PI factor, a novel type quorum-sensing inducer elicits pimaricin production in Streptomyces natalensis

A chemically novel autoinducer (PI factor) has been purified from cultures of the pimaricin producer Streptomyces natalensis ATCC27448. The chemical structure of the PI molecule was identified as 2,3-diamino -2,3-bis (hydroxymethyl) -1,4-butanediol. Pimaricin biosynthesis in S. natalensis npi 287, a mutant impaired in pimaricin production, was restored by supplementation with either A-factor from Streptomyces griseus IFO13350 or with PI factor. S. natalensis did not synthesize A-factor. The PI autoinducer was active at very low concentrations (50 to 350 nM). A threshold level of 50 nM  was required to observe the induction effect. The dose-response curve was typical of a “quorum sensing” type mechanism. The biosynthesis of PI factor was associated with cell growth of S. natalensis, both in defined and complex media. Supplementation of the wild type S. natalensis with pure PI (300 nM) resulted in a stimulation of 33 % of the production of pimaricin. These results indicate that the endogenous synthesis of PI factor is limiting for pimaricin biosynthesis in the wild type strain. This water-soluble PI factor belongs to a novel class of autoinducers in Streptomyces species different from the classical butyrolactone autoinducers. Since restoration of pimaricin production in the npi 287 mutants is conferred by both A-factor and PI factor, S. natalensis appears to be able to integrate different quorum signals from actinomycetes.

3 Pimaricin (Fig. 1) represents a prototype molecule of glycosylated polyenes [5] important for antifungal therapy and promising for its antiviral activity, stimulation of the immune response, and action in synergy with other antifungal drugs or antitumor compounds [10]. Pimaricin is produced by Streptomyces natalensis and it is widely used in the food industry to prevent mold contamination of cheese and other non-sterile foods (i.e. cured meats). Initial studies showed that the synthesis of this 26-membered macrolide tetraene in S. natalensis requires a complex polyketide synthase (PKS) [11]. The complete sequence and analysis of the PKS genes responsible for the biosynthesis of pimaricin has been reported.
The sequenced region of 85 kb encodes 13 PKS modules within five multifunctional enzymes, and 12 additional proteins that presumably catalyze post-PKS modification of the polyketide skeleton (tailoring enzymes), export and regulation of gene expression [12,13,14].
In an effort to understand the regulation of the commercially important antifungal pimaricin, it was of great interest to isolate and characterize the autoregulatory molecule of pimaricin biosynthesis in S. natalensis.
by guest on March 23, 2020 http://www.jbc.org/ Downloaded from 4 We describe in this article the isolation of a class of mutants that is defective in pimaricin production because it lacked a pimaricin inducer. Complementation tests of this mutant and purification of the inducer molecule led to the identification of a novel hydrophilic pimaricin inducer (PI) molecule.

Strains
The wild type Streptomyces natalensis ATCC 27448 was used as the parental strain to isolate different npi mutants. S. natalensis cultures were maintained on solid TBO sporulation medium (containing per liter tomato paste 20 g; oat meal 25 g; agar 25 g) as described previously [12]. Candida utilis (syn. Pichia jadinii) CECT 1061 was used as test strain in the bioassay of the antifungal activity of pimaricin.
Cultures of Streptomyces griseus IFO 13350 (formerly described as S. bikiniensis IFO 13350; S. Horinouchi, personal communication) were used to produce A-factor in YMPG medium [28]. S. griseus HH1 (an A factor negative strain) was used to quantify the A-factor inducing activity. lawn of C. utilis. Mutants that did not produce inhibition zones were selected. The lack of production of pimaricin of the selected mutants was confirmed in liquid cultures in SPG medium [29]. Mutants that did not revert in liquid medium cultures were further analyzed by complementation tests.

Culture media and quantification of pimaricin production in liquid cultures.
Four different complex media were used for quantification of the PI factor concentration and its relationship to pimaricin production. These include: NBG medium [nutrient broth (Oxoid) supplemented with glucose (5 g/l)]; YEME medium (yeast extract 3 g/l; peptone 5 g/l; malt extract 3 g/l and glucose 10 g/l); TSB medium (Difco) and YED medium (yeast extract 10 g/l; glucose 10 g/l). In addition two defined media were also used to quantify the inducer production: Streptomyces MM [30] and Lechevalier defined medium [31].
The production of pimaricin in liquid cultures was routinely quantified by spectrophotometric determination at 319 nm. A 0.5 ml aliquot of the culture was extracted with 5 ml of methanol and diluted with distilled water; the concentration of pimaricin was determined as described previously [12] using a pure sample of pimaricin (Sigma Chem. Co) as standard.

Complementation tests
Complementation tests were performed between pairs of the 35 stable non-producer mutants using standard co-synthesis methods in solid YED medium. Each pair of npi mutants were grown as lawn cultures. Agar plugs were taken out from each of the growth zones and the production of pimaricin was bioassayed using Candida utilis as sensitive organism.
Positive complementation was clearly detected by the production of pimaricin when the two non-producer mutants were placed close to each other, whereas control plugs from each of the two non-producer mutants gave no inhibition zone when assayed separately (Fig. 1).

Extraction and HPLC purification of the PI factor to homogeneity
The culture broth (15 liters) from S. natalensis wild type strain grown for 24 h in YED medium in a Braun Biostat C fermentor was concentrated 10-fold in a vacuum evaporator.
The The PI factor was purified by reverse phase HPLC using a Waters 600 unit coupled to a PDA 996 detector equipped with a Polarity C18 column (3.9 x 150 mm; particle size, 5 µm).

Determination of PI factor biological activity
The biological activity of PI factor was determined by its ability to induce pimaricin production by mutant strain S. natalensis npi287 in solid SPG medium. After allowing growth of S. natalensis npi287 for 2 days at 30 ºC, samples of culture broths (100 µl) or of different fractions from the PI factor purification process were added to wells (7 mm diameter) in the agar layer. The plate was then overlayed with a culture of C. utilis in soft agar and incubated for 24 h at 28 ºC. The diameter of the C. utilis inhibition zone after induction of pimaricin in strain npi287 was proportional to the amount of PI factor in the sample.

Acetylation of PI factor
Acetylation of PI factor (0.1 mg) was made with a mixture of acetic acid and acetic anhydride (4 ml each) for 24 h at room temperature in the dark. The resulting product was lyophilized and the acetylated PI was dissolved in dichloromethane. For the deacetylation process acetyl-PI was hydrolyzed by drop-wise addition of concentrated NaOH (5 g) in methanol (5 ml) and the conversion of acetyl-PI to PI was followed by TLC.

NMR Spectroscopy
The structure elucidation of PI factor was established by NMR spectroscopy, using a combination of 1D The ES+ mass spectrum was recorded on a HP 1100-MSD using CF 3 COOH 0.1% as the source of ionization.

Classes of S. natalensis mutants impaired in pimaricin biosynthesis and cross-feeding studies
A total of 384 non-producer mutants impaired in pimaricin (npi) biosynthesis (npi-1 to npi-384) were isolated after NTG mutagenesis as described in "Experimental Procedures".
Some of them reverted or were unstable. After several rounds of selection 35 stable npi mutants were selected (Table 1) and assayed in pair-wise complementation tests on solid YED medium (Fig. 1). Based on the results of the complementation tests, the mutants were divided into 11 classes (A to K in Table 1).
A group of seven mutants (npi-6, npi-12, npi-54, npi-64, npi-86, npi-98 and npi-137) (class H in Table 1) were unable to complement any other mutant class or vice versa, probably because these mutants were blocked in one of the pimaricinolide synthases [11,12] and the large intermediates that these enzymes produce could not diffuse out of the cells.
Non-producer mutants of classes A, B, C, F, and J were all able to complement npi287 (class G) ( Table 1). The results of these complementation tests indicated that npi mutants of classes A, B, C, F and J were able to produce a substance that complements pimaricin production in npi287 and that these mutants were blocked later in the biosynthetic inducer-requiring class G (see below) and, therefore, may also contain mutations related to the inducer biosynthesis.
Mutant npi287 recovers normal pimaricin production when supplemented with S. natalensis wild type culture broth or with A-factor from S. griseus Since initial studies indicated that mutant npi287 recovered pimaricin production in co-synthesis experiments with different S. natalensis mutant strains we tested its complementation with spent culture broths of the parental strain S. natalensis ATCC 27448.
Results showed that mutant npi287 recovered full pimaricin production levels when supplemented with culture broths of the S. natalensis wild type strain grown for 24 h in either YED, NB or YEME media suggesting that the inducer was secreted by the wild type strain.
Our first working hypothesis was that the pimaricin inducer might belong to the A-factor butyrolactone family since such class of inducing compounds are common among Streptomyces species [15,34]. To test this hypothesis mutant npi287 was supplemented with increasing concentrations of HPLC-purified A-factor from S. griseus IFO 13350. As shown in To discern if the pimaricin inducer was a butyrolactone or a different molecule we followed initially the purification procedure described for the virginia butanolides from Streptomyces virginiae [18]. The procedure is based on extraction of the butyrolactones with ethyl acetate under acidic pH conditions (pH 2.0) followed by concentration of the organic phase, and application of the concentrate through an active carbon column followed by stepwise elution with 50% methanol in water (v/v) or 100% methanol. As a control, A-factor was purified from cultures of Streptomyces griseus IFO 13350. As expected, A-factor was purified using this protocol and eluted in the 100% methanol fraction.
Surprisingly, in contrast to what occurs with butyrolactones, a significant proportion of the pimaricin inducing compound remained in the aqueous phase after ethylacetate extraction at either acidic, neutral or slightly basic (pH 7.5) pH values, suggesting that the inducer was a hydrophilic molecule.

The purified PI factor is different from A-factor
The PI factor was purified as indicated in Experimental Procedures. After elution from the Sephadex G10 column the biologically active fractions were further purified by reverse phase HPLC chromatography. Aliquots were derivatized with FMOC for an easy detection (Fig. 3). The pure PI compound was used for mass spectrometry and NMR analyses.
The S. griseus IFO13350 A-factor was purified by following its biological activity on the test strain S. griseus HH1, a mutant lacking streptomycin production due to its deficiency in A-factor biosynthesis. As indicated above, the purified S. griseus A-factor containing fraction elicited pimaricin production in the S. natalensis npi287 mutant (Fig. 2) as also did the crude culture broth of the streptomycin-producing parental strain of S. griseus IFO13350.
These results indicated that pimaricin production by our npi287 strain responded to A-factor.
Interestingly, the opposite tests were negative. Neither the wild type S. natalensis culture broth nor the HPLC-pure PI factor restored streptomycin production to the S. griseus HH1 mutant, further indicating that the PI factor was different from A-factor and specific for S. natalensis. In contrast to the well-known stimulation of sporulation of S. griseus HH1 exerted by A-factor, the pure PI factor did not stimulate sporulation of the wild type S. natalensis or the npi287 mutant.
These results, together with the capability of the PI factor (but not of A-factor) to react with FMOC, clearly supported the proposal that the nature of both compounds is different.
The ability of the PI factor to react with FMOC suggested the presence of amino groups in its To study whether any traces of PI factor might be produced by the S. griseus strain, comparative HPLC analysis of the pure PI and the A-factor containing fraction were performed. Results (Fig. 3) showed that the active PI factor derivatized with FMOC eluted at 11.0 min (Fig 3A and C), whereas no peak could be observed at 11.0 min in the A-factor chromatogram after FMOC derivatization (Fig. 3B).

Chemical structure of the PI factor
The 1 H NMR spectrum of PI Factor (Fig. 4) showed only a signal at δ 3.71 ppm as a singlet whereas its 13 C NMR spectrum contained two signals, which were assigned to one methylene group (CH 2 , δ 58.76 ppm) and one quaternary carbon atom at δ 60.87 ppm. This was confirmed by means of a DEPT study. Assignment of the carbon atom of the methylene group was carried out by using a HMQC spectrum, which showed a correlation peak via 1 J H,C with the methylene protons at δ 3.71 ppm. In the HMBC spectrum one key correlation peak via 3 J C,H was obtained between the methylene protons and the quaternary carbon atom at δ 60.87 ppm.
The low-field nature of the chemical shifts (δ H and δ C ) of the methylene group suggested the presence of an oxygenated group (-CH 2 OH). This was easily confirmed by comparison with the NMR spectra of the acetylated derivative PIa (Fig. 4). The 1 H NMR spectrum of the acetylated derivative (recorded in CDCl 3 ) showed two singlets at δ 1.25 and 4.43 ppm, which were attributed to a methyl group (CH 3 COO-) and a methylene group (CH 3 COOCH 2 -), respectively; and the 13 CNMR spectrum showed four signals at δ 20.70 ppm (CH 3 ), δ 58.10 ppm (C), δ 62.71 ppm (CH 2 ) and δ 170.60 ppm (CO). 12 The downfield shift of the quaternary carbon atom and the variation of the chemical shifts by the pH change (see Table 2), suggested the presence of an amine group. This was confirmed by means of the mass spectrum. The PI compound gave an ion at m/z 91 [M+2H] + /2 on the positive electrospray (ES+) indicating a double-charged species. On the basis of all available data the structure of compound PI is proposed to be 2,3-diamino-2,3bis(hydroxymethyl)-1,4-butanediol (Fig. 4).

Dose response: The PI factor works at low concentrations
The availability of the pure PI compound allowed us to quantify its inducing effect using the standard inducer assay with the npi287 strain (Fig. 5). The npi287 strain clearly responded to PI concentrations of 100 nM and the pimaricin production showed a linear response up to 350 nM. The inducer increased the diameter of the pimaricin inhibition zone up to a concentration of 350 nM and the assay was saturated at concentrations above 400 nM.
A threshold level of PI of about 50 nM was always required to detect the induction of pimaricin production. These results are consistent with a cooperative effect typical of a "quorum-sensing" type mechanism.
Following addition of PI factor (200 ng/ml) to npi287 mutant cultures, most of the PI factor was taken up by the cells and only 10 ng/ml remained in the culture broth 24 hours after addition. The PI level in the broth increased to 22 and 15 ng/ml at 48 h and 72, respectively, suggesting that a residual level is released after initial binding to the cells. The production of pimaricin by the parental strain S. natalensis ATCC 27448 in NBG, TSB and YEME media was stimulated by the addition of 300 nM PI factor in the three culture media. Since the wild type strain synthesizes endogenous PI factor, the stimulatory effect of exogenous inducer indicates that its biosynthesis is limiting for pimaricin production in the wild type. Furthermore the results indicate that PI factor is taken up by S. natalensis or at least it triggers at the membrane level a signal cascade leading to overproduction of pimaricin.

Kinetics of formation of PI factor in cultures of S. natalensis in defined and complex media
The time of synthesis of the PI factor is relevant to trigger the onset of pimaricin biosynthesis. Furthermore, the level of inducer formed may be limiting for total pimaricin accumulation. To analyze the time-course of PI factor biosynthesis, S. natalensis ATCC 27448 was grown in two defined media, namely MM for Streptomyces [30] and Lechevalier medium [31], and in four complex media TSB, NBG, YED and YEME (see Experimental Procedures) that are known to support high pimaricin production.
The two defined medium supported low yields of pimaricin (not shown). In both Streptomyces MM and Lechevalier media, PI factor was synthesized during growth and a peak (8 to 10 ng/ml) was observed at the end of rapid growth phase of Streptomyces natalensis (equivalent to the exponential growth phase of unicellular bacteria). Biosynthesis of pimaricin in these two defined media was parallel to the growth of the culture with a delay of about 12 h with respect to the inducer formation. These results indicate that the synthesis of PI is related to the growth phase. All complex media were found to support much higher levels of pimaricin production ranging from 620 µg/ml in TSB medium to 920 µg/ml in YEME medium (Fig. 6). In the four complex media the PI factor was formed earlier than pimaricin coinciding with the end of the rapid growth phase and reached levels of PI between 25 ng/ml in YEME and NBG and 45 ng/ml in TSB media that were clearly higher (about three-to four-fold) than the level of inducer in defined media. A special mention deserves the YED medium in which PI factor accumulated to levels of 140 ng/ml. Production of pimaricin in this medium did not match the high level of PI factor observed. These results indicate that biosynthetic steps other than the level of PI factor are limiting for pimaricin production above PI saturation levels in this medium.
To confirm that the PI inducer is produced during growth of S. natalensis cultures and is not present in the complex media used in these experiments all media were tested by HPLC before inoculation and after 48 h of incubation with S. natalensis. Results are shown in Fig. 7 for the NBG medium. There was no PI inducer in the complex medium before inoculation and it clearly accumulated after 48 h of incubation in this medium. To confirm the absence of PI, the NBG medium was primed with pure PI inducer. As shown in Fig. 7B, the HPLC elution profile of the PI-supplemented culture medium confirmed the lack of this molecule in the starting culture medium.
Strain specificity of autoinducers is common among Streptomyces species due to the formation of either different inducer molecules [18] or to differences in their stereochemistry  [37]. Several of these molecules including A-factor, the virginia butanolides and related compounds belong to the butyrolactone family [15,18,20]. The butyrolactones appear to occur in many but not all Streptomyces strains [34]. However other types of autoinducers occur in some actinomycetes [23,24]. A recent report described a novel boron-containing autoinducer [25] and it seems likely that novel types of "quorum sensing" effectors will be discovered.
As reported in this article a novel type of autoinducer has been found in the pimaricin producer S. natalensis. This autoinducer complements the non producer S. natalensis class G mutant. The molecule 2,3-diamino-2,3-bis(hydroxymethyl)-1,4-butanediol (PI factor) is symmetrical and has not been reported before in the microbial world. The tetraacetylated derivative of PI showed no inducing ability indicating that the hydroxyl groups of this molecule are strictly required for interaction with its receptor protein. The PI inducer has an entirely novel chemical structure, that is only distantly related to the homoserine lactone and the furanosyl diester [25] inducer families.
It is interesting that class G mutant npi287 also recovers pimaricin production when complemented with A-factor, suggesting that there is a mechanism of signal integration that switches gene expression in response to either PI factor or A-factor. However we have shown that S. natalensis does not produce A-factor. It will be interesting to know if the PI factor induction is mediated by a butyrolactone-type receptor [38,39,40]  The dose-response curve of pimaricin to increasing concentrations of PI factor shows a sigmoidal curve with a minimal threshold level. When PI factor was added to cultures of the npi287 mutant unable to produce it, PI was initially taken up by the cells and a small amount was then released into the culture broth. These results clearly indicate that the PI inducer shows a standard "quorum sensing" type of kinetics: the PI molecule is, therefore, secreted and signals to other cells that they must start to produce pimaricin in response to biomass accumulation, nutrient limitation or other environmental conditions. This "quorum sensing" type of response requires low concentrations of inducer [41], usually in the nanomolar range, as occurs with the PI factor.
An important question is whether higher levels of inducer increase further pimaricin biosynthesis. As shown in this article addition of 300 nM PI factor to cultures of the pimaricin producer wild type S. natalensis strain increased pimaricin production between 20 and 40% in different media. These results suggest that under standard culture conditions, the concentration of extracellullar PI factor in complex media is still limiting for pimaricin production. Therefore metabolic engineering of the PI factor biosynthesis is a subject of great interest for improving pimaricin production.
The characterization of the PI inducer opens the possibility of genetic modification of its biosynthesis to understand the molecular mechanism of signal transduction.