The small RNA (sRNA) Pxr negatively controls multicellular fruiting body formation in the bacterium Myxococcus xanthus, inhibiting the transition from growth to development when nutrients are abundant. Like many other prokaryotic sRNAs, Pxr is predicted to fold into three stem loops (SL1-SL3). SL1 and SL2 are highly conserved across the myxobacteria, whereas SL3 is much more variable. SL1 is necessary for the regulatory function of Pxr but the importance of SL3 in this regard is unknown. To test for cis genetic elements required for Pxr function, we deleted the entire pxr gene from a developmentally defective strain that fails to remove Pxr-mediated blockage of development and reintroduced variably truncated fragments of the pxr region to test for their ability to block development. These truncations demonstrated that SL3 is necessary for Pxr function in the defective strain. We further show that a highly conserved eight-base-pair segment of SL3 is not only necessary for Pxr to block development in the defective strain under starvation conditions, but is also required for Pxr to prevent fruiting body development by a developmentally proficient wild-type strain under high-nutrient conditions. This conserved segment of SL3 is also necessary for detectable levels of Pxr to accumulate, suggesting that this segment either stabilizes Pxr against premature degradation during vegetative growth or positively regulates its transcription.
Small RNAs (sRNA) play major roles in bacterial gene regulation1,2,3, including in the Gram-negative myxobacteria, which cooperatively construct multicellular fruiting bodies in response to starvation4,5. By base-pairing with mRNAs, sRNAs can modulate protein translation and mRNA degradation. The sRNA Pxr in the model myxobacterium Myxococcus xanthus was previously found to regulate the transition from vegetative growth when nutrients are abundant to the initiation of fruiting body development upon starvation4. Deletion of pxr from the developmentally proficient lab reference strain GJV1 causes robust fruiting body development and high levels of spore production when growth substrate is abundant, whereas GJV1 undergoes development only when growth resources are scarce or absent. Pxr is produced in two forms, one long (Pxr-L) and one short (Pxr-S), with Pxr-L presumably being processed by a ribonuclease to Pxr-S. Pxr-S appears to be the active form of Pxr blocking development, as Pxr-S levels are rapidly diminished in response to early developmental signals upon onset of starvation whereas Pxr-L remains present at high levels throughout development4.
The pxr gene appears to have originated in the base lineage of the Myxococcales suborder Cystobacterineae, as pxr homologs were found in almost all examined species within this suborder6. Divergent homologs from several Cystobacterineae species, including Stigmatella aurantiaca, as well as the inferred ancestral allele, have been shown to effectively control development in M. xanthus 7, thus indicating that pxr is likely to play a similar role in controlling development across species. pxr appears to have co-evolved with genes encoding a two-component regulatory system that are present upstream of pxr in all myxobacterial species examined to date8. This conserved two-component system (pxrR/pxrK) has been shown to regulate Pxr synthesis and processing8.
Pxr was identified due to a spontaneous mutation that restored developmental proficiency to the developmentally defective strain OC9. OC is a descendant of the developmentally-proficient lab reference strain GJV1 and evolved during 1000 generations of vegetative growth in a high-nutrient liquid environment10. During that evolution experiment, the OC lineage accumulated 14 mutations11 and concomitantly lost the ability to effectively develop and produce heat-resistant spores10. During a subsequent competition experiment, a spontaneous mutation in an OC cell restored developmental proficiency to the resulting mutant, PX9. That mutation occurred within the loop of the first of three predicted stem-loop structures (SL1-SL3) of Pxr and inactivated Pxr function4.
Strain OC is defective at development because Pxr-mediated blockage of development is not alleviated upon starvation (Fig. 1a). Deletion of pxr from OC removes that blockage and allows development to proceed4 (Fig. 1b), as does either deletion of the SL1 loop8 or a single-base substitution in the fifth base of that loop (C to A)4. Thus, Pxr SL1 is essential for control of M. xanthus development. While the functional roles of specific sequence elements in some non-coding regulatory sRNAs in other bacteria have been explored extensively12, it remains unknown whether other segments of the predicted Pxr structure other than SL1 are necessary for Pxr function. Here we examine the functional effects of a series of deletions from the 3′ end of the pxr coding region, including deletions that remove part or all of SL3, as well as deletion of a highly conserved eight base-pair segment of the SL3 stem.
The third stem loop of Pxr is required for control of development
We constructed several plasmids carrying sequences that all begin at the same position within the 3′-terminal region of the pxrR gene (previously known as Mxan_1078 and nla19)8. pxrR lies immediately upstream of pxr and is predicted to encode an NtrC-like response regulator that is part of a two-component signal transduction system along with the predicted histidine kinase gene pxrK (previously known as Mxan_1077)8. These plasmids extend variable lengths to include either the entire predicted pxr coding region (pPxr, pPxr+36 and pPxr+75) or only 5′ portions of pxr lacking part or all of SL3 (pPxr-27 and pPxr−46). Plasmids pPxr+36 and pPxr+75 include sequence extending 36 and 75 bases beyond the predicted 3′ end of pxr, respectively. Additionally, a plasmid carrying sequence also starting from the same position in pxrR as the other plasmids but not including the pxr coding sequence (pPxrΔ) was constructed as a control to test for any effect of plasmid-vector integration. All of these plasmids were integrated into a mutant of strain OC from which the pxr coding sequence was deleted (OC Δpxr) but which still retains the intergenic sequence between pxrR and pxr, thus creating merodiploids for the pxr region (Fig. 2a). All of the resulting transformants share the same 167 bp 3′ terminal sequence of pxrR and the 260-bp region upstream of pxr, including its predicted σ54 promoter and the upstream enhancer binding region.
As reported previously4, deletion of pxr from OC (strain OC Δpxr) restores fruiting body development and spore production to high levels relative to OC (which fails to aggregate into visible fruiting bodies and makes zero or few spores under standard developmental conditions (Figs 1 and 2a,c)). As expected, integration of the negative control plasmid pPxrΔ lacking pxr into OC Δpxr did not eliminate fruiting body formation or reduce spore production in the resulting transformant (Fig. 2a,c). Additionally, integration of all three plasmids carrying the entire pxr coding region (pPxr, pPxr+36 and pPxr+75) into OC Δpxr resulted in blockage of both fruiting body formation (Fig. 2a) and spore production (Fig. 2c), indicating that pxr is functionally expressed. Unlike complete copies of pxr, fragments of pxr lacking half or all of SL3 (pPxr−27 and pPxr−46, respectively) were unable to block development (Fig. 2a). Transformants produced spores at levels similar to those of GJV1 and OC Δpxr (Fig. 2c).
The experiments above were all performed in the context of the OC genomic background, which differs from its evolutionary ancestor GJV1 by 14 known mutations11. As shown both previously4 and here, deletion of pxr from GJV1 (creating strain GJV1 Δpxr) causes development to proceed and produces heat-resistant spores in the presence of abundant nutrients (e.g. 0.3% casitone, Fig. 2d), indicating that pxr controls the transition of M. xanthus from growth on abundant nutrients to development in response to starvation. To demonstrate that the results above are not likely to be specific to the OC background, we introduced a full copy of pxr on plasmid pPxr into strain GJV1 Δpxr. If the introduced pxr is active and performs the same function as the original native pxr, fruiting body development and spore production should be blocked. Indeed this is the case, as integration of pPxr restored pxr blockage of high-nutrient development, whereas integration of the corresponding plasmid vector lacking any pxr sequence (pPxrΔ) did not (Fig. 2d).
The third stem loop of Pxr is essential for Pxr accumulation
Consistent with our developmental assays, integration of all three plasmids carrying the entire pxr gene into OC Δpxr resulted in accumulation of high levels of Pxr-S, the short active form of Pxr that negatively regulates development (Fig. 2e, lanes 2–5). In contrast, integration of the two plasmids lacking part or all of SL3 (pPxr−27 and pPxr−46), yielded no visible Pxr transcript of the expected sizes (Fig. 2e, lanes 6–7). Rather, the respective transformants showed the same transcriptional profile as the control strain lacking pxr (Fig. 2e, lane 8), thus suggesting that the Pxr transcripts lacking part or all of SL3 may be unstable.
Interestingly, levels of the long form of Pxr, Pxr-L, were much lower in the merodiploids transformed with plasmids that included both the entire predicted pxr gene and a downstream extension (Fig. 2e, lanes 2–4) than were present in the transformant carrying only pxr with no extension (Fig. 2e, lane 5). This result implies that the region extending 36 bp downstream of the predicted pxr coding sequence reduces Pxr-L accumulation, but only in the context of these merodiploid transformants of OC Δpxr. Additionally, the accumulated levels of Pxr-L in all three merodiploids (Fig. 2e, lanes 2–5) bearing the pxr gene were significantly lower than that found in the control strain GJV1 (Fig. 2e, lane 1). This reduction may be due to the second copy of the pxr promoter and the upstream enhancer-binding region preceding the pxr-deleted allele in the merodiploids (Fig. 2a) that may sequester important transcription factors such as PxrR and sigma factor away from transcribing the functional Pxr.
A conserved 8-bp segment of Pxr SL3 is essential for blockage of development in both the OC and GJV1 genomic backgrounds
Comparison of pxr sequences across a phylogenetically diverse range of myxobacterial species revealed a highly conserved 8-bp segment within the Pxr SL3 stem (Fig. 3, the boxed region in the sequence alignment and within the simulated SL3). We therefore tested whether this stem segment is necessary for Pxr function by constructing an allele of pxr lacking the respective 16 bases (pxr^, Fig. 3b). Integrating of the plasmid carrying this allele (pPxr^) into OC Δpxr failed to block sporulation (Fig. 4a). Thus, like the two pxr alleles lacking either the entire SL3 sequence (pxr −46) or the 3′-terminal half of SL3 (pxr −27) (Fig. 2a,c), pxr^ is non-functional and the conserved 8-bp segment of the SL3 stem appears to be essential for Pxr function.
We also asked whether deletion of the conserved segment of SL3 would eliminate Pxr function in the GJV1 genomic background. To do so, we integrated pPxr^ into GJV1 Δpxr and tested for fruiting-body formation under high-nutrient conditions. In the presence of abundant nutrients, GJV1 does not form darkened fruiting bodies due to blockage of development by pxr but GJV1 Δpxr does (Fig. 4b). Integration of pPxr^ into GJV1 Δpxr failed to restore the blockage of fruiting body development under high-nutrient conditions and showed the same phenotype as GJV1 Δpxr (Fig. 4b).
Northern-blot analysis revealed that the 8-bp conserved segment of SL3 is necessary for accumulation of Pxr transcript, as no Pxr transcript of any size was observed for GJV1 Δpxr::pPxr^ (Fig. 4c). This segment of the SL3 stem might affect Pxr-transcript accumulation either by stabilizing Pxr-L once it has been produced or by facilitating transcription of pxr.
Non-coding sRNAs primarily modulate expression of target genes by base-pairing to their mRNAs and thus causing mRNA degradation or altered protein translation1,2. Pxr sRNA is highly expressed in the vegetative stage and processed into the putative active smaller form (Pxr-S) to presumably either directly repress developmental genes or enhance the expression of other developmental inhibitors when nutrients are abundant4. However, the direct targets of Pxr (if multiple) and their relationships to genes known to positively regulate the early stages of development remain under investigation.
To facilitate better understanding of how Pxr regulates development, we sought to identify segments pivotal to its function. Our results demonstrate the necessity of Pxr SL3 for regulating the Pxr-mediated developmental pathway. Most studies examining functional components of sRNA regulators have focused on sRNA interactions with regulatory targets (e.g. sequences involved in sRNA binding)13,14,15. In the case of Pxr, however, we have identified a highly conserved 8-bp segment within the third predicted stem loop (SL3:8 bp) that appears to be essential in maintaining levels of Pxr sufficient for effective negative regulation of fruiting body development. In particular, deletion of the invariable 8-bp stem abolishes the accumulation of Pxr transcripts and results in a pxr-null phenotype.
Comparison across all known pxr alleles exclusively identified in myxobacteria showed that the most variable sequences are located in the annotated SL3 (despite the highly conserved 8-bp segment core) while SL1 and SL2 are much more conserved as a whole6. Mutation or deletion within the loop of SL1 eliminates the negative regulatory function of pxr, but does not prevent accumulation of high levels of Pxr transcript, indicating that SL1 plays a key role in Pxr functionality, but not in its transcription or in its stability4,8. Such SL1-mediated Pxr functionality may involve target pairing or RNA-chaperone binding (although M. xanthus does not carry a homolog of hfq 16).
In contrast to the effects of a deletion in the SL1 loop8, deletion of SL3:8 bp in the GJV1 background strongly affects sRNA accumulation (Fig. 4c), which reflects a balance between sRNA synthesis and decay. The necessity of SL3:8 bp for Pxr accumulation might be due to an autoregulatory effect on transcription, as occurs with some other sRNAs17,18. However, there is no evidence to suggest this scenario. In particular, the response regulator pxrR, which has been shown to regulate pxr transcription8, is not predicted to be a target of Pxr (unpublished results).
Alternatively, it is possible that SL3:8 bp is a critical determinant of Pxr stability, such that Pxr transcripts of pxr^ are synthesized but rapidly degrade and are thus not detected on Northern blots. If the SL3:8 bp segment is crucial in protecting the transcripts from ribonuclease attack either by shielding the RNA from ribonuclease binding or by interacting with protective proteins, it is expected that deletion of SL3:8 bp will severely diminish Pxr accumulation. In E. coli, Hfq together with ribonucleases such as PNPase (polynucleotide phosphorylase) and RNasePH have been shown to stabilize some sRNAs and facilitate regulation of their targets19,20. In Pseudomonas putita, Hfq and Crc bind to CrcZ sRNA and protect it from degradation21. However, in those studies the sRNA regions conferring protection against degradation remain elusive. It has been documented that Hfq can recognize the 3′-polyU region of SgrS sRNA to maintain its stability22. While there is no homolog of hfq gene found in myxobacteria, it is possible that some myxobacteria proteins other than Hfq can perform a chaperone-like function to promote sRNA stability. Further analysis of the functional role of SL3:8 bp may identify any such chaperone-like proteins and provide further insights regarding how levels of the myxobacterial sRNA Pxr are regulated during the transition from growth to development.
Strain and plasmid descriptions and primer sequences used in this work are listed in Table 1.
Development assays and sporulation efficiency
The developmental phenotypes of the pxr-derivative strains as well as the controls were examined on TPM23 1.5% agar starvation plates. Mid-log phase cultures grown in CTT24 medium were centrifuged and resuspended in TPM liquid to a density of ~5 × 109 cells/ml. 50 µl of resuspended culture were spotted onto TPM agar plates and incubated at 32 °C. After four or five days, images of developmental phenotypes were taken with a Zeiss Stemi 2000 stereomicroscope. At day five of development the developing populations were harvested into 1 ml of ddH2O, incubated at 50 °C for two hours, sonicated twice for 10 seconds to disperse heat-resistant spores. All sonicated samples were serially diluted in ddH2O and then plated into CTT soft (0.5%) agar. The number of spores produced by each strain was calculated by multiplying the colony count of germinated spores on CTT soft agar by the relevant dilution factor. Development assays were performed in four temporally independent replicates for the experiments reported in Fig. 2c and two replicates for the experiments reported in Figs 2d and 4a.
Northern-blot analysis of Pxr sRNA
Pxr sRNA generated from each strain was analysed in a Northern-blot assay described previously4. Total RNA was prepared from mid-log phase cultures growing in CTT medium and small RNA fragments (<200 nt) were prepared with the MirVana miRNA isolation kit (Ambion). RNA concentrations were estimated by a NanoDrop spectrophotometer (Thermo Scientific). Equal amounts of RNA (0.7 μg) were electrophoresed in a 10% SequaGel (National Diagnostics). RNA transcripts were electro-transferred onto a BrightStar®-Plus positively charged nylon membrane (Ambion) at 180 mA for one hour and were fixed onto the membrane with a UV cross-linker. The membrane was hybridized with 100 pmol 3′Biotin-TEG-pxr oligo probe (Sigma, the probe is complementarily base-paired to the sequences within SL1) in 5 ml of UltraHyb-Oligo buffer (Ambion) overnight at 37 °C after completing pre-hybridization in 4 ml of UltraHyb-Oligo buffer for one hour at 37 °C. The positions and relative content of Pxr RNAs were detected and visualized with BrightStar® Biodetect non-isotopic kit (Ambion).
Plasmid and strain construction
All plasmids were cloned by a similar procedure. The Pxr sRNA coding region is mapped within a 413-bp intergenic region between Mxan_1078 and Mxan_1079 4,9. To make plasmids containing a series of the pxr 3′-terminal truncations, pxr fragments were first PCR-amplified using the same upstream primer located 167-bp upstream of the Mxan_1078 stop codon (GV367) and a series of downstream primers that generated fragments of desired length (Table 1). The resulting PCR fragments were gel-purified, cloned into the pCR-Blunt (Invitrogen) vector and were verified by sequencing. The downstream primers were designed to generate two fragments containing the complete predicted pxr coding region plus extensions of different length beyond the predicted end of pxr (pPxr+75 and pPxr+36), one fragment containing the entire pxr gene with no extension (pPxr), two fragments lacking part of the pxr gene (pPxr−27, which lacks the SL3 loop and the downstream portion of the SL3 stem, and pPxr−46, which lacks the entire predicted SL3 sequence) and one fragment lacking the complete pxr gene (pPxrΔ). To generate pPxr^ bearing the pxr derivative devoid of an 8-bp segment of the SL3 stem, a 48-mer primer missing the 16 respective nucleotides (GV683) was used together with GV367 to generate the PCR insert. All of the plasmids described above thus contain a region homologous to the M. xanthus GJV1/OC genomes (ranging from 427 bp to 612 bp in length) that allows integration via single crossing-over to generate cognate mero-diploid pxr-derivative constructs (Fig. 2a). The end positions of the 3′ primers used to generate inserts for plasmids pPxr, pPxr−27 and pPxr−46 on the predicted three-stem-loop structure of Pxr are shown in Fig. 2b.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This work was supported by U.S. National Institutes of Health Grant R01 GM079690 to G.J.V.