From transcriptome analysis to protein secretion: strategies to improve industrial protein production in Bacillus subtilis

Bacillus subtilis is widely used as a host for industrial enzyme production due to its safety profile and capacity to secrete proteins at gram-per-liter concentrations. Nevertheless, many protein products fail to reach such yields. The aim of this thesis was to advance our understanding of protein secretion pathways and bottlenecks in B. subtilis, to improve industrial protein production. Chapter 1 provides an overview of the protein secretion pathways in B. subtilis and outlines strategies to improve protein production, including targeting host factors and optimizing the expression construct itself. Additionally, transcriptome-based approaches, including RNA-seq and Ribo-seq, offer unbiased, genome-wide methods to identify production bottlenecks. These techniques were applied to investigate the production of XynA, AmyM, PrgA and GFPnb. Chapter 2 describes the development of GINtool, a data analysis tool designed to analyse regulon activity within transcriptome datasets. GINtool accounts for regulators that exert both activating and repressing effects on their target genes, it can be applied to analyze gene expression data based on functional categories or operon structures, and it can be readily extended to other types of omics data, such as proteomics. In Chapter 3, we applied RNA sequencing to analyze the production of XynA, PrgA, and GFPnb under constitutive and xylose-inducible expression systems. Induced expression resulted in a substantially larger number of differentially expressed genes than constitutive expression. Notably, transcriptome profiles were more strongly influenced by the expression system than by the identity of the produced protein. This highlights the sensitivity of RNA-seq analyses and emphasizes the importance of using experimental conditions that closely mimic industrial production processes to derive actionable insights for strain improvement. Despite these challenges, we achieved a 2.7-fold increase in PrgA production by comparing our transcriptome data with a previous study and deleting the transcriptional regulator CtsR. In Chapter 4, we investigated the underlying secretion bottlenecks of XynA and AmyM using RNA-seq and evaluated a range of mutant strains aimed at improving the production. Transcriptome analysis and Ribo-seq indicated that neither transcription nor translation accounts for the decline in XynA secretion at later time points. Instead, deletion of the intracellular protease LonA resulted in a 2.4-fold increase in XynA production. These findings suggest that targeting intracellular proteases represents a promising strategy to enhance protein production. In Chapter 5, we describe the development of a split-luciferase-based assay to monitor protein secretion levels in B. subtilis. A small peptide derived from the NanoLuc luciferase is fused to the secreted protein of interest, and secretion is quantified through complementation with the remaining luciferase component. We establish a medium-throughput protocol that enables rapid monitoring of the secretion of XynA, AmyM, PrgA, and GFPnb over time. In Chapter 6, we investigated how AmyM secretion is affected by replacing its signal peptide with YxkCss or SacBss, which range in hydrophobicity. We found that the YxkCss resulted in lower secretion levels and increased protein secretion stress compared to SacBss. To elucidate the basis of these differences, we performed RNA-seq and examined AmyM production in Ffh-depleted cells. However, these approaches did not yield clear mechanistic insights. This underscores the need for a more detailed understanding of protein secretion pathways in Gram-positive bacteria such as B. subtilis. Chapter 7 offers directions for future research: performing transcriptome analyses under conditions that more closely resemble industrial production processes and applying Tn-seq on large mutant libraries to identify genes that are beneficial or detrimental to the production of industrially relevant proteins. Taken together, this thesis provides an overview of how a combination of targeted and unbiased methodologies can be used to identify protein secretion bottlenecks and to advance our understanding of protein secretion pathways in B. subtilis.