1. Introduction
Secondary metabolites (SMs), also referred to as natural products or specialized metabolites, constitute an important group of biotechnological products. The relevance of SMs is primarily of a pharmaceutical nature, as it is not uncommon for SMs to display bioactivities relevant in the context of drug discovery initiatives [
1,
2]. Examples of pharmaceutically relevant SMs include penicillin, an antibiotic discovered by Alexander Fleming, and lovastatin, a cholesterol-lowering drug. In relation to public health security, it is vital to address emerging threats such as the antibiotic resistance of pathogens and the risk of pandemic outbreaks. Unsurprisingly, the need for novel medicines remains unabated [
3,
4]. While it is now possible to generate chemical libraries of potential drug leads by employing combinatorial approaches, the screening of remarkably rich and largely unexplored biosynthetic repertoires of microorganisms still provides an attractive alternative to synthetic methods. The structural complexity of natural products and the difficulty of designing economically viable synthetic routes to synthesize these molecules are crucial points to consider in drug discovery and development [
5,
6,
7]. On the other hand, uncovering the full biosynthetic potential of microorganisms is not a straightforward task. The production of SMs is strictly dependent on environmental conditions and, in contrast to the group of primary metabolites, is not indispensable in terms of sustaining life. To reveal the full SM catalog of a given microbial strain, one may resort to genetic manipulations or follow the bioprocess-based strategy of using unconventional methods of cultivation [
8,
9]. Microbial co-cultures serve this purpose ideally, as they set up the conditions for interspecies interactions that may ultimately lead to the awakening of metabolic pathways that are inactive in monocultures [
10,
11,
12,
13,
14,
15]. For some of the biosynthetic pathways, the stimulatory or inhibitory effects may be observable under the co-cultivation condition, as reflected, respectively, by the increased or decreased levels of SMs in the medium. Notably, the transition from mono- to co-cultivation does not require additional equipment costs unless a dedicated (e.g., membrane-based) system is required to separate the proliferating strains. Such efforts are undertaken to determine if the physical contact of the strains is necessary for SM induction. In other cases, however, the co-cultures can be successfully propagated on agar plates or in conventional laboratory flasks and bioreactors [
16,
17,
18]. It is important to emphasize the fact that the challenges of microbial co-cultivation are mainly associated with experimental design since the requirements and characteristics of all employed strains must be taken into consideration. One of the unwanted scenarios occurs when a faster-growing microbe exerts its dominance over a slower-growing partner and practically eliminates it from the co-culture [
19]. As a result, the SMs of only one strain are detected in the medium, and, depending on the research goals, the purpose of performing such co-cultivation experiments may be seen as questionable. To circumvent this problem, one may adjust the inoculum ratio or delay the inoculation of the faster-growing microorganism to allow the “slow grower” to develop its biomass prior to the moment of confrontation. The problem is that the bioreactor-based production of SMs in microbial co-cultures is greatly underexplored; the stirred tank bioreactor-based studies on SM production are still scarce [
20,
21,
22,
23], while the majority of co-cultures are investigated with the use of shake flasks. It seems surprising since the industrial production of biotechnologically relevant microbial SMs is mostly performed in stirred tank bioreactor systems. This topic is directly addressed in the present work, which is the first investigation of the stirred tank bioreactor co-cultures of
Penicillium rubens and
Streptomyces noursei, two potent and biotechnologically relevant producers of SMs. The filamentous fungus
P. rubens is mostly known for its ability to biosynthesize penicillin [
24], the first antibiotic produced on an industrial scale, while
S. noursei is an actinomycete equipped with the biosynthetic pathways leading to nystatin A1 [
25], a commonly prescribed antifungal drug. Furthermore, the repertoires of SMs exhibited by these two species encompass a broad array of molecules, including cycloheximide [
26] and chrysogine [
27], the molecules generated by
S. noursei and
P. rubens, respectively. As with any submerged co-cultivation process aimed at SM biosynthesis, it is of fundamental importance to study the impact of the co-culture initiation method on the outcomes of the bioprocess. Specifically, it is crucial to investigate the SM-related effects of the simultaneous inoculation of two strains as well as the delayed inoculation of the faster-growing strain. Moreover, the comparative analysis involving the co-cultures and the corresponding monocultures provides valuable information regarding the stimulatory or inhibitory effects associated with co-cultivation. Finally, it should be emphasized that each pair of co-cultivated microbial strains constitutes a unique biological system whose behavior cannot be predicted on entirely theoretical grounds. Hence, the performance of the co-culture in relation to the chosen inoculation routine must be investigated experimentally.
The aim of the study was to characterize the production of SMs in the stirred tank bioreactor co-cultures of P. rubens and S. noursei. The co-cultures initiated by performing the simultaneous inoculation of P. rubens and S. noursei were compared with the ones started by employing the delayed inoculation of S. noursei relative to P. rubens. All the investigated co-cultures were compared with their monoculture counterparts.
2. Materials and Methods
2.1. Strains
The strains Penicillium rubens ATCC 28089 and Streptomyces noursei ATCC 11455, purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA), were used in the study. The strains were maintained on agar slants, as recommended by the ATCC.
2.2. Cultivation Medium
The following liquid medium composition was used: glucose, 10 g L−1; lactose, 40 g L−1; yeast extract, 10 g L−1; KH2PO4, 1.51 g L−1; MgSO4·7H2O, 0.5 g L−1; NaCl, 0.4 g L−1; ZnSO4·7H2O, 1 g L−1; Fe(NO3)3·9H2O, 2 g L−1; biotin, 0.04 mg L−1; and phenylacetic acid, 0.25 g L−1. The following solution, added at 1 mL L−1, was employed as a source of trace metals: MnSO4, 50 mg L−1; Na2B4O7·10H2O, 100 mg L−1; CuSO4·5H2O, 250 mg L−1; and Na2MoO4·H2O, 50 mg L−1. The pH of the medium was adjusted to 6.5 with a 0.4 M solution of sodium and potassium carbonates prior to autoclaving at 121 °C.
2.3. Cultivation Conditions
The BIOSTAT® B stirred tank bioreactors (Sartorius, Goettingen, Germany) with a working volume of 5.5 L were employed in the study. The following experimental runs were performed: PRSN1, PRSN2, and PRSN3. In each experimental run, three bioreactors were used in parallel because the co-culture was always accompanied by the monocultures of P. rubens and S. noursei. The dissolved oxygen (DO) concentration was not controlled. The following aeration profile was used: the airflow rate was set at 1.5 L min−1 in the first 24 h of the run and was increased to 5 L min−1 until the end of the process, while the stirring speed was constant at 300 min−1. All experiments were performed in triplicate, and the standard deviation values were calculated in OriginPro 2017 software (OriginLab, Version b9.4.1.354 SR1, Northampton, MA, USA).
2.4. Co-Culture Initiation Methods
The spores of P. rubens were obtained through cultivation on a solid medium prepared by adding glucose (20 g L−1) and agar (20 g L−1) to potato broth (300 g of potatoes boiled in 500 mL of water). The ISP Medium 2 (BD, Franklin Lakes, NJ, USA), used according to the manufacturer’s instructions, was employed for the preparation of S. noursei spores. The spore suspension, obtained by removing the spores from agar slants with the use of a sterile pipette, was applied for the inoculation of sterile production medium in the bioreactor (150 mL for the monoculture, 150 + 150 = 300 mL for the co-culture). Following the completion of the inoculation procedure, the working volume in all bioreactors (i.e., mono- and co-cultures) was always equal to 5.5 L. The inoculation procedure was adjusted to reach the final spore concentration in the bioreactor equal to (1.0 ± 0.1) × 109 spores per liter. The following inoculation scheme was applied in the PRSN1, PRSN2, and PRSN3 co-cultures:
PRSN1: simultaneous inoculation of S. noursei and P. rubens;
PRSN2: S. noursei inoculation delayed by 24 h relative to P. rubens;
PRSN3: S. noursei inoculation delayed by 48 h relative to P. rubens.
2.5. Analytical Procedures
Liquid samples were collected from each bioreactor at 24 h intervals. The biomass was removed through filtration with the use of Munktell filter discs (grade 389.84 g m−2, diameter 150 mm). The liquid samples were frozen at −20 °C.
For the analysis of SMs in the liquid samples, ultra-high-performance liquid chromatography (AQUITY-UPLC
®) coupled with high-resolution mass spectrometry (ACQUITY–SYNAPT G2, Waters, Milford, MA, USA) was employed, as described in detail in [
23]. The column BEH Shield RP18 (reverse-phase), 2.1 mm × 100 mm × 1.7 μm, was used for SM analysis. For the analysis of glucose and lactose, the BEH Amide column (normal phase), 2.1 mm × 150 mm × 1.7 μm, coupled with an evaporative light scattering detector (Waters, Milford, MA, USA) was employed, as described in detail in [
23].
The identification of metabolites was based on the comparison of the experimental
m/
z values (at ESI
− mode) with the database records of The Natural Product Atlas [
28]. The identities and concentrations of penicillin G and nystatin A1 were determined using the analytical standards purchased from Sigma-Aldrich (Burlington, MA, USA). In the cases of the remaining SMs, for which the standards were not available, the identity was putatively assigned based upon the
m/
z similarity to the previously described
Streptomyces products (with the absolute error Δ
m/
z always below the value of 0.01) included in The Natural Product Atlas [
28], and the peak area representing the [M−H]
− ion was considered for semi-quantitative comparisons.
The volumetric uptake rates of glucose (rGLU) and lactose (rLAC) were obtained by approximating the concentration values with cubic b-spline function and differentiating the curves in time. PTC Mathcad 15 (PTC, Version 15.0, Boston, MA, USA) software was applied for this purpose.
The light microscope OLYMPUS BX53 (Olympus Corporation, Tokyo, Japan) with the software OLYMPUS cellSens Dimension Desktop 1.16 (Olympus Corporation, Version 1.16, Tokyo, Japan) was employed for the microscopic observations.
4. Discussion
The study represents a multi-angle investigation of stirred tank bioreactor co-cultures involving
S. noursei and
P. rubens. What emerges from the experimental datasets is a description of three distinct co-cultivation scenarios. It is convenient to start the discussion by considering the similarities between the DO profiles recorded for the mono- and co-cultures (
Figure 3). A similar pattern was observed in all three experimental runs for the monocultures of
S. noursei, i.e., a quick decrease to 0% after the inoculation, then the 24 h period at 0%, and, finally, the inflection back towards the 100% level. The question was whether the DO profiles recorded for the PRSN1, PRSN2, and PRSN3 co-cultures followed the same behavior as noted for the
S. noursei monocultures. In the case of the PRSN1 co-cultivation variant, the DO curve practically overlapped with the
S. noursei monoculture line (
Figure 3a). In PRSN2 co-culture, in which
S. noursei was inoculated 24 h after
P. rubens, the DO curve remained at the 0% level much longer than in PRSN1, i.e., for almost 110 h. The difference between PRSN1 and PRSN2 in terms of oxygen consumption was thus striking. Finally, the DO behavior in the PRSN3 co-culture was unique in two aspects. Firstly, the period from the moment of reaching the 0% level to inflecting back towards the saturation line lasted for about 72 h, i.e., it was longer than in the
S. noursei monoculture but, at the same time, not as long as in the PRSN2 co-culture. Secondly, the DO profile in the co-culture displayed a somewhat fluctuating trend following its descent to 0%, i.e., it remained at 0% only temporarily, in contrast to what was seen in the PRSN1 and PRSN2 co-cultivations. The DO curves indicated that the PRSN1 co-culture resembled the
S. noursei monoculture and that the simultaneous inoculation of the two microbes did not provide favorable conditions for the development of
P. rubens biomass, which was in turn associated with the company of a fast-growing
S. noursei. In other words, the fungus was dominated by the actinomycete, and the proliferation of
P. rubens mycelium was greatly inhibited under such circumstances. This observation was confirmed during the analysis of SMs originating from
P. rubens (
Figure 2). They were either absent or present in trace amounts in the PRSN1 co-culture. Another piece of evidence was provided when sugar concentrations were determined (
Figure 4). The lack of visible lactose consumption in the PRSN1 co-culture indicated that
P. rubens growth was blocked. The TICs (
Figures S1–S8) collected over the course of the PRSN1 run agreed well with the hypothesis of the domination of
S. noursei over
P. rubens in the simultaneously inoculated co-culture because the data representing the co-culture and the
S. noursei monoculture were highly similar. In the PRSN2 and PRSN3 co-cultivation variants, in which the inoculation of
S. noursei was delayed, the biomass of
P. rubens could freely develop under monoculture conditions for 24 and 48 h, respectively. The main question at this point was if
S. noursei could propagate in the presence of the already-developed biomass of
P. rubens. According to the results of the study,
P. rubens did not prevent
S. noursei from proliferating in the co-culture; however, the presence of the actinomycete was not manifested until the late phases of co-cultivation, as demonstrated with the microscopic analysis and the UPLC-MS assays. Firstly, the microscopic images (
Figures S28–S30) revealed the existence of
S. noursei biomass in the cultivation broth of PRSN2 and PRSN3 co-cultivation. Secondly, the SMs of
S. noursei became detectable in the late phases of the PRSN2 and PRSN3 co-cultures (
Figure 1 and
Figures S10–S27), which illustrated the “awakening” of the biosynthetic activity of the actinomycete. Finally, the consumption of lactose stopped in the final stages of these co-cultures (
Figure 4), which probably corresponded to the inhibition of
P. rubens by
S. noursei. It should also be mentioned that the DO profiles of PRSN2 and PRSN3 co-cultures showed inflection points towards the 100% level, which was a characteristic feature of
S. noursei monocultures. However, it took considerably more time for the co-cultures to reach the moment of DO inflection compared with the
S. noursei monoculture variants (
Figure 3). All in all, the results indicated that despite the domination of
S. noursei over
P. rubens, the former eventually came into play in the PRSN2 and PRSN3 co-cultures despite the growth-related advantage granted to the fungus by delaying the inoculation of the actinomycete.
Even though both PRSN2 and PRSN3 co-cultivation variants were based on the concept of delayed inoculation of
S. noursei to enable the growth of
P. rubens, they differed with respect to the time of delay. According to the results, the co-culture initiation approach had profound effects on the outcomes of the studied bioprocess, as reflected by the levels of SMs reached in the PRSN2 and PRSN3 co-cultures (
Figure 1 and
Figure 2). Perhaps the best illustration of the differences between these two co-cultures was provided through DO measurements (
Figure 3). As mentioned before, the DO profile recorded for the PRSN2 co-culture was characterized by a relatively long period of 0% oxygen level prior to displaying the behavior reminiscent of the
S. noursei monocultures, namely the inflection of the DO curve towards the 100% line. Such a scenario could be attributed to the fact that
S. noursei in the PRSN2 co-culture required more time for biomass development and for exerting its domination over
P. rubens in comparison with the PRSN3 counterpart, in which the time from
S. noursei inoculation to the characteristic inflection point was markedly shorter than in PRSN2. It all demonstrated that the events taking place during “
S. noursei vs.
P. rubens” co-cultivation were, at least to a certain degree, determined using the co-culture initiation approach. This leads to the question of the possible reasons for the differences between PRSN2 and PRSN3. They might have been associated with the development and changing characteristics of fungal pellets under submerged conditions. The pellets are not static units; their structures undergo changes as the cultivation progresses. For instance, the biomass growth takes place not only on the exterior of the pellet but also within its core region. While this topic has not yet been investigated, it is justified to assume that the ability of
P. rubens biomass to engulf (or “trap”) the spores of
S. noursei inside fungal mycelium changes over cultivation time, in concert with the structural transformations of
P. rubens pellets. In other words, the differences related to
P. rubens biomass age in PRSN2 and PRSN3 co-cultures were accompanied by differences with regard to the structural features of fungal pellets and, consequently, their capabilities to inhibit
S. noursei development. Quite surprisingly, the recorded results contradicted the intuitive presumption that the longer delay of
S. noursei inoculation (i.e., 48 h in PRSN3) would result in a longer period of
S. noursei adaptation and, possibly, an even more delayed onset of the domination exerted by
S. noursei over
P. rubens. In fact, the levels of
S. noursei SMs products were usually higher in PRSN3 than in PRSN2, their presence in the PRSN3 broth was typically confirmed after shorter post-inoculation periods than in PRSN2 (
Figure 1), and the inflection of the DO profile came much sooner in PRSN3 compared with PRSN2 (
Figure 3). In short,
S. noursei seemed to perform more efficiently in PRSN3, in which the actinomycete inoculation delay relative to
P. rubens (i.e., 48 h) was longer than in PRSN2 (i.e., 24 h). Another observation that seemed counterintuitive was the fact that the highest levels of several SMs of
P. rubens recorded in the co-cultures were reached not in the PRSN2 co-culture but in the PRSN3 counterpart (
Figure 2b–d,f). It was surprising, as the lactose uptake rates were visibly higher in the PRSN2 co-culture than in the corresponding PRSN3 variant (
Figure 4), and, as already mentioned, lactose consumption was attributed solely to
P. rubens. Supposedly, the mere fact of vigorous lactose utilization was not associated with the marked enhancement of fungal SM production. It seemed as if the relatively fast development of
S. noursei in the PRSN3 co-culture somehow stimulated several biosynthetic pathways in
P. rubens cells compared with the PRSN2 co-culture, i.e., the biochemical routes leading to chrysogine (
Figure 2b), benzylpenicilloic acid (
Figure 2c), adenophostin B (
Figure 2d), and preaustinoid D (
Figure 2f).
The fact that
S. noursei was able to grow and produce SMs after being introduced into the already developed
P. rubens culture agreed with the previous work on “
Streptomyces rimosus vs.
P. rubens” co-cultivation [
23], in which the actinomycete eventually showed its dominance over the fungus despite the inoculation-related disadvantage (i.e., a 24 or 48 h delay). Once more, the utilization of a substrate selectively consumed solely by one of the co-cultivated strains, namely lactose, proved to be an effective method of co-culture investigation. The present study also confirmed the importance of the co-culture initiation strategy for the outcomes of the microbial co-cultivation process. What made the “
S. noursei vs.
P. rubens” system unique was the fact that prolonging the inoculation delay of the actinomycete from 24 (as in PRSN2) to 48 h (as in PRSN3) was often beneficial for the co-culture in the context of SM production, especially with regard to the metabolites originating from
S. noursei. The previously studied “
S. rimosus vs.
P. rubens” co-cultivation did not reveal such tendencies [
23]. All in all, it is well-grounded to state that the “
S. noursei vs.
P. rubens” and “
S. rimosus vs.
P. rubens” co-cultures did not follow the same path, even though they shared some characteristics related to the survival of
Streptomyces sp. in the filamentous biomass of
P. rubens. It should be emphasized that the axenic culture was demonstrated to be a preferred system of production for the majority of
S. noursei and
P. rubens SMs identified in the present study. Still, the “
S. noursei vs.
P. rubens” co-culture was shown to be worth considering as an alternative to the axenic culture in the context of the production of several SMs, most notably desferrioxamine E and deshydroxynocardamine.
Despite the fact that the PRSN1 co-culture and the
S. noursei monoculture were highly similar in terms of DO profiles and sugar consumption, they were far from being equivalent in terms of SM production (
Figure 1). For instance, the levels of desferioxamine E, deshydroxynocardamine, and argvalin were elevated in the PRSN1 co-culture relative to the
S. noursei variant. On the other hand, in the case of actiphenol and obscurolide C2, the PRSN1 co-culture resulted in visibly weaker production than the one recorded for
S. noursei monocultivation. So, the presence of
P. rubens could not be regarded as negligible, even though the production of SMs was practically non-existent during PRSN1 co-cultivation owing to the domination of
S. noursei. In other words, even the blocked growth of
P. rubens did not prevent the fungus from affecting the SM production via
S. noursei. However, the mechanism responsible for this effect remains unknown, and further investigation would be required to elucidate its nature. Finally, it should be underscored that the stimulation of SM biosynthesis was associated with the inoculation scheme; e.g., the aforementioned enhancement of desferrioxamine E, deshydroxynocardamine, and argvalin production occurred exclusively in PRSN1.
The co-cultivation of
Penicillium with
Streptomyces was previously investigated in the context of SM induction in laboratory flasks. In a study by Wang et al. [
29], five unique SMs were obtained in the co-cultures of
Penicillium sp. WC-29-5 and
Streptomyces fradiae 007. More recently, Krespach et al. [
30] showed that arginoketides generated by
Streptomyces sp. activate the formation of secondary metabolites in
Penicillium isolates. In a different study, the influence of
P. rubens on the levels of
S. rimosus-derived SMs was examined in shake flask co-cultures [
31]. However, to the best of our knowledge, the only stirred tank bioreactor-based characterization of the “
Penicillium vs.
Streptomyces” co-culture system was previously reported by our group in relation to the
P. rubens/
S. rimosus case [
23]. In that context, the present study involving
P. rubens and
S. noursei provided examples of the unique features, as well as the similarities, that may be encountered among the
Penicillium vs.
Streptomyces bioreactor co-cultivations.