Elimination of PknL and MSMEG_4242 in Mycobacterium smegmatis alters the character of the outer cell envelope and selects for mutations in Lsr2.

Four serine/threonine kinases are present in all mycobacteria: PknA, PknB, PknG and PknL. PknA and PknB are essential for growth and replication, PknG regulates metabolism, but little is known about PknL. Inactivation of pknL and adjacent regulator MSMEG_4242 in rough colony M. smegmatis mc2155 produced both smooth and rough colonies. Upon restreaking rough colonies, smooth colonies appeared at a frequency of ~ 1/250. Smooth mutants did not form biofilms, showed increased sliding motility and anomalous lipids on thin-layer chromatography, identified by mass spectrometry as lipooligosaccharides and perhaps also glycopeptidolipids. RNA-seq and Sanger sequencing revealed that all smooth mutants had inactivated lsr2 genes due to mutations and different IS1096 insertions. When complemented with lsr2, the colonies became rough, anomalous lipids disappeared and sliding motility decreased. Smooth mutants showed increased expression of IS1096 transposase TnpA and MSMEG_4727, which encodes a protein similar to PKS5. When MSMEG_4727 was deleted, smooth pknL/MSMEG_4242/lsr2 mutants reverted to rough, formed good biofilms, their motility decreased slightly and their anomalous lipids disappeared. Rough delpknL/del4242 mutants formed poor biofilms and showed decreased, aberrant sliding motility and both phenotypes were complemented with the two deleted genes. Inactivation of lsr2 changes colony morphology from rough to smooth, augments sliding motility and increases expression of MSMEG_4727 and other enzymes synthesizing lipooligosaccharides, apparently preventing biofilm formation. Similar morphological phase changes occur in other mycobacteria, likely reflecting environmental adaptations. PknL and MSMEG_4242 regulate lipid components of the outer cell envelope and their absence selects for lsr2 inactivation. A regulatory, phosphorylation cascade model is proposed.

Introduction

Serine/threonine protein kinases (STPK) in mycobacteria appear to be involved in regulating cell growth, division and metabolism (Molle and Kremer, 2010). Of the 11 STK in Mycobacterium tuberculosis, only four are ubiquitous in mycobacterial species: PknA, PknB, PknG and PknL. Knockouts or overexpression of the essential PknA and PknB alter the shape of the bacilli, probably due to their involvement in the formation of the cell wall (Kang et al., 2005). Of the remaining 9 STK in M. tuberculosis, only PknL belongs to the same clade as PknA and PknB (Narayan et al., 2007, Wehenkel et al., 2008), but few studies have addressed its possible functions (Refaya et al., 2016) (Naz et al., 2021). PknL was shown to phosphorylate an adjacent and divergently transcribed, putative transcriptional regulator, Rv2175c (Canova et al., 2008, Cohen-Gonsaud et al., 2009), reducing its ability to bind DNA, but it is unknown what stimulates this phosphorylation, what genes Rv2175c regulates or what role PknL plays in the biology of the bacteria. This transcriptional regulator and the position of the gene encoding it are conserved in most mycobacteria.

Nine of the 11 STKs in M. tuberculosis, including PknA, PknB and PknL, contain transmembrane regions (Av-Gay and Everett, 2000). Based on the eukaryotic model for serine/threonine and tyrosine kinases, it is thought that the extracellular regions of the M. tuberculosis STPK may perceive a signal that activates the kinase regions within the bacteria. The extracellular region of PknB contains four PASTA (Penicillin-binding proteins And Serine/Threonine kinase Associated) (Bellinzoni et al., 2019) domains that are thought to interact with the peptidoglycan or its muropeptide precursors (Bellinzoni et al., 2019) to regulate PknB autophosphorylation. Mycobacterial PknL’s have <10 extracellular amino acids beyond the transmembrane region, not enough for an extracellular receptor motif, and Mycobacterium tuberculosis PknL does not appear to be autophosphorylated but rather is phosphorylated by PknB and PknJ (Baer et al., 2014). However, the PknL in bacteria such as Corynebacterium glutamicum (Schultz et al., 2009) contain extracellular PASTA domains, suggesting that in these bacteria PknL could also be involved in regulating the synthesis of components of the cell wall or cell envelope.

To study the role of PknL in mycobacteria, we constructed strains of Mycobacterium smegmatis lacking functional copies of pknL (MSMEG_4243) and the adjacent putative transcriptional regulator MSMEG_4242, as well as knockout strains containing intact, complementing copies of one or both genes. Some of the knockout strains formed smooth colonies distinct from the rough colony parent mc. The strains were characterized for phenotypes associated with the cell envelope, their lipid composition and transcriptomes.

Results

Construction of ΔpknL/ΔMSMEG_4242 mutants and complementing phenotypes

The 5′ end encoding the N-terminal regions of the adjacent and divergently transcribed genes pknL and MSMEG_4242 were replaced with a kanamycin resistance cassette (Fig. 1A). In a first experiment, the double mutant was constructed in the rough colony parental strain M. smegmatis mc as well as in strains containing integrating plasmid pYUB412int carrying either pknL, MSMEG_4242, or both genes (EB1 – 4, Fig. 1A & C,). Very few colonies were obtained, and all had a smooth morphotype (Fig. 2A). The deletion of both genes was initially confirmed by PCR amplification and restriction digests and subsequently by RNA-seq (data not shown). The mutant strains that contained plasmid pYUB412int with pknL and MSMEG_4242, separately or together, altered the morphology slightly, but the colonies remained smooth (Fig. 2A). The knockout experiment was repeated by another of the authors, using the same construct to eliminate both genes from the parent mc. In this second experiment many colonies were obtained, most of which were rough, but about 1/100 were smooth (Fig. 1D). Surprisingly, it was confirmed that the two genes had been eliminated in both smooth and rough colony morphotypes (data not shown). When the rough colony knock out strains were restreaked, smooth colonies appeared at a frequency of about 1/200 – 1/300 . In this second series of experiments, single knockouts of pknL and MSMEG_4242 were also constructed (Fig. 2B). Most of them showed the same rough phenotype as the double mutants and smooth colonies spontaneously appeared at the same frequency as they arose in the double mutant strains. The single knockout strains were not studied further. Both genes were also inserted, separately, into expression vector pMV261 and the plasmids were transformed into mc2155. The transformants showed no apparent alteration from the rough phenotype of the parent strain and were not studied further.

Fig. 1

A. Strategy for deleting both MSMEG_4242 (a putative transcriptional regulator) and MSMEG_4243 (pknL). A kanamycin resistance cassette replaced the 5′ terminal of both genes along with the intergenic region. This was inserted into plasmid pPR27, which carries a temperature sensitive mycobacterial origin of replication, the sacB gene and a gentamycin resistance cassette (Pelicic et al., 1997). B. Strategy for construction of strains lacking either MSMEG_4242 or MSMEG_4243 (pknL). C. Schematic diagram of strains originating from the first experiment to knock out both MSMEG_4242 and MSMEG_4243 (pknL) - EB1 – 4 and derivatives.D. Schematic diagram of strains originating from the second experiment to knock out both genes.

Fig. 2

.Fig. 2 Phenotypes of mutant strains. A. Colony Morphology of wt and mutant strains obtained in the first knockout experiment. B Colony Morphology of other strains in this study. C. Histogram showing measured length of 100 bacilli of each strain: wt, mc2155; KO, EB1- mc2155 ΔpknL/ΔMSMEG_4242; KO + 42, EB2 ΔpknL/ΔMSMEG_4242 + pYUB412intMSMEG_4242; KO + pknL, EB3 ΔpknL/ΔMSMEG_4242 + pYUB412intpknL.D.Histogram of the average lengths of 100 bacilli shown in C. Asterisk (*) indicates that the difference is statistically significant, P ≤ 0.05. E. Biofilms of the different strains, in duplicate. The +’s or –‘s indicate the quality of the biofilm. F. Sliding motility of the different strains, in duplicate.

The smooth colony mutant strains were inoculated into liquid media and grew slightly, but not significantly slower than the mc parent (data not shown). The uncomplemented smooth colony mutant EB1 (mc2155 ΔpknL/ΔMSMEG_4242) also formed longer bacilli (Fig. 2C & D) and appeared to show weaker acid-fast staining (Suppl. Fig. S1). Complementation with pknL and MSMEG_4242 restored both parental bacillus length and acid-fast staining. Strain EB1 was also two-fold more sensitive to rifampicin and cefoxitin than the parent mc2155 (Table 1 and Fig. S2). The sensitivity to rifampicin was restored to that of the parent by complementation with just pknL, while complementation with both pknL and MSMEG_4242 were required to regain parental cefoxitin sensitivity.

Table 1

Sensitivity of mutants and complemented mutants to Rifampicin and Cefoxitin.

	Rifampicin		Cefoxitin
	32 ug/ml	64 ug/ml	12.5 ug/ml	25 ug/ml
mc2155	++	++	++	++
Del pknL/4242/lsr2 smooth	++	--	++	--
Del pknL/4242/lsr2 smooth (4242)	++	--	++	+
Del pknL/4242/lsr2 smooth (pknL)	++	++	++	--
Del pknL/4242/lsr2 smooth (pknL/4242)	++	++	++	++
Del lsr2	++	+
Del lsr2 (lsr2)	++	++
Del pknL/4242 rough	++	++
Del pknL/4242 rough (pknL/4242)	++	++

Altered biofilms and sliding motility

Smooth colony morphotypes have been associated with alterations in the lipid composition of the outer cell envelope (Kocíncová et al., 2008), so we examined two phenotypes associated with cell envelope lipids – sliding motility (Recht and Kolter, 2001) and biofilm formation(Kocíncová et al., 2008). While the rough colony parent mc2155 forms strong biofilms and has moderate sliding motility, none of the smooth colony strains formed biofilms (Fig. 2E & 5), all had increased sliding motility (Fig. 2F & 5), and neither phenotype was restored to that of the parent by complementation with pknL and MSMEG_4242 (Table 2).

Fig. 5

Summary of colony morphology, biofilm formation and spreading motility. Colony morphology of the strains whose genotypes are shown on the right. The height of the bars in the histogram indicates the average diameter of the sliding motility halos of three independent experiments, with duplicates of each strain in each of the three experiments. One experiment is shown in Fig. 2G. Asterisks (*) indicate a significant difference in the diameter of the halo: *, P < 0.05; **, P < 0.005 The biofilms formed by each strain are indicated by +’s or –‘s according to the key shown below the colonies, based on three experiments, each with duplicates, as in Fig. 2F.

Table 2

Phenotypes of parent and mutant strains.

	pknL/4242	lsr2	4727	Biofilm	Motility (in mm*)	Morphology
A: mc2155	Yes	Yes	Yes	+++	38	rough
D: EB1	No	No	Yes	-	65	smooth
E: EB1 + p4242/4243	yes	No	Yes	-	68	smooth
I: EB1-Δ4727	No	No	No	+++	55	rough
H: mc2155 Δ4727	Yes	Yes	No	+++	20	rough
F: Δ4242/4243(R)	No	Yes	Yes	+	20#	rough but altered
G: Δ4242/4243 + p4242/4243	yes	Yes	Yes	++	50	rough
B: Δlsr2	Yes	No	Yes	- -	48	smooth
C: Δlsr2 + plsr2	Yes	yes	Yes	- -	28	rough
J: Δlsr2-Δ4727	Yes	No	No	-	4	very rough

*, Motility measured as millimeters of halo diameter.

#, irregular halo.

+++ Mature floating biofilm.

++ Mature biofilm.

+ Slow formation of biofilm.

- Abnormal biofilm.

- - No biofilm.

In contrast, the sliding motility of the rough mc2155 ΔpknL/ΔMSMEG_4242 mutant was about half that of mc2155 and was irregular and jagged, clearly different from the circular motility halos of all the other strains (Fig. 2F & 5). In addition, the rough ΔpknL/ΔMSMEG_4242 mutants could form only very weak, delayed biofilms. When the rough ΔpknL/ΔMSMEG_4242 was complemented with a plasmid containing pknL and MSMEG_4242, the strain formed biofilms only slightly weaker than the unmutated mc2155 parent and its sliding motility showed a regular halo larger than that of the parent (Fig. 2E, 2F & 5 and Table 2). Thus, while the reduced motility and poor biofilms in the rough colony mutants were attributable to the absence of pknL and MSMEG_4242, the increased motility and lack of biofilms in the smooth mutants was not reversed by intact copies of these two genes.

Mutations in lsr2 were responsible for the smooth colony phenotype

RNA-seq of the smooth colony mutants (EB1-4) revealed that all contained mutations that inactivated the lsr2 gene (MSMEG_6092) (Chen et al., 2008, Gordon et al., 2010). EB1 (ΔpknL/ΔMSMEG_4242) had a new IS1096 insertion into lsr2, while EB2 (ΔpknL/Δ4242- MSMEG_4242) had a different new IS1096 insertion into the lsr2 gene. RNA-seq showed an increase in the expression of IS1096 transposase tnpA from some of the IS1096 insertions in EB1 and EB2 (Table 3). Strain EB3 (ΔpknL/Δ4242-pknL) had a one base, frameshifting insertion into lsr2 (Fig. 3), while strain EB4 (ΔpknL/ΔMSMEG_4242 – pknL/MSMEG_4242), showed the same frameshifting but only in a minority of RNA-seq reads. This frameshifting mutation was also present in a small minority of reads from wt parent mc2155 and was seen in three reads from double knockout strain EB1. With successive passaging and restreaking, strain EB4 eventually reverted to a rough colony morphology. RNA-seq also showed increased expression of the IS1096 transposase tnpA from some of the other IS1096 insertions in EB1 and EB2 (Table 3), but no other new IS1096 insertions were found.

Table 3

RNA-seq results of strains EB1, EB2 and EB3 vs EB4 and/or wt mc2155

Fig. 3

RNA-seq reads in the region of lsr2 (MSMEG_6092): EB1 has an IS1096 insertion eliminating lsr2 transcription; EB2, has a different IS1096 insertion; EB3, showing insertion of a T that causes a frameshift; EB4, a small minority of reads show the same nucleotide insertion as in EB3; wt mc2155, a small minority of reads show the same nucleotide insertion as in EB3.

PCR amplification and Sanger sequencing showed that all smooth colony mutants isolated in the second knockout experiment, or smooth colonies that appeared spontaneously on restreaking these rough colony double mutants, also contained mutated lsr2 genes, including IS1096 insertions, a frameshift due to a nucleotide deletion and a mutation substituting an alanine for an aspartic acid at amino acid 35 (D35A) (Fig. S4). In contrast, the lsr2 genes were intact in all rough colony mutants. When strain EB1 (ΔpknL/ΔMSMEG_4242/Δlsr2) was transformed with a plasmid carrying an intact copy of lsr2, the colony morphology reverted to rough.

Analysis of cell envelope lipids

Thin-layer chromatography (TLC) of smooth colony strain EB1 showed anomalous bands, predominantly in the polar fraction, that disappeared when the strain was transformed with a plasmid carrying lsr2 and became rough (Fig. 4A & B.). These anomalous bands were not present on TLC of the rough colony ΔpknL/ΔMSMEG_4242 strains.

Fig. 4

Thin layer chromatography. Anomalous lipid bands in lsr2 mutant strains were eliminated either by inactivating the MSMEG_4727 (PKS5) gene (A), or complementing with a plasmid containing an intact copy of lsr2 (B). C. Mass spectrometry of wt mc2155 and EB1 total lipids, or D. fraction 8.

Mass spectrometry (MS) of cell wall lipids of the uncomplemented smooth colony mutant EB1 revealed anomalous peaks that were not present in parent strain mc2155 (Fig. 4C & D). The mass range of the peaks suggested that they were likely lipooligosaccharides A (LOS A) and C (LOS C) (Etienne et al., 2005), but there were also anomalous peaks in the size range of glycopeptidolipids (GPLs) (Etienne et al., 2005).

Elimination of MSMEG_4727 restored rough colony morphology in strain EB1.

The RNA-seq study of smooth colonies EB1 – EB3 also showed a marked increase in the transcripts of MSMEG_4727 and other downstream genes implicated in the synthesis of LOS (Table 3) (Etienne et al., 2009, Colangeli et al., 2007). MSMEG_4727 encodes a polyketide synthase similar to the PKS5 associated with the production of LOS in smooth colony Mycobacterium canetti. It was shown to be inactivated in rough colony M. canetti strains and M. tuberculosis (Boritsch et al., 2016). When the MSMEG_4727 gene was eliminated in strain EB1 to create EB1- Δ4727 (ΔpknL/Δ4242/Δlsr2/Δ4727), the colony morphology reverted to rough, although slightly different from the parent mc (Figs. 5 and S3). The EB1-Δ4727 strain was capable of forming normal biofilms and its sliding motility was reduced but still about 50% greater than that of parent mc (Fig. 2E, 2F, 5 and Table 2). When MSMEG_4727 was eliminated from parent strain mc, the resulting mutant could form normal biofilms but the sliding motility was reduced by nearly half. In contrast, when MSMEG_4727 was deleted from an lsr2 mutant obtained by transposon mutagenesis of mc strain NJS20 (Colangeli et al., 2007); there was no motility and the strain could not form biofilms (Kocíncová et al., 2008). These results confirm that the inactivation of lsr2 in the smooth colony pknL/4242 mutants led to increased production of LOS from the MSMEG_4727 synthesis pathway, which prevented the formation of biofilms but contributed to motility.

Discussion

This study was undertaken to explore the role of PknL in mycobacterial biology but led to unexpected results. All smooth colony pknL/MSMEG_4242 knockout strains harbored lsr2 genes that had been mutated by frameshifts, amino acid substitutions or IS1096 insertions, perhaps associated with the increased transcription of transposase TnpA seen on RNA-seq. The smooth mutants had two sets of phenotypes – those stemming from the lack of PknL/MSMEG_4242 and those caused by absence of Lsr2. The phenotypes of the smooth triple mutant smooth EB1 strain (ΔpknL/ΔMSMEG_4242/Δlsr2) that were reversed with a plasmid carrying pknL/MSMEG_4242 were: longer bacilli; poor acid-fast staining; and increased sensitivity to rifampicin and cefoxitin (Table 1 & Fig. S2). The phenotypes complemented by intact copies of lsr2 included the smooth colony morphology, the inability to form biofilms, increased sliding motility and additional lipid species, probably LOS-A, LOS-C and subfamilies of GPL (Etienne et al., 2005). These are reminiscent of the phenotypes of the original M. smegmatis strain ATCC607 that were lost when it evolved into variant mc2155(Etienne et al., 2005). Eliminating the PKS enzyme MSMEG_4727 in EB1(mc2155 ΔpknL/ΔMSMEG_4242/Δlsr2) reversed the smooth colony morphotype and reduced motility, but when the single lsr2 mutant was complemented with lsr2, it still could not form biofilms. This single lsr2 mutant strain was obtained from another lab (Colangeli et al., 2007) and may have acquired additional genetic changes, which appear to be frequent in M. smegmatis strains(Kocíncová et al., 2008, Etienne et al., 2005).

To fit these results into a regulatory model, we compared our findings with previously published studies. PknL was shown to be incapable of autophosphorylation but was phosphorylated by PknB and PknJ and colocalized with PknB to the bacillus poles and midplane septa (Baer et al., 2014), sites thought to be related to bacterial replication. A recent study found that a strain of M. tuberculosis lacking PknL showed decreased survival in macrophages and the spleens of infected mice, and also showed increased Minimum Inhibitory Concentrations (MICs) for isoniazid and ethambutol (Naz et al., 2021). We found no difference in resistance to isoniazid, but a two-fold increase in susceptibility to rifampicin and cefoxitin. These are large molecules, so changes in their susceptibility could reflect changes in cell wall permeability(Kolodziej et al., 2021). It was reported that the absence of Lsr2 in M. smegmatis increases susceptibility to rifampicin and nalidixic acid(Kolodziej et al., 2021), and that overexpression of PknB increased susceptibility to rifampicin and vancomycin(Le et al., 2020) likely due to alterations of the cell envelope. Zeng et al. found that depletion of either PknA or PknB from M. tuberculosis reduced the MICs for cephalosporins, carbapenems and rifampicin (Zeng et al., 2020). We found that a strain with a mutated lsr2 had increased susceptibility to rifampicin that was complemented with a parental copy of lsr2 (Table 1). However, there was a greater increase in susceptibility in smooth strains lacking PknL and MSMEG_4242 in addition to Lsr2, and the rifampicin susceptibility returned to that of the parent when these strains were complemented only with pknL and MSMEG_4242 (Table 1 and Suppl. Fig. 2). The lack of an increase in susceptibility in the rough ΔpknL/ΔMSMEG_4242 mutants suggests that the deletion of PknL and MSMEG_4242 has a synergistic effect with Lsr2 mutations in altering the cell envelope. It has been reported that Lsr2 also controls pigment in M. smegmatis (Kocíncová et al., 2008); but although some of the mutant strains had altered pigmentation, we did not analyze this phenotype.

In a study in M. tuberculosis using an antisense strategy, a strain with a PknL knockdown showed no change in colony morphology but, similar to our results, the mutant bacilli grew slower and were ~0.8 µm longer. They also found that the knockdown strain had lower viability over time but increased resistance to pH 5.5, SDS and lysozyme (Refaya et al., 2016). In contrast, a very recent study described shorter and wider bacilli in an lsr2 mutant strain(Kolodziej et al., 2021).

Several publications have addressed Lsr2, a nucleoid-associated protein that has been associated with cell cycle(Kolodziej et al., 2021) and is essential in M. tuberculosis (Colangeli et al., 2007, Kolodziej et al., 2021, Kolodziej et al., 2021) (Chen et al., 2008, Gordon et al., 2010) but not in M. smegmatis. The genes that Lsr2 appears to regulate – its regulon, is broad and overlaps with the PknB regulon (Alqaseer et al., 2019). It was previously observed that smooth colonies of M. smegmatis appear as spontaneous variants of M. smegmatis ATCC607, the parent strain of mc2155, as a result of insertions of IS1096 into lsr2 (Kocíncová et al., 2008). These hypermotile variants, which are unable to form biofilms, appeared in a starvation experiment with a relA (stringent response) mutant at a frequency of approximately 10-4 (Arora et al., 2008), higher than the expected frequency of about 10-6. In our rough colony pknL/MSMEG_4242 knockout mutants, the insertions, deletions, amino acid substitutions or IS1096 insertions in lsr2 occurred at a frequency of nearly 10-2, which is unexpectedly high but perhaps partially explained by the increased transcription of IS1096 transposase TnpA. The growth rate, bacillary length and rifampicin sensitivity of lsr2 smooth colony mutants were not previously reported to be different from those of the parent strains, although similar to our smooth strains, they showed increased sliding motility (Colangeli et al., 2007, Arora et al., 2008, Chen et al., 2006). Smooth colony lsr2 mutant strains also had anomalous lipid species, generally polar lipids identified as GPLs (Colangeli et al., 2007) that were present in greater amounts than in the parent strains (Kocíncová et al., 2008). GPLs have been shown to be essential for sliding motility (Recht et al., 2000). One study (Chen et al., 2006) found that two apolar mycolyl-diacylglycerols (MDAGs) were lost in an mc2155 strain with an insertion in lsr2, but a recent study convincingly showed that LOS are produced in lsr2 mutants(Kolodziej et al., 2021).

Apparently, M. smegmatis can have many types of variants. The original ATCC607, from which the transformation competent mc2155 was selected (Snapper et al., 1990), could not form biofilms and had low motility (Etienne et al., 2005), but a strain with an IS1096 inserted into lsr2 showed increased production of GPLs, increased motility and could not form biofilms. A similar phenotype was found in strains with IS1096 insertions upstream of MSMEG_0400, the first gene of the mps operon, whose encoded enzymes synthesize GPLs, and transcription of this operon was increased. Genome sequencing found that mc has an IS1096 insertion into this promoter region(Kocíncová et al., 2008). It was reasoned that Lsr2 negatively regulates the synthesis of GPLs by the mps gene cluster (Kocíncová et al., 2008) but the IS1096 insertions abolished the Lsr2 regulatory site. The RNA-seq studies of the smooth colony mutant EB1, which lacks an intact lsr2 gene, did not show an increase in the transcription of the genes in the mps cluster, perhaps because it is already highly expressed in mc (Kocíncová et al., 2008) due to the IS1096 insertion upstream of the mps operon that eliminated the Lsr2 binding site.

Our triple mutant EB1 (ΔpknL/ΔMSMEG_4242/Δlsr2) was shown by MS to have several lipid species that are not present in the parent mc2155, probably LOS-A, possibly LOS-C and perhaps GPLs. It was previously shown that the synthesis of these LOS, which were present in the original M. smegmatis ATCC607 but lost in the transformation competent M. smegmatis mc2155, depends upon the PKS protein encoded by MSMEG_4727 (Etienne et al., 2009). In strains EB1, 2 and 3 -- the smooth colony mutants with inactivated lsr2 genes, the transcription of MSMEG_4727 was several times greater than in the parent strain (Table 3), as previously reported for other lsr2 mutant strains (Colangeli et al., 2007, Kolodziej et al., 2021). When MSMEG_4727 was eliminated, the strain regained the rough morphology of the parent mc2155, formed strong biofilms and had reduced motility, although still 50% greater than the original mc2155 progenitor. Similar results were recently reported for another lsr2 mutant(Kolodziej et al., 2021).

Lsr2 is a two-domain protein composed of an N-terminal dimerization domain and a C-terminal DNA-binding domain. The crystal structure of the truncated N-terminal dimer domain (Summers et al., 2012) revealed an elongated molecule made up of a single four-stranded beta-sheet with two helices on one side. The carboxylate groups of the two Asp35, which are replaced by Alanine in some of the smooth mutants, are completely exposed to the solvent at each extremity (Figure S4). The N-terminal domain can also promote higher-order Lsr2 assemblies, providing a mechanism for the role of Lsr2 in the compaction and physical protection of DNA (Colangeli et al., 2009). Since Asp35 is directly engaged in dimer-dimer electrostatic contacts leading to higher oligomers, the Asp35-Ala substitution might limit the ability of Lsr2 to form these higher-order DNA binding assemblies.

Lsr2 binds to multiple DNA targets, but when phosphorylated by PknB on threonine 112, its ability to bind to DNA is abrogated (Alqaseer et al., 2019). Thus, phosphorylation of Lsr2 would lead to increased expression of the genes it represses, such as MSMEG_4727 (Kocíncová et al., 2008, Kolodziej et al., 2021). Perhaps PknL can also phosphorylate Lsr2, but as PknL cannot autophosphorylate (Baer et al., 2014), there may be a phosphorylation cascade whereby PknB phosphorylates PknL, which in turn phosphorylates Lsr2 (Fig. 6 A-C). However, there is currently no experimental evidence that PknL can phosphorylate Lsr2, which may only be phosphorylated by PknB (Fig. 6E).

Fig. 6

Hypothetical schemes of how PknL might function to produce the phenotype seen in the mutant strains. While the loss of lsr2 is responsible for several of the phenotypes in the EB series smooth mutant strains, the rough mutants have altered motility and biofilm formation despite having an intact lsr2 gene. The gene or genes responsible for the phenotypes of the rough strains are unknown. A. PknB phosphorylates PknL and PknL phosphorylates Lsr2, which abolishes its ability to bind DNA and thus relieves the repression of LOS-A and GPL synthesis. B. In an lsr2 mutated strain there is unrepressed synthesis of LOS-A and GPLs PknL phosphorylates Lsr2.C. In a pknL knockout, the unphosphorylated Lsr2 represses LOS & GPL synthesis. D. In an alternative scheme, PknB phosphorylates PknL (MSMEG_4243) and PknL phosphorylates the putative transcriptional regulator encoded by the adjacent gene MSMEG_4242, which then loses its ability to bind DNA and repress gene transcription, resulting in the synthesis of GPL and LOS-A. E.Another possibility is that PknB phosphorylates Lsr2 directly, without PknL serving as a phosphorylation cascade intermediate.

PknL has been shown to phosphorylate the protein encoded by the adjacent and divergently transcribed, putative transcriptional regulator Rv2175c, the M. tuberculosis equivalent of MSMEG_4242, which then loses its ability to bind DNA (Molle and Kremer, 2010). In an alternative model, the unphosphorylated MSMEG_4242/Rv2175c represses the expression of LOS and GPL synthesis and the repression is relieved when this transcriptional regulator is phosphorylated by PknL, which in turn is phosphorylated by PknB (Fig. 6D). The phosphorylation of Rv2175c or Lsr2 would abrogate their ability to bind DNA and repress transcription, which would induce the expression of the lipid-producing genes they regulate. This is consistent with other studies showing that phosphorylation by PknB has a negative effect on protein function, which also results in decreased synthesis and transport of mycolates to the envelope (Le et al., 2020, Kolodziej et al., 2021).

An intriguing finding from this study was that strains lacking pknL and MSMEG_4242 accumulate mutations or insertions in lsr2 at a very high frequency, ~ 10-2. Presumably this is because the lack of PknL somehow causes stress in the bacteria, which is reflected in the increased transcription of tnpA, the IS1096 transposase. The rough colony pknL/MSMEG_4242 mutants did not show altered lipid bands by TLC but had a reduced sliding motility that was aberrant, jagged and asymmetric (Fig. 2F), as well as weak, delayed formation of biofilms (Fig. 2E), suggesting that PknL is involved with the synthesis of cell envelope components. Other viable M. smegmatis variants, including ATCC607, also lack sliding motility but show a much lower frequency of insertions into lsr2. However, the very abnormal motility in the rough ΔpknL/ΔMSMEG_4242 mutants may reflect severe, unknown alterations in the cell envelope. Perhaps the LOS or other lipid species produced by lsr2 inactivation somehow compensate for these undefined defects. It is interesting that in EB4, the ΔpknL/ΔMSMEG_4242 strain with integrated pknL/MSMEG_4242, a frameshifting nucleotide deletion was seen in very few reads. The EB4 mutant was originally smooth but changed its phenotype with passaging and the RNA-seq was performed after many passages. This could indicate that the defect that led to the lsr2 mutations was alleviated by the integrated complementing genes, or alternatively, possible contamination with the parent strain. This same frameshifting mutation was seen in reads from EB3, in a minority of reads in the parent strain and was even present in three reads from EB1.

The phenotypic changes produced by the inactivation of lsr2 suggest a phase change from a strain that forms biofilms and has moderate motility, to one that does not form biofilms, has high motility and is therefore more likely to spread rather than attach to a surface. The lipid components of the outer cell envelope seem to spontaneously vary in M. smegmatis, giving rise to many variants (Kocíncová et al., 2008), which may be a mechanism for survival in different environments(Kolodziej et al., 2021) (Wang et al., 2008). In M. abscessus, the rough colony strains lack GPLs present in smooth morphotypes and are more virulent in animal models, but the smooth M. abscessus variants are perhaps better suited for survival in environmental niches(Roux et al., 2016, Bernut et al., 2014, Howard et al., 2006). In contrast with M. smegmatis, though, deletion of lsr2 in M. abscessus has no effect on the GPL profile, at least under laboratory growth conditions (Le Moigne et al., 2019). Similarly, it was recently suggested that rough morphotypes of M. kansasii might be an adaptation favorable for human pathogenicity (Luo et al., 2021). Perhaps the smooth M. canettii, or its recent ancestors, existed in an environmental niche but lost their LOS when they evolved into the rough, aggregative pathogen M. tuberculosis (Boritsch et al., 2016), which is not known to grow outside of mammalian hosts. Although we do not know exactly how PknL functions, nor exactly which lipid components determine sliding motility, biofilm formation and colony morphotype, this study suggests that both Lsr2 and PknL are involved in regulating components of the outer cell wall of M. smegmatis, but this regulation seems complex. Likely the conserved transcriptional regulator adjacent to PknL (MSMEG_4242/Rv2175c), as well as additional regulators such as other nucleoid-associated proteins (Ghosh et al., 2013) or other STPK could also be involved in the synthesis(Le et al., 2020) and export(Le et al., 2020, Pérez et al., 2006, Melly et al., 2021) of the components of the cell envelope that determine these complex phenotypes.

Methods

Construction and selection of mutants.

Vector pUC19, modified to eliminate the BamHI site, was used to clone a 4450 bp fragment amplified from a boiled lysate of mc with primers PknLXbaIFw and PknLXbaIRv (Table S1). This fragment included MSMEG_4242, MSMEG_4243 (pknL) and 1500 pb on either side of these genes. This construct was then cut with BamHI to eliminate the 5′ end of MSMEG_4242, the intergenic region and half of MSMEG_4243 (pknL), including the kinase motif, and then religated with a kanamycin resistance cassette cut from pUC4K with BamHI. The entire insert, with the Kan cassette, was cut with XbaI and ligated into pPR27.

The pPR27-ΔpknL/ΔMsmeg4242::kanR construct and control vector pPR27 and modified pPR17-pJV53 were electroporated into M. smegmatis mc2155, and also M. smegmatis mc2155 strains containing chromosomally integrated pYUB412int plasmids with MSMEG_4242, MSMEG_4243(pknL), or both genes. The electroporated strains were then plated onto 7H10-OAD (Oleic Acid, Albumin Dextrose) with kanamycin and gentamycin and incubated at 32 °C. for 4–5 days(Pelicic et al., 1997). Colonies from these plates were inoculated into 5 ml of 7H9-OAD-kanamycin (25 µg/ml) and grown until saturation for 4 days at 32 °C. From these liquid cultures, 5 ul (first expt.) or 100 ul (second expt.) were plated onto 7H10-OAD-kan (25 µg/ml) with 10% sucrose and grown at 39 °C. until colonies appeared (10 – 15 days). All of the colonies that appeared were grown in 7H9-OAD-Kan (25 µg/ml) at 39° C (first experiment) or 37 °C (second experiment), and the second crossover event was confirmed by the inability to grow on 7H10-OAD plates with gentamycin 5 µg/ml.

To construct knockouts of single genes, first plasmid pUC119 was cut with PstI and filled-in with Klenow (NEB) to make pUC2P, and separately the HindIII site was similarly eliminated to make pUC2H. Two 1500  bp fragments, one containing MSMEG_4241 and MSMEG_4243(pknL) and the other MSMEG_4242 and MSMEG_4244 (Fig. 1Ba) were amplified from an M. smegmatis boiled lysate with primers 1KOFw4242-1KORv4242, 2KOFw4242-2KORv4242 and 1KOFw4243-1KORv4243, 2KOFw4243-2KORv4243 (Table S1). The amplified fragments containing MSMEG_4242 were cut with PstI, eliminating most of this gene, then religated and inserted into the XbaI site of pUC3P to make pUCKO4242 (Fig. 1Bb). The amplified fragments containing MSMEG_4243(pknL), were cut with HindIII, religated and then inserted into the XbaI site of pUC2H to make pUCKO4243 (Fig. 1Bc). The aph kanamycin resistance cassette was then inserted into the PstI site of pUCK04242 and the HindIII site of pUCK04243 (Fig. 1B d & e). The plasmids were transformed into E. coli XL1Blue and plated on agar with kanamycin and carbenicillin. The fragments MSMEG_4242::KanR and MSMEG_KO4243::KanR were then cut from the pUC plasmids and ligated into vector pPR27 and electroporated into M. smegmatis mc2155 for allelic exchange, as described above.

The lsr2 mutant strain was the kind gift of Roberto Colangeli, along with lsr2 containing, complementing plasmid pMP161.

Confirmation of mutants.

Gentamycin sensitive colonies were grown in 7H9-OAD-Kan and 1 ml of culture was centrifuged, resuspended in 150 ul sterile distilled water and boiled for 10 min. From this boiled lysate, 10 ul were used in a PCR amplification with primers Msmeg_4242pknLsmeg412Fw y pknLsmeg412Rv (Table S1), and the product was sequenced in CeSAAN (IVIC, Caracas, Venezuela) and subjected to restriction analysis. The elimination of the two genes was also subsequently confirmed by RNA-seq.

Cloning of complementing integrative plasmids.

Vector pYUB412int was transformed into E.coli K12 ER2925 (dam) in order to use the BclI cloning site, and grown in LB broth with carbenicillin 50 ug/ml. The amplified fragments containing MSMEG_4242, MSMEG_4243(pknL), or both genes (Table S1) were ligated into the pYUBint vector using the BclI, or PacI sites, and then electroporated into mc2155 (Snapper et al., 1990) and plated onto 7H10-OAD-Hygromicin 50 µg/ml, and the presence of the inserts verified by PCR amplification and restriction digests.

Ziehl-Neelsen staining

Ziehl-Neelsen staining was performed using the traditional method. Slides spread with 10 µl of a bacterial suspension were fixed over a flame. They were then covered with phenol fucsin, heated over a flame three times during a period of ten minutes, and then washed with distilled water. The slides were then covered with acid alcohol for 3 min and again washed with distilled water. Finally, the slides were covered with methylene blue for one minute, washed, left to air dry and examined with a 100X objective microscope.

Measurements of bacilli length.

From exponential phase cultures (DO600 = 0,8–1), 10 µl were placed on glass microscope slides and air dried. After adding 2 µl of glycerol, the slides were examined with a Nikon E-600 microscope using a 100X oil immersion objective. DIC (Differential Interference Contrast) images were taken, blinded to strain identification, with a digital Hamamatsu camera using the Methamorph program. The lengths of 50 bacilli of each strain were measured with Image-J software.

Motility assay.

Colonies were inoculated into 7H9 with 10% OAD and 0.05% Tween 80 and grown to mid log phase (D.O600nm = 0.4 – 0.6). A toothpick was dipped into the culture, excess drops were allowed to fall off, and it was stabbed into the center of a Petri dish containing 7H9 media without glycerol and 0.3% agarose. The diameter of the motility halo was measured after 2 days of incubation at 37 °C.

Biofilm assays

Biofilm formation was assessed by inoculating 20 ul of a bacterial culture (OD600, 0.8 to 1.0) onto the wells of a 6 well plate, each containing 4 ml of M63 salts supplemented with 10% glucose, 1 mM CaCl2 and 1 mM MgSO4 and then incubated without shaking at 30 °C for 5 days.

Determination of rifampicin sensitivity

The strains were grown in 7H9-OAD at 37 °C. until reaching log phase (1–5 × 108 CFU/ml), diluted 1:100 and then 10 ul of the dilution was inoculated onto 2 mls of 7H10-OAD with rifampicin 32 or 64 µg/mL in a 24 well plate and incubated at 37 °C for 3 to 5 days until confluent growth was observed. Each strain was tested in triplicate.

RNA-seq analysis

The strains were grown in 7H9-OADC at 37 °C, RNA was isolated, reverse transcribed and sequenced as previously described (Solans et al., 2014). Illumina reads were mapped against the M. smegmatis mc2155 reference sequence (NCBI RefSeq: NC_008596.1) with bowtie2 (Langmead and Salzberg, 2012) v.2.0.0-beta7. Gene counts (based on the NC_008596.1 annotation) were obtained with htseq-count tool (https://htseq.readthedocs.io/en/release_0.11.1/count.html). To minimize count bias from overlapping genes, overlapping gene regions (except for nested genes) were excluded and the option ‘-m intersection-strict’ was used. Differential gene expression analysis was done with the DESeq package (Anders and Huber, 2010) v.1.12.1. Alignments were visualized in the IGV viewer(Thorvaldsdottir et al., 2013). The RNA-seq data is available under BioProject reference PRJNA735510.

To look for new IS1096 insertions, the transcriptome of each mutant was assembled de novo using SPAdes v.3.15.2 with the “--rna” option (Bushmanova et al., 2019). Transcripts containing the IS1096 sequence were identified using BLASTn and the IS1096′ flanking sequences were blasted against the mc2155 reference genome to infer their location. IS1096 insertion sites from the transcriptomes were compared to the known sites in the reference genome (Wang et al., 2008) to determine whether their location corresponded to a known lS1096 insertion site in the reference genome or constituted a new insertion site unique to the mutant.

Lipid extractions and (HP)TLC analyses

Bacteria were cultured to Log growth phase in 7H9-OADC, collected by centrifugation and washed twice with PBS. Apolar and polar lipid fractions were extracted from 50 mg of wet-weight bacterial pellet and the polar lipid fraction analysed by thin-layer liquid chromatography as previously described (Besra, 1998). The total extractable lipids were recovered from wet bacterial pellets by three successive extractions using distinct mixtures of CHCl3:CH3OH (1:2, 1:1 and 2:1, v/v). The fractions were pooled, washed with water and dried. For thin-layer chromatography (TLC) analysis, equivalent weights of total lipid fraction from each strain were spotted on silica gel 60 plates (Merck), which were developed either in CHCl3:CH3OH (9:1, v/v) or in CHCl3:CH3OH:H2O (65:25:4, v/v/v), as indicated. Compounds were revealed by spraying with 0.2% (w/v) anthrone in concentrated H2SO4 followed by heating at 100 °C.

The relative abundance of the different types of lipids from each strain was determined by loading a fixed amount (5 μg) of lipid mixture onto a high-performance TLC (HPTLC) silica gel 60 plate (Merck) with a Camag ATS4 apparatus. The plate was developed in appropriate organic solvent mixtures using a Camag ADC2 device and stained by immersion in 10% (w/v) CuSO4 (in H3PO4:CH3OH:H2O, 8:5:87, v/v/v) with a Camag CID3 apparatus, followed by heating at 150 °C for 20 min. Lipids were quantified by absorption measurement at 400 nm with a Camag Scanner 3 device using Wincats software.

MALDI-TOF MS and NMR spectroscopy

Matrix assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) and MS/MS analyses were performed in the positive ionization and reflectron mode, using the 5800 MALDI-TOF/TOF Analyzer (Applied Biosystems/ABsciex) equipped with a Nd:YAG laser (349 nm wavelength). MS and MS/MS spectra were acquired with a total of 2,500 shots at a fixed laser intensity of 4,000 (instrument-specific units) and 400-Hz pulse rate for MS, and a total of 5000 shots at a fixed laser intensity of 6000 (instrument-specific units) and 1000-Hz pulse rate for MS/MS. For MS/MS data acquisition, the fragmentation of selected precursor ions was performed at a collision energy of 1 kV using air as collision gas. Lipid samples were dissolved in chloroform and were directly spotted onto the target plate as 0.5 µl droplets, followed by the addition of 0.5 µl of matrix solution {10 mg/ml of 2,5-dihydroxybenzoic acid (Sigma-Aldrich) in CHCl3/CH3OH (1:1, v/v)}. Samples were allowed to crystallize at room temperature. MS data were acquired using the instrument default calibration.

CRediT authorship contribution statement

Estalina Báez-Ramírez: Conceptualization, Methodology, Formal analysis, Investigation, Visualization, Writing - review & editing. Luis Querales: Conceptualization, Methodology, Formal analysis, Investigation, Visualization. Carlos Andres Aranaga: Conceptualization, Methodology, Formal analysis, Investigation, Visualization, Writing - review & editing. Gustavo López: Conceptualization, Methodology, Formal analysis, Investigation, Visualization. Elba Guerrero: Conceptualization, Methodology, Formal analysis, Investigation, Visualization. Laurent Kremer: Conceptualization, Methodology, Resources, Supervision, Funding acquisition, Visualization, Supervision, Writing - review & editing. Séverine Carrère-Kremer: Methodology, Investigation. Albertus Viljoen: Methodology, Formal analysis, Investigation, Visualization, Writing - original draft, Writing - review & editing. Mamadou Daffé: Methodology, Formal analysis, Resources, Visualization, Supervision. Françoise Laval: Methodology, Formal analysis, Investigation, Visualization. Stewart Cole: Resources, Supervision, Funding acquisition, Writing - review & editing. Andrej Benjak: Methodology, Software, Formal analysis, Investigation, Visualization, Writing - original draft, Writing - review & editing. Pedro Alzari: Formal analysis, Funding acquisition, Writing - original draft, Writing - review & editing. Gwenaëlle André-Leroux: Methodology, Investigation. William R. Jacobs Jr.: Resources, Supervision, Funding acquisition, Writing - review & editing. Catherine Vilcheze: Methodology, Formal analysis, Investigation, Writing - review & editing. Howard E. Takiff: Conceptualization, Methodology, Formal analysis, Resources, Visualization, Supervision, Investigation, Data curation, Funding acquisition, Writing - original draft, Writing - review & editing, Project administration.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.