Bursts in biosynthetic gene cluster transcription are accompanied by surges of natural compound production in the myxobacterium Sorangium sp

Abstract

A better understanding of the genetic regulation of the biosynthesis of microbial compounds could accelerate the discovery of new biologically active molecules and facilitate their production. To this end, we have investigated the time course of genome-wide transcription in the myxobacterium Sorangium sp. So ce836 in relation to its production of natural compounds. Time-resolved RNA sequencing revealed that core biosynthesis genes from 48 biosynthetic gene clusters (BGCs; 92% of all BGCs encoded in the genome) were actively transcribed at specific time points in a batch culture. The majority (80%) of polyketide synthase and non-ribosomal peptide synthetase genes displayed distinct peaks of transcription during exponential bacterial growth. Strikingly, these bursts in BGC transcriptional activity were associated with surges in the net production rates of known natural compounds, indicating that their biosynthesis was critically regulated at the transcriptional level. In contrast, BGC read counts from single time points had limited predictive value about biosynthetic activity, since transcription levels varied >100-fold among BGCs with detected natural products. Taken together, our time-course data provide unique insights into the dynamics of natural compound biosynthesis and its regulation in a wild-type myxobacterium, challenging the commonly cited notion of preferential BGC expression under nutrient-limited conditions. The close association observed between BGC transcription and compound production warrants additional efforts to develop genetic engineering tools for boosting compound yields from myxobacterial producer strains.

FIGURE 1

Growth curve of So ce836. The blue curve shows the growth of So ce836 over six time points (logistic curve $y=L/\left(1+{\mathrm{e}}^{-k\left(x-x0\right)}\right)+b$, with $L=5.434,k=0.347,x0=4.225,b=26.083$). The exponential phase (days 1–8) is indicated by the pink area and black dashed line. Each data point represents mean cell counts per mL on a logarithmic scale (five biological replicates per growth time with 10 technical replicates each). Error bars indicate the relative standard error of the mean.

FIGURE 2

Temporal transcription patterns of 8311 differentially expressed genes in the So ce836 genome. The deviation from mean transcription (z‐scores) is plotted over growth time. A deviation value of 0 indicates that a gene is transcribed at its mean level calculated over all time points. A value of 2 indicates that transcription levels deviate from the mean by two standard deviations. The genes were grouped using k‐means clustering with a group size of five. The mean transcription of individual genes over time‐point replicates is plotted in grey and median values for the whole transcription group are plotted in blue with error bars indicating first and third quartiles. The pink‐shaded area indicates the exponential growth phase.

FIGURE 3

Temporal transcription of core biosynthetic genes of NRP‐ and PK‐related BGCs. The deviation from mean transcription (z‐score) is plotted over growth time for each BGC associated with the production of NRPs, PKs, or their hybrids. Numbers in brackets indicate the number of differentially expressed core genes and the total number of core genes in the BGC, respectively. Each curve represents the time‐course of transcription of one core biosynthetic gene. Data points indicate the means over replicates at a time point and error bars represent the adjusted standard deviations. Differentially transcribed genes are represented in dark red, genes without a significant difference in transcription over the time‐course experiment in grey. The last panel shows the temporal transcription of 44 genes encoding ribosomal proteins for comparison. Pink shading indicates the exponential growth phase.

FIGURE 4

Net production rates of icumazole A (A) and epothilone A (B), and transcription of core genes in the underlying BGCs. The images above the plots show the chemical structures of icumazole A (left, PubChem CID 101557675; https://pubchem.ncbi.nlm.nih.gov/compound/Icumazole‐A) and epothilone A (PubChem CID 448799; https://pubchem.ncbi.nlm.nih.gov/compound/Epothilone‐A). The upper panels (blue) represent the relative concentration of natural products as obtained from LC/MS analyses (HPLC based UV light at 304 nm, 15.27 min retention time for icumazole A; LC/MS mass peaks at 494.2571 m/z, 9.05 min retention time for epothilone A). Boxplots over five replicates per time point are shown. The curves were generated by monotone piecewise cubic interpolation over median concentrations. The second panels (green) show the net compound production rate per myxobacterial cell, based on the slope of the compound concentration per mL (blue curve) at a certain time point normalized by the mean cell count per mL at that time point. Error bars indicate propagated errors. The lower panels (red) represent the deviation from mean transcription (z‐scores) of core biosynthetic genes in the icumazole and epothilone BGCs, respectively (see also regions 19 and 25_3 in Figure 3). The zero‐line marks the mean transcription of a gene over all time points. Pink shading indicates the exponential growth phase.

FIGURE 5

Temporal distribution of maximal gene transcription of BGC core genes associated with NRP‐ and PK‐production. Out of 81 core genes from NRP‐ and PK‐related BGCs, 75 were differentially expressed in our experiment. This figure shows for each time point how many of the differentially expressed core genes were maximally transcribed at that time. Pink shading indicates the exponential growth phase.

FIGURE 6

Distribution of BGC read counts on their day of maximal transcription. For each BGC core gene the mean read counts per kilobase after DESeq2 normalization was calculated over all replicates on the day of maximal BGC transcription. Plotted here is the median over all core genes in a BGC. Each type of BGC is represented in a different colour. The ‘other BGC types’ include resorcinol‐, phenazine‐, and phosphonate‐producing BGCs. Please note that the first two bins and the last one do not have the same range as the other bins. With maximally three reads per kilobase, the epothilone BGC is located in the left most bin. In contrast, the icumazole BGC had maximally 428 reads per kilobase.