The ocean covers over 70% of the Earth’s surface and harbors numerous unique habitats distinctly different from terrestrial architecture. Marine bacteria have been discovered and isolated from these unique habitats, ranging from open water environments, seabed sediments, hydrothermal vents, and even as symbionts from specified marine organisms (Di Donato et al., 2018). Their adaptation to these habitats allows them to exhibit various functions and characteristics unique from their terrestrial counterparts (Alex et al., 2023). Due to these features, marine bacteria have been frequently targeted by industry and pharmaceutics as genetic resources for highly active enzymes or novel lead compounds (Lindequist, 2016). Recent advancements in sequencing technologies have also allowed for more comprehensive bacterial diversity analyses, leading to the discovery of countless novel bacteria that have not been characterized or exploited (Escudeiro et al., 2022). Thus, the demand for more useful marine bacteria is much higher now and needs to be addressed.

For the exploitation of useful bacteria, it is important for the target bacteria to be cultivated since it allows users to metabolically engineer and optimize them for the efficient and mass production of important compounds (Jung et al., 2021). Over the years, much effort has been placed into establishing techniques to isolate novel and useful bacteria from the environment (Rodrigues and de Carvalho, 2022). In addition to the conventional Fluorescence In Situ Hybridization (FISH) approach, where specific bacteria can be fluorescently stained, identified, and isolated using flow cytometry or micromanipulation, recent efforts for large-scale screening and high-throughput systems using microfluidics and in combination with single-cell analytical techniques have allowed for a comprehensive understanding of bacterial communities (Moon et al., 2023), construction of near-complete draft genomes of novel and unique bacteria (Kogawa et al., 2018), and elucidation of highly efficient enzymes applicable for downstream processes (Baysoy et al., 2023). However, originally designed for nucleic acid analyses, most of these techniques have incorporated cell fixation or lysis, resulting in many isolated bacteria to be inapplicable for subsequent cultivation.

Thus, there is still a high dependency on conventional direct cultivation in which researchers try to establish culture media and mimic the environmental conditions in standard laboratory settings. Nevertheless, this approach is tedious and time-consuming, and, in many cases, since cultivation conditions are hard to replicate, many bacteria still remain uncultivable. Factors that have been speculated to majorly contribute to this difficulty include undefined growth conditions (Kapinusova et al., 2023) or the dependency of the bacteria on specific scaffolds, metabolites, and nutrients only available from its host or surroundings (Zengler and Zaramela, 2018). Thus, to promote the cultivation of environmental bacteria, an approach to cultivate bacteria in their original environment has been introduced (Chaudhary et al., 2019). Co-cultivation has also been introduced as an alternative approach based on the necessity of some bacteria to depend on other bacteria within their community (Watterson et al., 2020). Although these approaches have been successful in the cultivation of several bacteria, it is still challenging to cultivate environmental bacteria as pure cultures. However, among these factors, one other possible factor that is often missed and could provide a large impact to the direct cultivation of bacteria is the presence of impurities within the bacterial fractions prepared from environmental samples. Environmental samples can contain various organic and inorganic materials, including soil particles, organic debris, and minerals, creating complex matrices that may interfere with bacterial growth (Wilpiszeski et al., 2019) or particles that have a similar size or density to bacteria, making it challenging to distinguish between them (Fatoyinbo et al., 2014).

Currently, the preparation of the bacterial fractions for use with the bacterial isolation techniques is generally performed using conventional differential centrifugation or filtration, but these methods can still result in fractions that contain commendable amounts of impurities (Vaishampayan et al., 2010; Nnadozie et al., 2015). In this work, to demonstrate the significance of removing impurities and their impact on bacterial isolation, we employed two approaches, dielectrophoresis (DEP) and fluorescent D-amino acids (FDAA), to attain clean bacterial fractions from environmental samples. Using the marine sponge Halichondria okadai as a model, clean bacterial fractions were attained, and the distribution and diversity of bacterial populations were determined using nanopore sequencing and compared with differential centrifugation. Finally, the viability of bacteria attained using these approaches was evaluated using direct cultivation. Through our findings, we believe that DEP and FDAA can be applied to prepare clean bacterial fractions for the further exploration and investigation of diverse bacteria from complex environmental samples.

Attaining clean bacterial fractions with less or no impurities from environmental samples is crucial as this greatly influences the possibility of isolating diverse and useful bacteria. Therefore, to attain such clean bacterial fractions, we needed approaches that could distinctly identify bacteria based on their unique characteristics and features. Having searched for possible solutions, we found two approaches that fulfilled this criterion. DEP employs generating non-uniform electric fields via predesigned electrodes that allow for manipulating and capturing polarized particles within microfluidic systems (Wakizaka et al., 2020). Viable bacteria with intact cellular membranes have intracellular ion homeostasis, resulting in them being highly polarized and, therefore, applicable for manipulation with DEP. As a result, DEP can be used to specifically detect, separate, and characterize viable bacteria from impurities present within environmental samples. FDAA, on the other hand, is D-amino acid residues artificially labeled with fluorophores and is designed to be incorporated into the peptidoglycan (PG) layer during cell membrane synthesis for specific labeling of actively growing bacteria (García-Heredia et al., 2018). Hence, we employed DEP and FDAA and tested their applicability in attaining clean bacterial fractions. From our results, where we used the bacterial fraction from the marine sponge H. okadai, in addition to attaining relatively clean bacterial fractions comprised of viable bacteria ( Figures 1 , 2 ), our bacterial distribution and diversity analysis via nanopore sequencing also showed differences in bacterial populations attained using both approaches ( Table 1 ).

From the bacterial distribution and diversity analysis of the DEP-captured bacteria sample, we found that Pseudomonas and Psychrobacter were the most abundant, covering 98.7% of the total bacterial population ( Table 1 ). This correlated with the bacterial distribution from the centrifuged bacteria sample, where Pseudomonas and Psychrobacter were also most dominant at 84.9%. Interestingly, we observed that the populations of Pseudomonas and Psychrobacter were inverted in the DEP-captured bacteria sample, where Pseudomonas was at 61.3%, followed by Psychrobacter at 37.4%. We found this as an interesting observation because the bacterial genera captured by DEP, excluding Psychrobacter, all had motile characteristics. Referring to the principle of DEP and the design of PixeeMo^®, we speculated that it was easier to capture motile bacteria with this system since the probability of these bacteria approaching the electrodes is higher in comparison to non-motile bacteria. Psychrobacter, however, still covered a high percentage of the DEP-captured bacteria sample population simply because they were at high abundance in the final H. okadai bacterial fraction. Being a non-motile bacteria, Psychrobacter may also have the tendency to attach themselves to some of the impurities rather than being in suspension (Staloch et al., 2022), resulting in them to be also removed during the DEP capture process. In addition to Pseudomonas and Psychrobacter, although covering an abundance of only 0.3% in average, motile bacteria such as Aeromonas, Shewanella, Buttiauxella, Serratia, and Exiguobacterium were also captured by DEP.

In contrast to the bacteria captured by DEP, from the FDAA-stained bacteria sample, we found that Clostridium was most abundant, accounting for 73.5% of the total bacterial population ( Table 1 ). The dominance of Clostridium within the FDAA-stained bacteria sample was intriguing, as this genus consists of obligate anaerobes (Maier et al., 2014). As described, the FDAA is incorporated into the PG layer of bacteria during cell growth. Since sBADA staining of the bacteria was performed for 24 h not in anaerobic conditions, it is hard to assume that growth occurred. However, bacteria classified as Clostridium are known to have spore-forming characteristics (Talukdar et al., 2015). The non-optimal growth conditions for Clostridium could have possibly triggered the formation of spores, hence promoting the incorporation of sBADA. The large abundance of Clostridium in the FDAA-stained bacteria sample compared to the centrifuged bacteria sample could be explained by selective sorting during flow cytometry. Among the bacteria from the FDAA-stained bacteria sample, one other spore-forming genus, Paenibacillus, was also identified (Grady et al., 2016). Although not showing high abundance compared to Clostridium, this genus was also only identified from the FDAA-stained bacteria sample. The low abundance of Paenibacillus in the FDAA-stained bacteria sample could be due to their initial abundance within the centrifuged bacteria sample ( Table 1 ). The distribution of other genera within the FDAA-stained bacteria sample comprised mainly of gram-negative bacteria but was at much lower abundance compared to the centrifuged and the DEP-captured bacteria samples. Particularly for Psychrobacter and Pseudomonas, this is probably due to the staining conditions where the samples were stained with sBADA in 1 x PBS. Although we risk the possibility of bacteria losing their viability when incubating in PBS, we avoided using culture media since we wanted to avoid bias growth but instead attain a more diverse bacterial population.

Next, we tested the viability of the bacteria attained from each approach by plating the samples on MB agar plates ( Figure 3 ). Here, we used MB as the culture medium as it is considered a rich media commonly used to isolate marine bacteria. From the centrifuged bacteria sample, upon performing 16S rRNA analysis, we found that they represented the two dominant bacteria, Psychrobacter and Pseudomonas ( Figure 3 ). This was expected as both genera can grow on MB and cover more than 80% of the total bacterial population ( Table 1 ). Likewise, in the DEP-captured bacteria sample, the presence of Pseudomonas was confirmed. However, in contrast to the centrifuged bacteria sample, although representing the second most abundant genus, Psychrobacter was not identified. Surprisingly, Psychrobacter was also not identified from the FDAA-stained bacteria sample. Even though it is hard to determine the exact reason why the growth of Psychrobacter from the DEP-captured and FDAA-stained bacteria samples was not observed, the removal and absence of impurities including possible metabolites, proteins or compounds from the centrifuged bacteria sample during DEP and FDAA treatment may suggest that Psychrobacter from H. okadai could be dependent on such constituents present within the marine sponge to support growth (Staloch et al., 2022). Besides Pseudomonas, the growth of Exiguobacterium was also confirmed from the DEP-captured bacteria sample ( Figure 3 ). Since the abundance of this bacterium was very low ( Table 1 ) and all the other low-abundant genera are also motile and can grow on MB, we assumed that the identification of Exiguobacterium was random.

In the direct cultivation of the FDAA-stained bacteria sample, the growth of two genera, Pseudomonas and Stenotrophomonas, was confirmed ( Figure 3 ). Although, both these bacteria were not highly abundant, cultivation of these bacteria suggests that bacteria isolated by FDAA were viable ( Table 1 ). Clostridium, which represents the most abundant genus, was not identified since we knew that bacteria from this genus are mostly obligate anaerobes, and the growth conditions were not ideal for their cultivation. Subsequent attempts to further cultivate the Clostridium with the proper growth conditions were performed but were not successful. Here, we speculated that the cultivation of Clostridium could have been hindered by other factors rather than viability. Subsequent to Clostridium, from the nanopore sequencing analysis, Saccharospirillum, Psychrobacter, and Brevundimonas were the next most abundant, but no colonies representing neither of these bacteria was identified. As described, we speculated that the growth of Psychrobacter could be its dependency on organic compounds or molecules present within H. okadai. Looking at the characteristics of Saccharospirllum and Brevundimonas, where some bacteria from both genera have been reported to be non-motile (Choi et al., 2011) or adopts a sessile lifestyle (Sperling et al., 2019), we hypothesized that both these bacteria in H. okadai also have similar features to Psychrobacter, hence, no colonies were identified.

H. okadai has brought much interest since unique natural compounds, including halichondrin B, have been identified (Hirata and Uemura, 1986), yet only a few reports on the bacterial distribution in H. okadai (Abe et al., 2012; Jeong et al., 2015) have been published. Thus, in addition to evaluating the distribution and viability of bacteria attained using DEP and FDAA, we also highlight the ability of these approaches to identify unique and useful bacteria. From the DEP-captured bacteria sample, Exiguobacterium and Serratia were newly identified ( Table 1 ). Both genera have been reported from marine sponges, where they harbor enzymes applicable for industrial applications (Shanthakumar et al., 2015; Jadhav et al., 2020). The FDAA-stained bacteria sample, on the other hand, showed the presence of 4 unique genera which were Acidibacter, Bradyrhizobium, Paenibacillus, and Saccharospirillum, among which, as far as we know, Acidibacter, Bradyrhizobium, and Saccharospirillum have not been reported from marine sponges. All 3 genera are commonly found in freshwater and terrestrial environments and have been reported to be associated with wastewater bioremediation (Oren, 2015; Ratnasari et al., 2021) or plants (Saranraj et al., 2021). However, several genera have recently been reported from marine environments. Bradyrhizobium and Saccharospirillum, for example, have recently been found abundant in bacteriocytes of a nudibranch (Zhukova et al., 2022) and associated with a marine dinoflagellate (Yang et al., 2020), respectively, indicating they do serve as possible candidates for harboring novel natural compounds. In contrast, Paenibacillus has been identified in marine sponges and has been associated with antibacterial compounds (Grady et al., 2016). With the employment of DEP and FDAA, we believe these approaches could assist in isolating and discovering novel, unique, and useful bacteria.

In summary, we showed that DEP and FDAA can be used to prepare clean bacterial fractions for the investigation and exploration of viable and useful bacteria from complex environmental samples. Both approaches have been designed to capture or stain bacteria based on their distinct characteristics, allowing for the identification of novel bacteria unique to each approach apart from conventional differential centrifugation. The differences in bacterial populations between the centrifuged bacteria sample and samples attained via both approaches strongly indicate that the removal of impurities from bacterial fractions does indeed have an impact on bacteria isolation. Although further optimization of the DEP and FDAA approaches may lead to the discovery of more unique bacteria, they still have their limitations. Thus, in addition to further optimization of these approaches, other new approaches designed for targeted isolation of bacteria should also be explored.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: DDBJ Sequence Read Archive; BioProject: PRJDB17192 (PSUB021998); BioSample: SAMD00664446-SAMD00664448 (SSUB027844); Experiment: DRX502901-DRX502903 (moritets-0003_Experiment_0001-0003); Run: DRR519015-DRR519017 (moritets-0003_Run_0001-0003). The nanopore sequencing raw data has been submitted to the DDBJ Sequence Read Archive (DRA) with the accession number DRA017584.

The manuscript presents research on animals that do not require ethical approval for their study.

JY: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Validation, Writing – original draft, Writing – review & editing. MK: Data curation, Investigation, Methodology, Writing – original draft. AK: Investigation, Methodology, Writing – original draft. AZ: Investigation, Methodology, Writing – original draft. TM: Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Writing – original draft, Writing – review & editing.