Introduction

Filamentous Actinobacteria are prolific producers of bioactive compounds. These metabolites are mostly used as weapons that provide protection against other microorganisms and phages in the environment (Abrudan et al. 2015; Kronheim et al. 2018; van der Heul et al. 2018). This is particularly useful for filamentous organisms, given that they generally lack the ability to make flagella for escaping dangerous situations. In addition, these bacteria are able to generate resistant spores that can invade new environments after their dispersal. Germination of spores leads to the formation of 1–2 germ tubes, which grow by tip extension, thereby establishing filamentous cells called hyphae. Branching of hyphae leads to the formation of a multinucleated vegetative mycelium, which forages and acquires nutrients by decomposing polymeric substances. Stressful conditions (such as nutrient depletion) induce program cell death (PCD) of the mycelium, which in turn triggers morphological and chemical differentiation (Claessen et al. 2014). This developmental transition leads to the formation of specialised hyphae that grow into the air, and the onset of production of a suite of bioactive compounds (Barka et al. 2016). Eventually, the aerial hyphae metamorphose into chains of grey-pigmented spores. Mutants that are unable to establish an aerial mycelium are called bald (bld), while those that are not capable to form spores are called white (whi) after their whitish color (Chater 2001; Claessen et al. 2006).

Genome mining has been instrumental for the revival of drug discovery (Bentley et al. 2002; Ichikawa et al. 2010). Many of the biosynthetic gene clusters that specify bioactive natural products are contained on giant linear plasmids (Chater and Kinashi 2007; Kinashi 2011; Medema et al. 2010; Mochizuki et al. 2003). Although linear replicons are rare in many bacterial taxa, they are common in Actinobacteria (Chen 1996; Kirby and Chen 2011). In fact, Streptomyces chromosomes (between 8 and 10 Mb in size) are also linear and typically comprise a “core region” containing the essential genes, and two variable “arms” with lengths ranging from 1.5 to 2.3 Mb (Hopwood 2006). Like linear plasmids, the linear chromosomes are capped by terminal proteins bound to the 5′ end of the DNA (Nindita et al. 2015). The chromosomal ends are genetically unstable, and readily undergo large (up to 2 Mb) DNA rearrangements. Such rearrangements can lead to circularization of the chromosome, exchange of chromosomal arms or the formation of hybrid chromosomes due to recombination between the linear plasmids and the chromosome (Hoff et al. 2018). This wide range of genomic rearrangements is believed to be caused by transposition or homologous recombination, occurring actively within the chromosome or between the chromosome and linear plasmids (Chen et al. 2002). Not surprisingly, these changes have profound effects on differentiation and specialised metabolite production (Leblond and Decaris 1994; Zhang et al. 2020).

Here we characterized genetic instability in Kitasatospora viridifaciens. This tetracycline producer was originally classified within the genus Streptomyces, it was recently shown to belong to the genus Kitasatospora (Girard et al. 2014). Protoplast formation and regeneration leads to the emergence of colonies that are no longer able to differentiate, which we attribute to the deletion of a 1.5 Mb segment of the right chromosomal arm and concomitant loss of most of the sequences contained on the large megaplasmid KVP1.

Methods

Strains and media

The strains used in this study (Table 1) are derivatives of K. viridifaciens DSM40239 (DSMZ). For protoplast preparation a spore suspension (106 spores ml−1) was grown for 48 h in a mixture of TSBS-YEME (1:1 v/v) supplemented with 5 mM MgCl2 and 0.5% glycine. Protoplasts were prepared as described (Kieser et al. 2000), with the difference that the lysozyme concentration was increased to 10 mg ml−1. Serial dilutions of protoplasts were plated on R5 (Kieser et al. 2000) or MYM medium (Stuttard 1982) at 30 °C. Regenerated protoplasts were streaked twice to single colonies on MYM before selecting the three independent bald colonies (B3.1, B3.2, and B3.3) that were further analyzed. Bald colonies were used as inoculum on liquid cultures of TSBS. Genomic DNA was isolated after 2 days of growth at 30 °C.

Table 1 Strains used in this study
Full size table

Whole genome sequencing and analysis

For genomic DNA isolation strains were grown in Tryptic Soy Broth medium containing 10% sucrose until mid-exponential phase. Next, chromosomal DNA was isolated as described previously (Kieser et al. 2000) and sequenced by BaseClear (Leiden, The Netherlands). Alignments of Illumina reads were performed using CLC Genomics Workbench 8.5.1. Raw Illumina (Hiseq 2500 system) sequences of the bald strain B3.1 were imported and mapped to the reference genome of K. viridifaciens DSM40239 (NCBI reference sequence: NZ_MPLE00000000.1) through the “Map reads to reference” function in the NGS core tools. Mismatch cost was set to 2 and non-specific matches were handled by mapping them randomly.

Pulsed-field gel electrophoresis

For Pulsed-Field Gel Electrophoresis (PFGE), 106 spores ml−1 of K. viridifaciens were inoculated in 25 ml of TSBS with 0.5% glycine or LPB (Ramijan et al. 2018). Cultures were grown at 30 °C agitating at 200 rpm for 16 and 40 h, respectively. Mycelial pellets were harvested by centrifugation at 4000 rpm for 15 min. The preparation of plugs for PFGE was performed as previously described (Zhang et al. 2020). Plugs were made with SeaKem Gold agarose (Lonza, Switzerland), and the genomic DNA in the plugs was cut with AseI. Plugs were run using a CHEF-DR II PFGE system (Biorad, USA). For efficient separation of fragments, samples were run in two conditions: a switching time of 60–125 s for 20 h, or a switching time of 2.2–75 s for 19 h, both at 200 V.

Quantitative real time PCR

Aliquots of 5 ng of DNA were used as a template in quantitative real time PCR. We used primers for both chromosomal (atpD and infB) and KVP1 (allC, tetR, parA and orf1) genes (Table 2). The PCR reactions were performed with the iTaq Universal SYBR Green Supermix Mix (Bio-Rad) using 5% DMSO, according to the manufacturer’s instructions. Reactions was performed in duplicate using a CFX96 Touch Real-Time PCR Detection System (Bio-Rad). To normalise the relative amount of DNA, the wild-type strain was used as a control, using the atpD gene as a reference.

Table 2 Primers used for quantitative real time PCR
Full size table

Results

Genomic characterisation of Kitasatospora viridifaciens

We previously sequenced K. viridifaciens and identified KVP1 as a novel megaplasmid (Ramijan et al. 2017). Analysis of the biosynthetic gene clusters (BGCs) using antiSMASH 5.0 (Blin et al. 2019) located 11 BGCs on KVP1 and 33 clusters on the chromosome (Fig. S1). One of the BGCs showed high homology to the BGC for chlortetracycline (Fig. S2). To test experimentally whether KVP1 is indeed a plasmid, we analysed genomic DNA of the wild-type strain with Pulsed-Field Gel Electrophoresis (PFGE). In the lane containing uncut DNA, a fragment with an estimated size between 1,600,000 and 2,200,000 bp (Fig. 1a, boxed region) was evident. This fragment is consistent with a genetic element that migrates independently of the chromosomal DNA. Digestion of the DNA with AseI revealed multiple DNA fragments, including two large fragments at 1,541,168 and 1,695,864 bp (see arrowheads in Fig. 1b). By further adjusting the switching time to 2.2–75 s, well-separated fragments with sizes ranging from 565,000 to 945,000 bp were identified (arrowheads Fig. 1c). Combining these different PFGE runs (Fig. 1d) allowed us to map the AseI fragment spectrum to the in silico genome assembly of K. viridifaciens (Fig. 1e–f). Altogether, these results confirmed that KVP1 is a megaplasmid and verified the predicted AseI sites in the chromosome.

Fig. 1
figure 1

KVP1 of K. viridifaciens is a megaplasmid. Pulsed-Field Gel Electrophoresis of genomic DNA of K. viridifaciens grown in TSBS (a, c) or LPB (b) medium. DNA was separated using switching times of 60–125 s for 20 h (a, b) or 2.2–75 s for 19 h (c). d The composite gel shows fragments larger than 1,125,000 bp (derived from the gel shown in panel b), and fragments smaller than 1,020,000 bp (derived from panel c). The solid rectangles indicate fragments derived from the chromosome, while the dashed rectangles indicate fragments derived from KVP1. Predicted in silico maps of the KVP1 megaplasmid and chromosome of K. viridifaciens are shown in panels (e) and (f), respectively

Full size image

Protoplast formation and regeneration leads to morphological diversity due to deletions and rearrangements in the chromosome and KVP1

The identification of KVP1 as a megaplasmid prompted us to analyse if it was distributed homogenously throughout the mycelium. For this, protoplasts were generated using a standard lysozyme-based protocol. Surprisingly, when protoplasts were allowed to regenerate on MYM agar plates, many colonies had developmental defects. Although the majority of colonies formed grey-pigmented spores (yellow circles Fig. 2a), a significant number of colonies was brown and failed to develop (red circles Fig. 2a). Sub-culturing of these so-called bald colonies (referred to as “b” in Fig. 2c) resulted in three morphological phenotypes, namely grey-pigmented colonies (B1), bald colonies (B2) or colonies with a variety of phenotypes (B3).

Fig. 2
figure 2

Protoplast formation and regeneration yield colonies with developmental defects (a) Protoplasts of K. viridifaciens regenerated on MYM medium yields grey-pigmented colonies (yellow dotted circles) and brown colonies (red dotted circles). The grey colonies retain their morphology in subsequent subcultures (b). c Subculturing of the brown colonies reveals morphological heterogeneity: some colonies appear grey-pigmented (B1), while others are bald (B2) or display a variety of phenotypes (B3). Bald colonies in B3 are indicated with black dashed circles

Full size image

To rule out that the MYM medium, which lacks osmoprotectant agents, was the main cause of the morphological differences, protoplasts were also allowed to regenerate on the more commonly used R5 medium (Kieser et al. 2000). None of the colonies that arose after regeneration of protoplasts were able to differentiate on R5 medium, which is typical of K. viridifaciens (Fig. S3A). To analyse this further, 149 colonies were randomly selected from R5 agar plates and subsequently streaked onto MYM agar (Fig. S3B). After 7 days of growth, 77% of the colonies had an apparent wild-type morphology and produced grey-pigmented spores, while 23% of the colonies were defective in development. This demonstrates that the observed morphologically heterogeneity detected after protoplast formation and regeneration is medium-independent.

The (partial) loss of megaplasmids is known to cause morphological defects similar to those observed here (Álvarez-Álvarez et al. 2014). To test if loss of KVP1 explains the change in phenotype, three B3-type bald colonies (encircled in Fig. 2c) were streaked onto MYM agar plates (Fig. 3a). All three colonies showed severe morphological defects and failed to produce spores even after 14 days of cultivation (Fig. 3a). Total DNA was then extracted from the three lineages (B3.1, B3.2, B3.3), and analysed for genes that served as markers for either the chromosome or for KVP1. Quantitative real-time PCR detected the chromosomal gene infB in the wild-type strain and the three tested lineages before the 20th cycle of amplification (Fig. 3b). Conversely, the allC gene located on the KVP1 plasmid was only detected in the wild-type strain (Fig. 3c). Similarly, we were unable to detect other KVP1-specific genes, namely orf1, parA or tetR (Fig. 3d). These results strongly suggested the loss of KVP1 in these colonies.

Fig. 3
figure 3

Bald colonies have lost the KVP1 megaplasmid. a Three independently isolated bald strains (B3.1, B3.2 and B3.3) are unable to sporulate on MYM medium, unlike the wild-type strain. Quantitative real-time PCR showed the presence of chromosomal gene infB in both the wild-type and bald strains (b). In contrast, the allC gene located on KVP1 is only present in the wild-type strain (c). d The relative abundance of four megaplasmid genes (orf1, parA, tetR, allC) in comparison to the abundance of the chromosomal gene infB suggest that KVP1 is lost in strains B3.1, B3.2 and B3.3

Full size image

To corroborate the loss of KVP1, next-generation sequencing of the total DNA of strain B3.1 was performed. While the wild-type strain showed the expected distribution of reads over genome and plasmid (3,882,545 and 1,112,768 respectively), the number of reads mapping to KVP1 was dramatically underrepresented in B3.1 (Fig. 4a, b, right panels with 5,005,998 and 143,387 reads for the chromosome and KVP1, respectively). These KVP1-mapped reads corresponded to 164,769 bp of the plasmid, mostly located on the terminal end of the right arm (black box in Fig. 4a, right column). Interestingly, the number of reads mapping to the right arm of the chromosome was also dramatically decreased in B3.1 (Fig. 4b right column, dashed red rectangle). A more detailed analysis indicated that most chromosomal sequences between 6,261,000 and 7,725,700 bp (Fig. 4c, arrow) were absent in B3.1. Apparently, strain B3.1 had not only lost the majority of sequences contained on KVP1, but also a major part of its right chromosomal arm. Further investigation of the deletion sites revealed an insertion sequence (IS) immediately adjacent to the chromosomal deletion start (around 6,261,000 bp) in B3.1 (Table 3). This IS contains the BOQ63_RS37135 gene encoding the transposase likely involved in moving this element. Furthermore, close inspection of KVP1 sequences still present in strain B3.1 also identified a flanking IS element containing the BOQ63_RS06880 transposase (Table 4). Altogether, these results demonstrate that major chromosomal and megaplasmid rearrangements and DNA loss occur during protoplast formation and regeneration, which is likely mediated via transposition events.

Fig. 4
figure 4

Whole genome sequencing reveals major chromosomal and megaplasmid deletions in strain B3.1. Alignment of Illumina reads of the wild type (left) and B3.1 strain (right) against KVP1 (a) and the chromosome (b). Please note the high coverage of KVP1 sequences (panel A) detected in the wild type (1,112,768 reads) compared to those of strain B3.1 (143,387 reads). Similarly, a high coverage in the right arm of the chromosome (dashed red rectangle) is observed for the wild-type (b, left panel) in comparison to strain B3.1 (b, right panel). c–e A more detailed characterisation reveals that all reads between 6,261,000 and 7,725,700 are absent in the right arm of the chromosome of strain B3.1

Full size image
Table 3 Transposition elements in the K. viridifaciens chromosome
Full size table
Table 4 Transposition elements in the right terminal region of KVP1
Full size table

Discussion

Actinobacterial genomes readily undergo rearrangements, whereby more than 1 Mb of genomic DNA can be lost (Hoff et al. 2018; Redenbach et al. 1993; Zhang et al. 2020). Here, we provide evidence that protoplast formation and regeneration in K. viridifaciens can lead to profound genomic rearrangements in the chromosome as well as loss of (large parts of) the megaplasmid KVP1. Given that these genomic rearrangements translate into major phenotypic variations, caution should be taken when using protoplasts for creating mutants, in particular when using strains that carry natural plasmids.

Filamentous Actinobacteria grow by tip extension and develop multinucleated mycelia. Little is known on how these bacteria regulate the abundance and spatial distribution of chromosomes and extrachromosomal plasmids within the mycelium. Maintaining large plasmids such as KVP1 is costly, given that such elements can comprise about a fifth of the entire genome. Megaplasmids are often reservoirs for biosynthetic gene clusters, and their interactions with the chromosome have been suggested to be a driving force for horizontal gene transfer (Kinashi 2011). Loss of such plasmids not only affects morphological development but may also influence the production of specialised metabolites whose gene clusters are contained on the chromosome. In Streptomyces hygroscopicus elimination of pSHJG1 increased the production of validamycin A (Lu et al. 2017), while holomycin yield was boosted when pSCL4 was lost in Streptomyces clavuligerus (Álvarez-Álvarez et al. 2014; Charusanti et al. 2012). Whether the loss of KVP1 has a similar effect on production of specialised metabolites in K. viridifaciens remains to be elucidated.

We observed that close to one fourth (23%) of the colonies derived from regenerated protoplasts of K. viridifaciens were defective in aerial growth and sporulation. Similar morphological defects have been described for S. clavuligerus, Streptomyces lividans and Streptomyces coelicolor upon loss of their plasmids (Álvarez-Álvarez et al. 2014; Hsu and Chen 2010). While the loss of KVP1 in K. viridifaciens may explain the arrest in morphological development, we here show that such plasmid-lacking derivatives can also carry other profound deletions in the chromosome, which could equally well contribute to this phenotype. By sequencing one revertant that had lost KVP1 we found that this strain had also lost approximately 1.5 Mb of the right arm of the chromosome. Such genetic instability is typical of streptomycetes, and can affect morphological differentiation, but also phenotypic traits associated with natural products, such as pigmentation, antibiotic biosynthesis and antibiotic resistance (Leblond and Decaris 1994; Zhang et al. 2020). Transposable elements were suggested as the principal cause of genetic instability (Lin and Chen 1997). The loss of KVP1 and the chromosomal deletions in the right arm could be the consequence of replicative transposition between the chromosome and the megaplasmid (Fig. 5). Notably, while most KVP1-located sequences were absent in the sequenced strain, including those required for autonomous replication of this megaplasmid, we identified a high coverage of sequences originally located at the end of the right arm of KVP1. This could be explained by an exchange between the right arms of KVP1 and the chromosome. A replicative transposition event between two linear replicons often results in the loss of chromosomal terminal regions and recombination of transposable elements (Chen et al. 2002). The presence of IS elements located at the chromosomal and plasmid termini could provide the basis for homologous recombination between these DNA molecules. It has been previously shown that homologous copies of IS elements could serve as substrates for the recombination machinery, creating chromosomal rearrangements in the genomes of Lactococcus lactis and Escherichia coli (Daveran-Mingot et al. 1998; Gaffé et al. 2011). In the case of colony B3.1 that had arisen from protoplast regeneration, our results suggest that a possible cause of the genomic rearrangement was the replicative transposition of an IS, exerted by the BOQ63_RS37135 transposase (black arrow in Fig. 5b). Following replicative transposition, a double-stranded break occurs at the site of transposon excision. This break is repaired by recombination with homologous genes located on IS elements present in the right arm of KVP1 (shown as a grey arrow in the KVP1 right arm), which are abundantly present (see Table 4, genes highlighted with **). This recombination might force the interchange of terminal arms. This hypothetical model would explain the genome size reduction of B3.1 (6,787,546 bp).

Fig. 5
figure 5

Proposed model for the genomic rearrangements and deletions identified in strain B3.1. Replicative transposition of a transposable element (black arrow) located on the chromosome creates a double-stranded break (DSB). Subsequent repair of the DSB by homologous recombination between the chromosome and KVP1 plasmid (carrying numerous transposable elements), leads to an exchange of the right arms of both replicons and changes in their size. Please note that the larger KVP1 megaplasmid variant is lost in strain B3.1

Full size image

We previously identified several genes required for plasmid replication and segregation on KVP1 (Ramijan et al. 2017). Interestingly, bioinformatics analysis indicated that the conserved tpg gene, which encodes the terminal protein required for capping of the telomeres and thereby contributing to its protection, is absent in K. viridifaciens. Instead, we identified homologues of the Tac-Tpc system, encoded by the adjacent BOQ63_42425 and BOQ63_42430 genes, respectively. The tac/tpc genes were first identified in the linear Streptomyces plasmid SCP1 (Huang et al. 2007). In K. viridifaciens, the tac-tpc genes are located on the right arm of the chromosome and no copies are found on KVP1. This implies that the encoded proteins are required for maintenance of both the chromosome and plasmid telomeres. The use of the Tac-Tpc system by K. viridifaciens also indicates that, like in streptomycetes, different systems are used for maintenance of telomeres in Kitasatospora species, given that Kitasatospora setae contains the tpg gene (Ichikawa et al. 2010).

The frequency of aberrant phenotypes after protoplast regeneration is higher than the phenotypic heterogeneity obtained after outgrowth of spores, which typically is in the order of 1% (Gravius et al. 1993; Hoff et al. 2018; Leblond et al. 1989; Lin and Chen 1997; Redenbach et al. 1993; Zhang et al. 2020). An explanation for the high frequency of aberrant mutants in colonies arising after protoplasting may relate to the activation of transposable elements contained in the terminal regions of the chromosome and/or the KVP1 plasmid. The activation of transposases are typically stimulated by stressful conditions, such as radiation, oxidative stress, temperature or inhibitory concentrations of metals and antibiotics (Vandecraen et al. 2017). It was recently demonstrated that elevated levels of osmolytes induces hyperosmotic stress (Fuchino et al. 2017; Ramijan et al. 2018), which are conditions that are also used during preparation of protoplasts. This stress could also stimulate transposition events and consequent chromosomal rearrangements. Consistent with this idea is that other cell wall-deficient cells, called L-forms, which have likewise been exposed to osmotic stress conditions, carry chromosomal deletions. In this context it is interesting to note that in three independent L-form lineages of K. viridifaciens deletions in the right chromosomal arms were found in addition to loss of KVP1 (Ramijan et al. 2018). These three strains retained a similar region of KVP1 in their genomes, with a size of 164,773 bp for alpha and M1, and 164,642 bp for M2. These are very similar to the KVP1-sequences remaining (164,769 bp) in the bald protoplast regenerant B3.1 in terms of length and content. Notably, in strain B3.1 as well as the three independent L-form mutants the BOQ63_42425 and BOQ63_42430 genes are still found. Given that these genes are likely involved in telomere maintenance (see above), their presence could suggest that the chromosomes remain linear after these dramatic recombination steps. However, this awaits further experimental evidence.

Chromosomal rearrangements are often detrimental for the fitness of a unicellular organism. However, it was recently shown that in Streptomyces chromosomal rearrangements may increase the diversity and production of specialized metabolites, including antibiotics (Zhang et al. 2020). A division of labour strategy would allow a colony to have a mixture of mutant and wild-type chromosomes, where the mutant cells are virtually sterile and become specialized in the production of antibiotics, while the cells containing wild-type chromosomes are efficient spore producers (Zhang et al. 2020). Thus, while some genetic variation may naturally exist within the mycelium, we expect that exposure to high levels of osmolytes, associated with growth and subsequent protoplast formation, generates stress and dramatically increases chromosomal changes. This study provides a starting point to further characterise these changes and to investigate their consequences, which may lead to exciting new insights into the biology of these prolific antibiotic producers.