Fig 1.
Schematic representation of the sequencing and data analysis methodology.
(A) A description of the experimental pipeline. RNA-Seq (left column) and term-seq (middle column) libraries were prepared according to protocols described in [34,35]. 5’ end-Seq libraries (right column) were generated by dividing the total RNA into 5’ polyphosphate treated (processed) and untreated (unprocessed) samples and subsequently processed according to protocols described in [22,64]. White and grey lines correspond to RNA and cDNA, respectively and colored blocks represent unique sequence barcodes. Illumina sequencing adaptors with index barcodes are also indicated. (B) A brief description of the analysis pipeline. Raw reads from all three sequencing methodologies obtained from three environmental conditions (early-log and mid-log growth phase in the presence and absence of vancomycin) were de-multiplexed and aligned to T4 (NC_003028.3). Based on the reads mapped, single nucleotide coverage was calculated. Coverage of the 5’ end of the reads calculated for the 5’ end-Seq was used to determine the transcription start sites. Coverage of the 5’ end of the reads from term-seq was used to determine the transcription termination sites. RNA-Seq coverage was used to calculate the read-through across candidate 5’ untranslated regions with early transcription terminators. Read-through responses of the candidates in different media conditions (rich and poor media, media with and without uracil) were analyzed to identify environment-responsive RNA regulators.
Fig 2.
Genome-wide map of the Streptococcus pneumoniae TIGR4 transcriptional landscape.
A map of all the transcriptional features identified. 1. Paired-end RNA-Seq coverage of the plus strand. 2. Paired-end RNA-Seq coverage of the minus strand. 3. Annotated operon structures on the plus strand. Tile colors represent classification of the operons according to their number of high confidence TSS and TTSs: 1) traditional operons consisting of multiple genes with a single TSS and a TTS (green); 2) multiTSS operons consisting of multiple genes with internal TSSs but one TTS (blue); 3) multiTTS operons consisting of multiple genes with a single TSS but multiple internal TTSs (red); and 4) complex operons consisting of multiple genes with multiple internal TSSs and TTSs (orange). To avoid clutter, simple operons consisting of a single gene with a single TTS and TSS are omitted. 4. Annotated operon structures on the minus strand. 5. 141 novel putative intergenic regulatory elements in blue. 6. 742 high confidence transcriptional start sites (TSS) in green. 7. 1864 enriched transcriptional termination sites (TTS) in red. 8. Log transformed processed/unprocessed ratio of each high confidence TSS. 9. Log transformed coverage of each predicted TTS. 10. Positions of the known structured non-coding RNAs and small RNAs. Figure was generated using Circos [79].
Fig 3.
Distribution of operon types in the genome and frequency of transcriptional features within non-traditional operons identified using StringTie.
(A) The pie chart describes the distribution of the types of operons present in T4 annotated using only high confidence TSSs. A total of 388 multigene and 343 single gene operons were identified, which can be divided up in 47% simple operons (single gene transcriptional units with a single TSS and TTS; gray), 26% complex operons (multi-gene operons with multiple TSSs and TTSs; orange), 10% traditional operons (multi-gene operon with a single TSS and TTS; green), 1% multiTSS operons (blue), and 16% multiTTS operons (red). (B) The clustered histogram describes the distribution of genes and transcriptional features in non-traditional operons, where gray represents the numbers of genes in the multigene operons, green represents the number of TSSs within operons, red represents the number of TTSs within operons. Two-gene operons are found most frequently in the non-traditional operons with one internal TSS and TTS.
Fig 4.
Variability in expression levels between genes in the same operon when grown in rich (SDMM) and poor (MCDM) media conditions.
RNA-Seq coverage maps of complex operon/regulon including the TSSs in green and TTSs in red. Size of the transcriptional features represents log transformation of the processed/unprocessed ratio for TSSs and coverage for TTSs. Arrows on the TSS represent the strand and direction of transcription. There are little to no reads mapped to the strand opposite to coding regions. Coverage was normalized to the number of uniquely mapped reads in each library and log10 transformed for representation (A) A complex 9-gene operon (SP_1018–SP_1026) encoding thymidine kinase, GNAT family N-acetyltransferase, peptide chain release factor 1, peptide chain release factor N(5)-glutamine methyltransferase, threonylcarbomyl-AMP synthase, N-acetyltransferase, serine hydroxymethyltransferase, nucleoid-associated protein, and Pneumococcal vaccine antigen A respectively. Genes SP_1022-SP_1026 are expressed to greater levels in MCDM than in SDMM unlike genes SP_1018-SP_1021. (B) The maltose regulon, an example of two operons of different complexities working together in response to maltose in the medium. Complex operon SP_2108-SP_2112 shows greater expression in MCDM than SDMM in comparison to the multiTTS operon SP_2106-SP_2017.
Fig 5.
Validation of the regulatory activities of FMN, TPP riboswitches and a putative 5’-UTR regulatory candidate in different nutrient conditions.
The relative expression and average RNA-Seq coverage of SP_0178 (FMN) (A) and SP_0716 (TPP) (C) increases in poor (MCDM) medium compared to rich (SDMM) medium, potentially compensating for the depletion of the specific ligand. Error bars represent standard error of the mean. Expression of SP_0178 (FMN) (B) and SP_0716 (TPP) (D) is reduced when the poor medium is supplemented with respective ligands thus confirming the regulatory activities of FMN and TPP riboswitches. (FMN- Riboflavin; TPP- Thiamine). Expression is relative to the qPCR internal control and thus for instance an expression level of 0.5 means that the expression of tested gene is 50% of the expression of internal control. Error bars represent standard error of the mean across three technical replicates. SP_1782 encoding ribosomal protein L11 methyltransferase (E) decreases in poor media (MCDM) compared to rich media (SDMM).
Fig 6.
Mechanism, structure and regulatory activity of pyrR regulatory RNA element and its mutants in the presence and absence of uracil.
(A) Schematic representation of the proposed mechanism of regulation of pyr operon by pyrR RNA element. In the presence of UMP, PyrR binds to the pyrR RNA and results in the formation of a premature terminator, disrupting the anti-terminator formed when UMP is low, resulting in transcription termination. (B) RNA-Seq coverage map across the pyr operon (SP_1278–1276) showing premature transcription termination and consequently decreased expression of its genes downstream of the pyrR regulator when grown in defined medium (CDM) in the presence of uracil (yellow) compared to the absence of uracil (blue). Coverage was normalized to the number of uniquely mapped reads in each library. TSSs are in green and the size represents the log transformation of the processed/unprocessed ratio and TTSs are in red and size represents the log transformation of the coverage. (C) qRT-PCR determining the expression of the first genes in the pyr operon (SP_1278) in the presence and absence of uracil validates the RNA-Seq observation. Expression is relative to the qPCR internal control. (D) Secondary structure of the S. pneumoniae pyrR RNA regulatory element in ‘off’ conformation. Boxed in red and yellow are bases that were deleted in M1 and M3 mutations, respectively. Bases boxed in green were replaced by indicated bases to make mutation M2. Highlighted in grey are nucleotides that would base pair to form the anti-terminator when the riboswitch is in the “on” conformation. (E) A representative qRT-PCR quantification of the expression of SP_1278 transcript from pyrR RNA mutant strains cultured in defined medium with or without uracil, corresponding to the regulatory activity of pyrR RNA mutants. While the WTc (wild type with chloramphenicol resistance cassette) decreases expression in the presence of uracil, M1 is insensitive to the ligand. M2 and M3 result in either constitutive or reduced expression of the pyr operon. ~2 fold higher expression in M1 compared to WTc in the absence of uracil could be the result of endogenous uracil having a slight inhibiting effect on the wild type. Expression is relative to the qPCR internal control. Error bars represent standard error of the mean across three technical replicates.
Fig 7.
Regulation by pyrR regulatory element is important, but not essential for in vitro growth of S. pneumoniae.
(A-B) In vitro growth curves of mutants when cultured in defined media with (20 μg/ml) or without uracil. While mutant M2 (green) does not display a growth defect, mutants M1 (orange) and M3 (maroon) have growth defects that are restored in the presence of uracil indicating that a functional pyrR RNA element is important, but not absolutely essential for in vitro growth of S. pneumoniae. WT (blue) does not have a growth defect in the tested conditions. (C-D) Representative in vitro growth curves of mutants cultured in media with (15 μg/ml) or without 5-FOA, a toxic uracil analog. All strains show varying degrees of defects in the tested conditions indicating that a drug targeted against the secondary structure can severely and specifically hamper growth. Mutant M3 was not included in this assay as it cannot be grown without uracil. Error bars represent standard error of the mean across three biological replicates.
Fig 8.
Regulation by the pyrR RNA element is crucial for in vivo survival and virulence of S. pneumoniae.
1x1 competition assays (mutant vs wild type T4) reveals fitness defects of pyrR RNA mutants in a mouse infection model. While M1 (A) and M3 (C) display severe defects in all the tested in vivo environments namely lung, blood and nasopharynx, M2 (B) has less of a defect. Significant change in fitness (p<0.0125) are indicated by asterisks (*). Each data point represents a single mouse.