis for the comparison of bPAC cells grown under restrictive and permissive conditions resulted in 233 identiĄed GO-terms. As in the comparison with the WT strain the GO-term Ťsmall molecule biosynthetic processŤ (GO:0044283) was identiĄed, but in contrast DAPs were found to be mostly of low abundance for the restrictive growth condition in comparison to the permissive condition. Furthermore, as observed in the comparison to the WT the term Ťoxidoreductase activity, acting on the CH-OH group of donors, NAD+ or NADP+ as acceptorŤ (GO:0016616) was identi- Ąed. Another GO-term with high enrichment factors was ŤdetoxiĄcationŤ (GO:0098754). Here, DAPs were of higher abundance mostly for growth-restricted bPAC cells indicating cellular stress due to the lack of cAMP supply. Nevertheless, DAPs associated to the generation of precursor metabolites and energy (GO:0006091) were found to be for the most part of higher abundance for the growth-restricted bPAC strain, which indicates a change in energy availability for growth-restricted bPAC cells in comparison to the per- missive condition. This Ąnding is in accordance to the results described in section 2.2.1. Here many proteins involved in the TCA cycle and the respiratory chain were found to be high abundant in the comparison of bPAC cells cultured under restrictive and permissive conditions. Furthermore, the term Ťergosterol metabolic processŤ (GO:0008204) relat- ing to the production of β-carotene and the GA4 was obtained. Hence, for the bPAC strain, induced growth-restriction may have a direct impact on the production of valu- able target molecules within the MELICOMO project. This Ąnding is further discussed in section 3.2. 46 CHAPTER 2. RESULTS 2.2.4 Proteomics of Light-Controlled S. cerevisiae Ű Summary Proteomics data obtained for the Cdc48-psd3, Clb2ΔDB-psd3 and bPAC CCM yeast strains had a good coverage in regard to the provided data base of S288c/SK1 gene products and was a solid data basis for further analysis after quality Ąltering. For CCM cells of the Cdc48-psd3 and bPAC strain, a clear impact of the light condition on the pro- teome was observed. Furthermore, both CCM strains were clearly distinguishable from corresponding WT strains as demonstrated by the heatmaps, respectively. In contrast for the Clb2ΔDB-psd3 strain differences between restrictive and permissive growth condition of Clb2ΔDB-psd3 cells and in general differences between the Clb2ΔDB-psd3 and the WT strain were signiĄcantly smaller. In order to examine beneĄcial differences on the proteome level for the production of the target molecules within the MELICOMO project, DAPs were analyzed for the CCM yeast strains. The analysis of DAPs revealed that many proteins of the ergosterol biosynthesis superpathway were differentially abundant for the examined comparisons of Cdc48-psd3 and the bPAC strain. This Ąnding implies different production capacities for the CCM and the WT strains for β-carotene and GA4, as precursors of both molecules branch off the ergosterol biosynthesis superpathway. For the Cdc48-psd3 strain most of the pathway proteins were higher abundant for Cdc48-psd3 cells cultured under restrictive growth condition in comparison to the WT strain and the permissive growth condition. For growth-restricted bPAC cells, identiĄed DAPs were evenly distributed in comparison to the WT strain and slightly less abundant in comparison to the permissive light condition. These results are discussed in more detail in section 2.26. Interestingly, an acetyl-CoA synthetase isoform was of high abundance in both comparisons for Cdc48-psd3 and bPAC grown under restrictive light conditions. This Ąnding may imply an increased availability of acetyl-CoA as the initial precursor of the mevalonate pathway with beneĄcial effects on the production of β-carotene and GA4. Furthermore, Pho8 involved in the depletion of the cordycepin precursor 3ŠAMP was found to be of lower abundance for growth- restricted Cdc48-psd3 cells in comparison to the WT strain. The analysis of relative abundances for proteins of the central metabolism revealed that enzymes involved in glycolysis/glucose fermentation were found to be overall lower abundant in the comparison of the growth-restricted Cdc48-psd3 and the WT strain. Furthermore, many proteins involved in the TCA cycle and the respiratory chain were found to be lower abundant for growth-restricted Clb2ΔDB-psd3 cells in comparison to the permissive condition. Nevertheless, differences in abundance for these comparisons were rather small. For bPAC cells cultured under restricted growth condition overall higher abundances for proteins involved in the TCA cycle and the respiratory chain were observed in the comparisons to the permissive condition and to the WT strain. These Ąndings might indicate a in general higher energy and precursor availability for the bPAC strain under restrictive growth conditions. 2.2. PROTEOMICS OF LIGHT-CONTROLLED S. CEREVISIAE 47 In the GO-term analysis for DAPs of both analyzed comparisons for the Cdc48-psd3 strain, abundance differences for proteins involved in the metabolic process of α-amino acids and oxidoreductase activity were observed. For the bPAC strain, small molecule biosynthetic processes and oxidoreductase activity were impacted as found in the GO- term analysis. These observations point towards a light-dependent different metabolic state for both CCM strains, which might be an explanation the increased production of the desired target molecules within the MELICOMO project. Summarizing the results of the proteomics experiment for light-controlled S. cerevisiae, meaningful proteomic insights were obtained for the Cdc48-psd3 and the bPAC strains. These Ąndings help to explain the increased productivity for the light-controlled CCM strains in comparison to WT cells and effects of light-controlled growth-restriction for both CCM strains. For the Clb2ΔDB-psd3 strain the obtained results were less convincing and the analysis of the obtained data raised the question whether the desired cell cycle arrest phenotype was induced for Clb2ΔDB-psd3 cells during cultivation at all. 2.3. BIOANALYSIS FOR PRODUCTION STRAIN ESTABLISHMENT 49 The applied LC-method allowed a good separation of β-carotene from cell extracts, while maintaining well deĄned peak shapes. Nevertheless, for standard solutions β-carotene showed a limited solubility in 50 % ACN 50 % MeOH (≈ 5 mg/L) and powdery β-carotene precipitated under the cooled storage conditions of the autosampler for higher concen- trations. Cell extracts were found to contain β-carotene concentrations of 10 mg/L and higher, presumably kept in solution by coextracted substances. As visualized by Ągure 2.27, beside the β-carotene peak, a fair amount of other signals were visible at λ= 450 nm for cell extracts, presumably caused by β-carotene precursor lycopene or cis and trans isomers. [31] [74] Intensities of these signals seemed to vary in between batches of analyzed samples and were usually not as prominent as for the shown samples. 2.3.2 3ŠAMP and Cordycepin Production The potential anti-cancer drug cordycepin was one of the main target molecules to be produced in a metabolic engineering approach within the MELICOMO project. Cordy- cepin is produced by the fungus Cordyceps militaris from cellular 3ŠAMP, a precursor also available in limited amounts in S. cerevisiae due to the degradation of RNA. [107] During the MELICOMO project, Bastian Pook focused on the generation of S. cerevisiae strains providing an increased 3ŠAMP supply and on engineering strains for cordycepin production. Within the present work, sample preparations and mass spectrometric methods were established in order to enable 3ŠAMP and cordycepin quantiĄcation from S. cerevisiae cultures and a summary of the obtained results is presented below. 2.3.2.1 LC-MS QuantiĄcation of 3ŠAMP In C. militaris 3ŠAMP is generated from adenosine by the enzyme Cns3. Here, Cns3 cat- alyzes also the reaction of adenosine to pentostatin, a molecule regulating cellular levels of cordycepin by inhibiting adenosine deaminase activity. [107] Within the MELICOMO project, Bastian Pook generated plasmids for the expression of CNS3 controlled by the blue light-sensitive transcription factor LexACry2/Cib1-VP16, which allowed Cns3 pro- duction in S. cerevisiae under blue light. Furthermore, different shortened CNS3 variants were generated in order to eliminate pentostatin production of Cns3. Within the present work, 3ŠAMP was quantiĄed by LC-MS from strains expressing CNS3 in full length (strain yDS495 + pDS263), different shortened constructs of CNS3 (strains yDS495 + pBP02, yDS495 + pBP03 and yDS495 + pBP04) and the WT strain (yDS495). Culture media and cell extracts were provided by Bastian Pook. Here, 3ŠAMP was detected in corresponding samples of cell extracts but not in samples prepared from culture media. Figure 2.28 shows representative EICs of the separation of 3ŠAMP and 5ŠAMP and the linear regression of calibration standards for 3ŠAMP. Data plotted for the linear regression is summarized in table 5.28 in the appendix section. 2.3. BIOANALYSIS FOR PRODUCTION STRAIN ESTABLISHMENT 51 Table 2.5: Obtained 3ŠAMP concentrations determined for the cell extracts and calcu- lated for the corresponding culture (equivalent to half of the sample concen- tration). Standard deviations are denoted for samples measured in triplicate. light c 3ŠAMP [µm] c 3ŠAMP [mg/L] strain condition (sample) (culture) yDS495 + pDS263 D 25.7 ± 2.0 4.5 ± 0.3 L 27.1 ± 4.4 4.7 ± 0.8 yDS495 + pBP02 D 26.5 ± 2.3 4.6 ± 0.4 L 27.3 ± 1.2 4.7 ± 0.2 yDS495 + pBP03 D 18.1 ± 2.0 3.1 ± 0.3 L 12.1 ± 2.1 2.1 ± 0.4 yDS495 + pBP04 D 10.6 ± 2.0 1.8 ± 0.3 L 8.8 ± 1.8 1.5 ± 0.3 yDS495 (WT) D 2.8 ± 0.2 0.5 ± 0.0 L 4.2 ± 0.1 0.7 ± 0.0 As shown in table 2.5, resulting 3ŠAMP concentrations were very similar for samples generated from strain yDS495 + pDS263, which was expressing full length CNS3 and for strain yDS495 + pBP02 expressing a shortened construct. These two yeast strains exhibited the highest 3ŠAMP concentrations measured within the experiment. SigniĄ- cantly less 3ŠAMP was found in samples for the strain yDS495 + pBP03 and even smaller concentrations were determined for samples generated from strain yDS495 + pBP04. Both strains were expressing shortened variants of CNS3. For samples generated from WT cells lacking Cns3, the lowest 3ŠAMP concentrations were determined. As visualized in Ągure 2.29, the light condition did not signiĄcantly inĆuence 3ŠAMP production, as for most strains similar concentration ranges were obtained for cells grown in darkness and under blue light conditions. However, higher 3ŠAMP yields were indeed expected for the 3ŠAMP production strains cultured under blue light conditions, because the ex- pression of CNS3 was under the control of the blue light-sensitive transcription factor LexACry2/Cib1-VP16. Noteworthy, pentostatin, which is also known to be generated by Cns3, could not be detected for any strain within the analysis. Due to the usage of negative ion mode for the experiment a deĄnitive conclusion on the absence of pento- statin can not be drawn, because positive ion mode might would have been necessary to ionize pentostatin. Furthermore, pentostatin is not charged and might get excreted from the cell, which would explain the absence in cell extracts. [52] 2.3.2.2 LC-MS/MS QuantiĄcation of Cordycepin In C. militaris cordycepin is generated from 3ŠAMP via the deduced intermediate 2Š-carbonyl-3Š-deoxyadenosine by the enzymes Cns2 and Cns1. [107] The corresponding S. cerevisiae production strain yBP03 was generated by Bastian Pook and has CNS2 genomically integrated, while CNS1 is expressed from plasmid. Both genes are under 2.3. BIOANALYSIS FOR PRODUCTION STRAIN ESTABLISHMENT 57 Table 2.8: Comparison of cordycepin concentrations for culture media obtained by ex- ternal calibration and standard addition. biol. light c cordycepin [mg/L] strain repl. condition ext. cal. std. add. deviation yBP03 1 D 0.55 ± 0.00 0.50 ± 0.01 9.1 % 1 L 1.04 ± 0.03 0.92 ± 0.05 11.5 % yJT42 1 D 0.96 ± 0.01 0.85 ± 0.04 11.5 % 1 L 1.87 ± 0.10 1.86 ± 0.05 0.5 % yJT42 2 D 1.14 ± 0.04 1.06 ± 0.01 7.0 % 2 L 3.55 ± 0.04 3.65 ± 0.06 2.8 % WT 1 D NC NC NC 1 L NC NC NC As shown by table 2.8, obtained results for both quantiĄcation methods deviated from each other in the range of 0.5 to 11.5 %. In the comparison, no clear trend was observed in regard to higher or lower quantiĄcation results of a respective method. Standard addition as an internal calibration method compensates for matrix effects and changes of sensitivity towards the analyte during the analysis but requires more time as for each quantiĄcation 10 measurements were necessary (5 standard additions measured in dupli- cate). For quantiĄcation by external calibration, 3 measurements for each sample were sufficient (measurements of calibration standards not taken into account). Summarizing the comparison, both methods have proven to be suitable for quantiĄcation of cordycepin from 1:100 diluted culture media. 2.3.3 Gibberellin A4 Production Gibberellin A4 (GA4) was one of the main target molecules within the MELICOMO project and the S. cerevisiae GA4 production strain candidate yJS13 was constructed by Dr. Johannes Scheffer during his doctorate. [92] He integrated the plant genes AtCPS, CmKS, AtKO, PsKAO2, CmGA20ox, PsGA3ox and AtAtr2 into the yeast genome in or- der to generate GA4 from endogenous GGPP. Furthermore, the LexA-EDLL transcription factor was genomically integrated to enable constitutive expression of the recombinant genes. Unfortunately, Ćow cytometric analyses using a recombinant GA-reporter did not result in robust evidence for cellular GA4 production and GA4 was not detected in mass spectrometric analyses of cellular extracts either. As a possible explanation for this Ąnding Dr. Johannes Scheffer observed that CmKS was not clearly identiĄed by western blot analyses, when the construct was individually expressed from plasmid. Moreover, investigations into the availability of each individual protein of the recombinant pathway for yJS13 by western blotting were troublesome, due to similar molecular weights of the pathway proteins and the usage of YFP as a fusion tag for all constructs. In order to investigate into this issue a proteomics experiment was performed for the yJS13 strain within this work. For the analysis, whole proteome digests were generated for biological 58 CHAPTER 2. RESULTS triplicates of the yJS13 and the WT strain. Samples were measured using the timsTOF Pro instrument and obtained data was analyzed via the PEAKS workĆow. Table 2.9 summarizes obtained sequence coverage for proteins of the recombinant pathway. As each of the proteins was modiĄed with the YFP-tag, untagged sequences were used for the database search in order to avoid cross attributions. Table 2.9: Obtained sequence coverage for proteins of the recombinant pathway for biological triplicates of the GA4 production strain yJS13 and the WT strain. sequence coverage [%] WT yJS13 replicates 1 2 3 1 2 3 AtCPS 0 0 0 50 50 50 CmKS 0 0 0 0 0 0 AtKO 5 5 0 56 53 52 PsKAO2 0 0 0 40 40 38 CmGA20ox 0 0 0 37 41 39 AtAtr2 0 0 0 43 27 33 LexA-EDLL 0 0 0 38 33 33 As visualized by table 2.9, each protein of the recombinant pathway apart from CmKS exhibited a good sequence coverage which was well distributed over the full protein se- quence (data not shown). In regard to the WT no to very little coverage was obtained for the target sequences. This experimental result veriĄes that the yJS13 strain did not express the gene of the ent-kaurene synthase, thereby disrupting the pathway towards GA4, which explains the lack of GA4 production observed in previous experiments. Subsequently, Dr. Christof Taxis transformed the yJS13 strain with plasmids carrying genes of ent-kaurene synthases originating from plants (Oryza sativa and Arabidopsis thaliana) and fungi (Fusarium fujikuroi and Fusarium mangiferae). In Ćow cytomet- ric analyses using a recombinant GA-reporter, cells transformed with plasmids for the production of either OsKS or AtKS, showed signiĄcantly increased Ćuorescence (com- munication with Dr. Christof Taxis, data not shown). In order to quantify the amount of GA4 produced by the newly generated GA4 production strains, extracts of the culture media and cells were analyzed by mass spectrometry. 60 CHAPTER 2. RESULTS Table 2.10: QuantiĄcation of GA4 by LC-MS from media extracts. GA4 concentrations are shown for the measured samples (100-fold concentrated) and for the extracted media. Standard deviations are denoted for samples measured in triplicate. Strains labeled with * were used as negative controls and carried the plasmids pRS313, pRS314 and pRS316. supplemented cGA4 (sample) cGA4 (medium) strain ent-kaurene synthase [nm] [ng/L] yGA01 OsKS 83 ± 9 274 ± 31 yGA02 AtKS 82 ± 18 271 ± 59 yGA03 AtKS 51 ± 24 171 ± 80 yGA04 AtKS 53 ± 15 175 ± 49 yJS13 * Ű ND NC yJT23 * Ű ND NC As shown in table 2.10 standard deviations of GA4 concentrations for the measured samples were very high. This behavior was not observed for the measured calibration standards. In order to generate samples containing sufficient GA4 for quantiĄcation, sam- ples had to be analyzed 100-fold concentrated compared to the original extract volume. Hence, extracted contaminants were also highly concentrated in the sample, which might cause ion suppression during the LC-MS measurement and leads to an overall problem- atic behavior of the sample during the measurement. Resulting GA4 concentrations of samples for strains yGA01, yGA02, yGA03 and yGA04, producing either OsKS or AtKS, were found to be similar. Corresponding GA4 con- centrations calculated for the growth media were in the range of a few hundred ng/L. Similar GA4 concentrations were obtained in subsequent LC-MS experiments, which were performed to repeat the LC-MS analysis of growth media extracts for the same GA4 production strains (data not shown). 2.3.3.2 Summary Ű Bioanalytics for Production Strain Establishment Within this work various HPLC and LC-MS methods were established for the quantiĄ- cation of target molecules and a selection of exemplary analyses are presented above. Here, samples were generated from S. cerevisiae cells or culture media and analytes were quantiĄed by external calibration or standard addition. The applied methods allowed quantiĄcation of target molecules in the range of ng/L to mg/L and the obtained re- sults helped to select promising production strains for subsequent optimization cycles. In summary, instrumental analytics played an important role for the development and es- tablishment of S. cerevisiae strains for the production of valuable secondary metabolites within the MELICOMO project. 61 Chapter 3 Discussion Within the present work, advanced bioanalytical tools were applied for the character- ization of light-controlled S. cerevisiae and for the quantiĄcation of target molecules in the context of metabolic engineering. Section 2.1 summarized results for biomass composition determined for light-controlled CCM yeast strains and the effect of growth restriction on the proteome of S. cerevisiae was further analyzed by the proteomics ex- periment evaluated in section 2.2. Subsequently, a representative selection of HPLC and LC-MS experiments was shown by section 2.3, which were established within this work in order to quantify β-carotene, 3ŠAMP, cordycepin and GA4 from cultures of different S. cerevisiae production strains. In the following chapter, the results obtained in the present work are discussed in the light of Ąndings published in literature. 3.1 Biomass Analysis of Light-Controlled S. cerevisiae The biomass composition determined for growth-restricted Clb2ΔDB-psd3 cells was signif- icantly different from the biomass composition of the WT strain and the obtained biomass compositions were used as data basis for a Ćux balance analysis (FBA) performed by the group of Prof. Dr. Zoran Nikoloski. Most publications discussing FBA parameters, highlight the importance of precisely adjusted biomass compositions to the respective experimental conditions, [27] [78] [79] and only few sources claim that biomass composition did not strongly impact Ćuxes through the central carbon metabolic network for their calculations. [111] Nevertheless, due to the laborious and error prone process of biomass analysis, research groups often rely on published data for their Ćux balance modeling. [111] Corresponding data on biomass composition for the model organism S. cerevisiae is read- ily available but was often generated for strain backgrounds, no longer commonly used in present laboratories. Furthermore, changes introduced into the genome, e.g. within metabolic engineering approaches, may lead to the formation of distinct phenotypes, which might not be represented by literature biomass data. Accordingly, implications 62 CHAPTER 3. DISCUSSION drawn from FBAs with poorly adjusted biomass composition may lead to misinterpreta- tions in regard to Ćuxes through the metabolic network. Hence, for the light-controlled CCM strains used in the MELICOMO project, analyses of biomass composition prior to FBA were necessary based on the characteristic phenotypes described in the literature. Growth-restricted Clb2ΔDB-psd3 and Cdc48-psd3 cells appear mostly large budded [89] [43] and altered PKA activity due to the substitution of Cyr1 by bPAC is expected to have a broad impact on cellular processes e.g. on glycogen storage. [70] Furthermore, biomass compounds could be accumulating during cell cycle restriction, with the before men- tioned consequences on the FBA calculations. In order to compare the obtained biomass data to literature, table 3.1 summarizes the determined biomass composition for the Clb2ΔDB-psd3 strain and data published by Lange et al. Table 3.1: Comparison of biomass composition for the Clb2ΔDB-psd3 strain (S288C derivative) cultured under restrictive growth condition with literature data. Denoted ranges of literature data originate from varied dilution rates and glucose feeds applied during aerobic fermentation of the S. cerevisiae strain CEN.PK113-7D. this work literature WT Clb2ΔDB-psd3 Lange et al. [64] protein 23.6 ± 1.2 % 31.7 ± 1.7 % 32.3 - 45.7 % carbohydrates 43.7 ± 3.6 % 34.1 ± 2.3 % 31.6 - 45.4 % RNA 3.6 ± 0.4 % 5.9 ± 0.6 % 4.3 - 7.9 % DNA 0.13 ± 0.01 % 0.11 ± 0.02 % 0.4 - 0.5 % lipids 6.4 ± 0.3 % 7.6 ± 0.4 % 7.2 - 10.2 % Analyzed Clb2ΔDB-psd3 cells were found to contain more protein and RNA and less carbo- hydrates in comparison to the WT strain. Lange et al. observed a simultaneous increase in protein and RNA content with a decrease in carbohydrate content when the dilution rate was increased. [64] Hence, biomass composition of growth restricted Clb2ΔDB-psd3 cells reĆected culture conditions with high nutrient availability, while biomass composi- tion of WT cells was similar to culture conditions with limited nutrient supply. For future experiments biomass analysis could be scaled down, focusing on the determination of protein, RNA and carbohydrate content, which were found to be signiĄcantly different in the two analyses. Excluding lipid quantiĄcation from the analysis would reduce the preparative expense signiĄcantly and allow to perform the experiment in 200 mL culture scale. Here, it would be interesting to include the analysis of Clb2ΔDB-psd3 cells cul- tivated under permissive condition in order to distinguish between changes induced by strain-dependent differences and growth restriction. In addition, remaining glucose in the medium should be monitored to ensure that nutrient availability is similar for both strains under sampling conditions, as differences in nutrient availability would affect biomass composition. [64] 3.2. PROTEOMICS OF LIGHT-CONTROLLED S. CEREVISIAE 63 Data on biomass composition published by Lange et al. was chosen as a reference due to the representative range compared to other biomass compositions available in litera- ture. [37] [79] [82] Remarkably, the percentages of the individual biomass compounds deter- mined within the present work are all below or on the low end of the range of the reference data. Accordingly, for both analyzed strains, the sum of the determined biomass com- pounds did not sum up to 100 % and even if an additional 8.2 % of the dry mass was attributed to residual water, metals, inorganic phosphorus and sulfate as described by the literature, [64] the remaining undetermined biomass fraction was still > 10% for both examined strains. An explanation for this Ąnding might be that within the present work analyte recoveries were only determined for RNA and DNA. Lange et al. reports recovery factors of 0.82 for protein and 0.95 for carbohydrates using highly similar biomass quan- tiĄcation methods. An application of these recovery factors on the results summarized by table 3.1 would reduce the undetermined biomass fraction to signiĄcantly less than 10 %. Furthermore, lyophilization for 36 h and drying for an additional 24 h at 50 °C might not have entirely removed the intracellular water from the cells and it is possible that a higher share should be attributed to remaining water. Lange et al. freeze-dried their cell pellets for 48 h and even dried the cells further at 70 °C for 48 h, which might have removed considerably more water from the cells. However, conclusions drawn from the direct comparison of both strains is likely to be unaffected, because Clb2ΔDB-psd3 and the corresponding WT cells were treated identically and remaining undetermined fractions are highly similar in magnitude. Nevertheless, effects on FBA due to undeter- mined biomass can not be excluded for reasons discussed above. 3.2 Proteomics of Light-Controlled S. cerevisiae The analysis of proteomics data for the Cdc48-psd3 and bPAC yeast strains revealed deep insights into strain-dependent differences between the growth-restricted production state of CCM cells and the WT strain, as well as into differences between the restrictive and permissive light condition for both yeast strains. Unlike for the two before mentioned strains, the impact of blue light on the proteome for Clb2ΔDB-psd3 cells was negligi- ble and in the comparison to WT cells only minor differences were observed for the Clb2ΔDB-psd3 strain. These observations suggest, that within sample generation the de- sired cell cycle arrest was not sufficiently induced for Clb2ΔDB-psd3 cells and the following discussion will focus on the analysis of proteomics results obtained for the Cdc48-psd3 and bPAC CCM strains. In the following analysis, results of the comparison between restricted CCM and the WT strain are interpreted as a combination of strain-dependent differences and differences induced by the light condition. Here, strain-dependent refers to differences already present in the comparison between the permissive growth condi- tion of the CCM and the WT due to genetic modiĄcation. Whereas conclusions drawn 64 CHAPTER 3. DISCUSSION from the comparison of the CCM strain grown under restrictive and permissive growth condition point speciĄcally at effects induced by the light condition. Compared to WT cells, growth restriction results in an altered metabolic state for the Cdc48-psd3 and the bPAC strain As visualized by the heatmaps provided in section 2.2.1.1, the protein abundance proĄle of Cdc48-psd3 cells grown under permissive conditions was more WT-like in comparison to cell cycle arrested Cdc48-psd3 cells. On the other hand, growth restriction induced by cAMP depletion had a very broad impact on the proteome as visualized by the heatmaps for the bPAC strain in section 2.2.1.3. Here, protein abundances were found to be highly different between both light conditions for the bPAC strain and in comparison to the corresponding WT strains. This observation highlights that the proteome of bPAC cells cultured under permissive light conditions and the corresponding WT strain differ strongly due to the artiĄcial cAMP level of bPAC cells, which is determined by blue light intensity and thus cannot be adapted to nutrient availability. Consequently, in contrast to the Cdc48-psd3 strain, the permissive growth condition for bPAC does not translate into a WT-like state of the cells. It is therefore not surprising that DAPs identiĄed in both comparisons between restrictive and permissive growth conditions for the Cdc48-psd3 and bPAC strains overlapped only slightly, as shown by the Venn diagram in Ągure 2.20 B. In total, only 13 DAPs were shared by both analyses, representing 6.5 % for the Cdc48-psd3 and 2.5 % for the bPAC strain. Nevertheless, the GO-term analysis for restrictive and permissive growth condition pointed out that both strains had an altered redox activity in common and most of the assigned DAPs were higher abundant for the growth-restricted condition compared to the permissive growth condition. This might imply a favorable metabolic state for the production of valuable molecules via recombinant pathways. On the other hand, the overlap between DAPs identiĄed in both comparisons for the growth-restricted CCM and the WT strain was signiĄcantly higher, as shown in Figure 2.20 A. Here, 117 DAPs were shared between the two comparisons, representing 23 % of the total number of DAPs identiĄed for each of the two comparisons. Accordingly, results of the corresponding GO enrichment analyses overlapped, indicating altered catabolic and anabolic capacity (GO:0016616, Ťoxidoreductase activity, acting on the CH-OH group of donors, NAD+ or NADP+ as acceptorŤ) and an overall increased stress response (GO:0035770, Ťribonucleoprotein granuleŤ) for both growth-restricted CCM strains compared to their respective WT strains. Here, in particular, the observed effects on metabolism could have a positive impact on the production of molecules via recombinant pathways for growth-restricted CCM strains. 3.2. PROTEOMICS OF LIGHT-CONTROLLED S. CEREVISIAE 65 Increased β-carotene production in growth-restricted Cdc48-psd3 and bPAC cells could be explained by higher abundance of acetyl-CoA synthetase isoforms In comparison to the WT strain and normalized to cell dry weight Dr. Christof Taxis observed a 18-fold increased β-carotene production for the growth-restricted Cdc48-psd3 strain and a 5-fold increased β-carotene production compared with the permissive con- dition (unpublished manuscript). For the bPAC strain, Dr. Jonathan Trauth reported a 2-fold increased β-carotene production for growth-restricted bPAC cells in comparison to the WT strain normalized to dry mass and to cell culture volume. [100] Within the present work, proteomics data was analyzed for both CCM strains in order to provide possible explanations for the elevated β-carotene yields. In this section, the discussion of the data will focus on proteins involved in the biosynthesis of ergosterol (heatmap in Ągure 3.1 below), since the recombinant synthesis pathway for β-carotene branches off the ergos- terol biosynthesis superpathway from the isoprenoid FPP, and more general insights will be discussed later in a broader context. Conclusions drawn from the analysis should be applicable to the recombinant production of GA4, since FPP is a common precursor. Acetyl-CoA is the initial precursor of the ergosterol biosynthetic pathway and is generated by the acetyl-CoA synthetase isoforms Acs1 and Acs2. Remarkably, for both analyzed CCM yeast strains Acs1 or respectively Acs2 was found to be high abundant for growth- restricted CCM cells in comparison to the WT strain and to the permissive growth condition. In the Cdc48-psd3 strain, Acs2 was twice as abundant compared to the WT and was 10 % higher abundant in comparison to the permissive growth condition. For the bPAC strain Acs1 abundance was even 3.5-fold increased for the restrictive growth condition in comparison to the permissive growth condition and an impressive 6-fold in- creased abundance was observed when compared to the WT strain. Literature reports that overexpression of ACS1 increases the production of the sesquiterpenoid amorpha- diene in a corresponding S. cerevisiae production strain, which was also overproducing key enzymes of the mevalonate pathway. [95] Therefore, the observed high abundance of acetyl-CoA synthetase isoforms is likely to increase Ćux through the ergosterol biosyn- thetic pathway for growth-restricted Cdc48-psd3 and bPAC cells, with beneĄcial effects on the recombinant production of isoprenoid-derived compounds. For the growth-restricted Cdc48-psd3 strain Hmg1 was found to be higher abundant in comparison to the WT strain. The HMG-CoA reductase isoform Hmg1 is predominantly responsible for catalytic activity in the mevalonate pathway under aerobic conditions and marks a rate-limiting step in ergosterol biosynthesis. [12] In numerous metabolic engineering approaches, a truncated construct of HMG1 is overexpressed in order to increase the availability of FPP for sesquiterpene production. [53] [6] However, overex- pression of full length Hmg1 was demonstrated to have a negative effect on ergosterol production and was found to be responsible for the formation of nucleus-associated arrays of stacked endoplasmic reticulum (ER) membranes. [69] Nevertheless, the mod- 66 CHAPTER 3. DISCUSSION erately increased abundance observed for Hmg1 may lead to higher Ćuxes through the mevalonate pathway for growth restricted Cdc48-psd3 cells in comparison to the WT strain. However, this observation does not explain the higher β-carotene production of the growth-restricted Cdc48-psd3 compared to the permissive growth condition, since similar Hmg1 abundances were observed for both conditions. In addition many pro- teins involved in ergosterol biosynthesis downstream FPP were among the DAPs for the Cdc48-psd3 strain. For these proteins, strong abundance differences were observed in the comparison between Cdc48-psd3 and the WT strain. In the