Fusion of Biotechnology and Craftsmanship: Bacterial Treatment to Improve Bashofu Fiber Extraction

ABSTRACT

Bashofu is a traditional Okinawan textile made from thin banana fibers. The raw materials derived from banana leaf sheaths are composed of plant fibers and unwanted constituents such as the plant cuticle layer. The unwanted constituents are hand-scraped by the traditional way that follows boiling the raw materials in a mild alkali solution. However, even after this mild degumming, the plant cuticle layer of current materials can still be too hard to be hand-scraped from the fibers. For smooth fiber separation, the unwanted constituents should be specifically degraded before hand-scraping. Fatty acid polyesters are the main components of plant cuticle layer and are not present in fibers. We attempted to specifically degrade the materials by Stenotrophomonas sp. with the ability to degrade Tween-20, as a result, the treated materials became softer and thinner with uniform thickness. Such changes in the morphology of the material allowed the plant cuticle layer to be easily separated from the banana fibers during hand-scraping, and the cross section of the extracted fibers was not affected by this bacterial treatment. This treatment would be used as a minimal improvement of the traditional Bashofu making in the future and would reduce the hard work for elderly artisans.

摘要

芭蕉芙是冲绳传统的纺织品，由薄香蕉纤维制成. 香蕉叶鞘的原料由植物纤维和植物角质层等不需要的成分组成. 不需要的成分是用传统的方法手工刮除的，即在弱碱溶液中煮沸原料. 然而，即使经过这种温和的脱胶，目前材料的植物角质层仍然太硬，无法用手从纤维上刮下来. 为了使纤维顺利分离，不需要的成分应在手动刮擦前进行专门降解. 脂肪酸聚酯是植物角质层的主要成分，不存在于纤维中. 我们试图通过狭窄单胞菌（Stenotrophomonas sp.）对材料进行特异性降解. 通过降解Tween-20的能力，处理后的材料变得更软、更薄、厚度均匀. 这种材料形态的变化使植物角质层在手工刮擦过程中很容易与香蕉纤维分离，并且提取纤维的横截面不受这种细菌处理的影响. 这种处理将在未来作为对传统巴绍夫制作的最小改进，并将减少老年工匠的辛勤工作.

KEYWORDS:

• Bashofu
• banana fiber
• plant cuticle layer
• Stenotrophomonas sp.
• fatty acid ester
• cultural property

关键词:

• 芭蕉布
• 香蕉纤维
• 植物角质层
• 狭窄单胞菌属
• 脂肪酸酯
• 文化财产

Introduction

Bashofu is a traditional Okinawan banana fiber textile made from Itobashou (Musa balbisiana var. liukiuensis) plant fibers, which are thinner (50–100 μm) than other banana fibers like Abaca (Cai et al. 2016). Nowadays, only experienced artisans are adept at the fine craft of Bashofu making, employing a production method (Fig. S1A) that has remain unchanged for the past 400–500 years. The Bashofu making method of the Kijoka area of Ogimi village, located in the northern region of Okinawa, has been a national cultural property since 1974. In the process, fibers are first degummed in the U-daki step before unwanted constituents are removed using traditional tools in the U-biki step (Fig. S1B, Kijoka no Bashofu Hozonkai 2009). Briefly, during U-daki, raw materials (Figure 1a) are boiled in a solution made mildly alkaline (pH 11). This allows pectin (Fuchigami 2014; Wang et al. 2003) and hemicellulose (Shiiba et al. 1992) to be extracted from the material. Next, the fibers are mechanically separated from other unwanted constituents, such as plant mesophyll cells and cuticle layer, during U-biki.

Figure 1. Raw Bashofu material.

(a) Leaf sheaths (approximately 1.2 m) were peeled off, and the abaxial axis parts (fiber rich) were separated, which were classified into Waha (hard), Nahau (soft), and Nahagu (very soft) according to their hardness. Prepared materials were tied for boiling in a mild alkali solution. (b) Raw plant material is covered with plant cuticle layer. Microscopic images were obtained by the technique reported by Buda et al. (2009). The main components of cutins in plant cuticle layer are polyesters of fatty acid esters, which are not found in fibers. Fibers of Itobashou plant are made of hollow plant cells (sclerenchyma/mesophyll fiber bundles and vascular tissue fibers).

However, such traditional steps, while capable of extracting fibers from materials in good (soft) condition, are not always able to extract fibers from all materials. Since soft materials are obtained from slow-growth Itobashou plants, artisans have deliberately grown Itobashou banana plants slowly by traditionally trimming most of leaves off them at the appropriate growth stage several times per year. However, due to fewer and the aging of artisans, the good condition of the material plant is not managed nowadays.

Recently, for materials that are too hard (e.g., materials called Waha), artisans of Kijoka have raised the pH of alkali solution to 12 (checked with pH test paper) or increased the boiling time; however, these extra treatments result in weakened fibers that can be easily broken during U-biki. On the other hand, this alkali degumming cannot make hard materials enough soft to separate fibers mechanically during U-biki. U-daki is not effective for fiber extraction from such hard materials, which make matters worse that sometimes artisans suffer from tendonitis, because extraction of fibers from hard materials requires a great deal of force during U-biki.

Moreover, for the subsequent thread making steps, artisans require long fibers that do not need to be frequently and laboriously hand tied, such is the case with short fibers. Consequently, broken and short fibers cannot be used for thread making; this is an important issue given the perpetual shortage of Bashofu fibers for craft making. Therefore, Bashofu artisans have been seeking a more efficient way for separating Itobashou fibers from unwanted constituents while they want to conserve their traditional method. Bashofu is valuable as a national cultural property only when it is made using the traditional method (Important Intangible Cultural Property), so only minimal improvements that assist this traditional process would be allowed.

Banana fiber has attracted much attention in recent years from the perspective of utilizing the large amount of edible banana agricultural waste; not only for apparel products but also as a material for bioethanol (Sawarkar et al. 2022), polymer materials (Motaleb et al. 2021), and cellulose nanofibers (Kumar et al. 2019). Recently, extractions of oligosaccharides as functional foods from banana fibers have been reported (Díaz et al. 2021). Banana fibers are modified by various methods including chemical/bio-retting, and obtained fibers are characterized using instrumental analysis like FT-IR, XRD and microscopic technique (Badanayak, Jose, and Bose 2023). In many cases the fibers of fully grown edible banana pseudostem are decorticated by machine, and the obtained fibers are thick and rough surface. As a yarn material, the combination of alkali treatment with ultrasonic method has been used to reduce the surface coarseness of banana fiber, and smooth and thin fibers have been obtained (Twebaze et al. 2022). However, such extractions or modifications using chemicals or machines are for industrial purposes or utilizations of pseudostem as agro-waste, which is not suitable for the traditional Bashofu making.

In general, plant fiber extraction is conducted by either a combination of chemical, biological (bio-retting), or mechanical (hand-scraping and machine extraction) treatments (Sangamithirai and Vasugi 2020). While U-daki and U-biki are chemical (mild alkali boiling) and mechanical (hand-scraping) treatments, respectively, the use of biological treatment for fiber extraction in Bashofu making has yet to be explored because nonspecific and excessive degradation of Itobashou fibers is a major drawback of Bashofu fiber extraction by bio-retting. To avoid over retting, conditions of bio-retting have been investigated (Badanayak, Jose, and Bose 2023), furthermore, microorganisms (Aspergillus Niger) for degradation pectin in banana pseudostem have been selected (Sarma and Deka 2016). However, the bio-retting for agro-waste banana pseudostem is difficult to apply directly to thin Itobashou fiber extraction in the Bashofu production.

The anatomy of the raw material derived from Itobashou plant is shown in Figure 1b. Bashofu fibers are made from the hollow cells of Itobashou plant, which are a kind of sclerenchyma fiber cells or vascular tissue fiber cells. The (cell) walls of these hollow cells are secondary cell walls, which are formed from cellulose fibril bundles held together by hemicellulose, pectin, and lignin. Therefore, excessive degradation of these substances in fibers may cause the fibers themselves to degrade. Nevertheless, for fiber extraction, degradation of pectin is a necessary first step (U-daki step in Bashofu making method) since it connects the plant fibers to adjacent mesophyll cells and cuticle layer.

The accumulation of lignin in secondary cell walls is considered to give plants their hardness. In Bashofu making, conventional degumming (boiling in a mild alkali solution) does not result in the release of a substantial amount of lignin (Nomura et al. 2017); therefore, lignin degradation is a possible means of improving fiber extraction. However, excessive lignin degradation may also result in broken fibers.

From the above, we theorized that the specific degradation of substances other than pectin and lignin may contribute to the effective extraction of Itobashou fibers. To achieve this, we introduced a biological treatment step between U-daki and U-biki to target the removal of plant cuticle layer attached to the fibers; this can be difficult to achieve with the boiling in mild alkali solution of traditional Bashofu making. The presence of this hard cuticle on the fibers prevents their smooth separation in U-biki.

Plant cuticle layer forms the hard surfaces of plants, which is mainly composed of cutins, cutans, and other components such as waxes (Fernández et al. 2016; Yeats and Rose 2013). Interestingly, cutins or waxes are mainly composed of fatty acid esters (Li and Beisson 2009; Yeats and Rose 2013), and fatty acid esters are not contained in plant secondary cell walls. Previous reports have shown that cuticle layer degradation is an important step for fiber extraction. Fungus that can degrade cuticle layer can be used for fiber extraction at the early stages of hemp field-retting (Fernando et al. 2019). Moreover, the use of cutinase to degrade cutins has been investigated for scouring of cellulose (Degani, Gepstein, and Dosorets 2002).

In this study, we selected bacterial strains in the soil of Itobashou fields in Kijoka and identified a strain of Stenotrophomonas sp. as the best candidate. Members of the Stenotrophomonas genus are found in soil and plants (Ryan et al. 2009), and Stenotrophomonas nitritireducens is related to conversion of fatty acids (Kang et al. 2017; Yu et al. 2008). We tested the ability of our isolated strain Stenotrophomonas sp. to degrade substances related to fiber extraction and found that the strain was able to degrade the fatty acid ester Tween-20 but showed poor growth on medium containing pectin and no decolorization of lignin agar plates. Furthermore, we found that materials treated with this Stenotrophomonas sp. had decreased and more uniform thickness, and the fibers could be easily extracted mechanically from the treated materials. The biological treatment with this Stenotrophomonas sp. did not affect the cross section of the extracted fibers.

This study also addresses how treatment with this bacterial strain could be incorporated into the traditional method of Bashofu making.

Materials and methods

An overview of this study and the flowchart of experiments are depicted in Figure 2. Details of the experiments are described below.

Figure 2. Overview of this study and the flowchart of experiments.

Bacteria selection and isolation

Selection of bacteria

We prepared nine inoculum samples from Itobashou fields of Kijoka. Samples #1–7 were from soil and #8 was from the rotten pseudostem of an Itobashou plant. Sample #9 was water suspended sediment from a ditch adjacent to the Kijoka fields. Phosphate buffered saline (pH 7.4, 137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, and 2 mM KH₂PO₄) was added to samples #1–8, and the samples were incubated at 25°C for 3 h in a shaking incubator at 250 rpm.

Itobashou plant materials were prepared by the Bashofu Textile Studio (Kijoka). Leaf sheaths were peeled from Itobashou pseudostems, and each leaf sheath was split into abaxial and adaxial sides. Only the abaxial side, which contains many long fibers, was used as Bashofu fiber material (Figure 1). The raw materials were degummed in a boiling alkali solution at the Bashofu Textile Studio. We used either raw or degummed material for the bacterial selection process.

We have summarized the origins of the bacterial inoculum samples (Fig. S2A) in Fig. S2B and the selection steps in Fig. S2C. Control cultures not containing inoculum solution were also included at each selection stage (Fig. S2D).

For the first selection round, 300 μL of each extract or sample #9 was added to 3 mL of M9 medium (1× M9 salt, 2 mM MgSO₄, 0.1 mM CaCl₂, 0.1% casamino acids) containing raw Itobashou plant material (0.3 g) (Fig. S2D). The tubes were shaken at 250 rpm for 4 days at 25°C, followed by incubation without shaking at room temperature (8 days). For the second and third-1 round selections, 150 μL of culture from the previous round was inoculated into fresh M9 medium containing fresh raw material. Incubation of the cultures was conducted by the conditions in Fig. S2C. The shaking temperature was increased to 30°C from the third-2 selection round.

For the fourth and fifth selection rounds, alkali-degummed material was added to 5 mL of M9 medium together with 500 μL of culture from the previous selection round.

Colony isolation

The culture from inoculum sample #1 after the fourth selection round was appropriately diluted and plated on R2A agar plates (BBL^TM, 251258; Becton, Dickinson and Company, Franklin Lakes, NJ), then the plates were incubated at 30°C for 1–2 days. Several colonies were carefully picked and streaked on fresh R2A agar plates, and the plates were incubated. Each colony carefully picked from the streaked plates was streaked again on fresh plates, and then incubated (repeated at least two times). Isolated colonies from the plates were used for colony polymerase chain reaction (PCR; see section “Identification of bacterial strains by 16S rRNA sequence analysis”) amplification of a DNA sequence corresponding to a partial 16S rRNA region. Colonies that were positive for the sequence were cultured in tryptic soy broth (TSB, Bacto^TM 211825; Becton, Dickinson and Company) liquid medium for 1–2 days at 30°C at 250 rpm and were subsequently used for identification of the bacterial strains by whole 16S rRNA analysis.

Identification of bacterial strains by 16S rRNA sequence analysis

A DNA sequence corresponding to a partial 16S rRNA region (V4–V5) of the isolated strains was amplified by colony PCR according to a commercial method (Phusion High-Fidelity PCR Master mix, M0531; New England Biolabs, Ipswich, MA) with PCR primers U515F (5’-GTGYCAGCMGCCGCGGTA) and CC926R (CCCCGYCAATTCMTTTRAGTT) in our laboratory. The amplified sequences were analyzed by Sanger sequencing. Moreover, for a few of isolated bacterial strains, a DNA sequence corresponding to the entire 16S rRNA gene (~1,500 bp) was amplified and analyzed by NIPPON STEEL Eco-Tech Corporation (Chiba, Japan) using the following primers: 27F (5’-AGAGTTTGATCMTGGCTCAG-3’), 518r_f (5’-CCAGCAGCCGCGGTAAT-3’), Bac1055YF (5’-ATGGYTGTCGTCAGCT-3’), 518 r (5’-ATTACCGCGGCTGCTGG-3’), 1055YF-R (5’-AGCTGACGACARCCAT-3’), and pH (5’-AAGGAGGTGATCCAGCCGCA-3’). BLAST analysis was conducted for identification of the bacterial strains.

Prior to analysis for isolated strains, dominant bacteria of the culture #1 from the fourth selection round were identified by NIPPON STEEL Eco-Tech Corporation. For this identification, partial 16S rRNA of genomic DNA was PCR amplified with U515F and 926 R (CCGYCAATTCMTTTRAGTT) and analyzed using SILVA for taxonomic classification.

Characterization of Stenotrophomonas sp.

Culture conditions

Bacterial strain #13 (hereinafter “strain #13”) and the E. coli TOP10 control were individually grown in TSB liquid medium at 30°C for 18–22 h in a shaking incubator at 250 rpm. From the pre-cultures (OD₆₀₀ = 8.0), 5 μL each was added to 2.5 mL aliquots of M9 liquid medium containing each of the following carbon sources (solid chemicals, 1.0% [w/v]; liquid chemicals, 1.0% [v/v]): sodium lignin sulfonate (SL, L0098; Tokyo Chemical Industry Co., Ltd., Tokyo, Japan), pectin (26235–82; Nacalai Tesque Inc., Kyoto, Japan), Tween-20 (Tw20, 28353–14; Nacalai Tesque Inc.), Tween-80 (Tw80, 35703–62; Nacalai Tesque Inc.), linear alkylbenzene sulfonate (LAS, 195–07682; FUJIFILM Wako Pure Chemical Industries Ltd., Osaka, Japan), sodium dodecyl sulfate (SDS, 02873–62; Nacalai Tesque Inc.), amyl laurate (AL, L0012; Tokyo Chemical Industry Co., Ltd.), and butyl laurate (BL, L0083; Tokyo Chemical Industry Co., Ltd). We used casamino acids (casa, 0.1%; Bacto^TM, 223050; Thermo Fisher Scientific Inc., USA), an amino acid mixture frequently used as a supplement of M9 minimal medium, as a (positive) growth control. The effect of glycerol (0.4%; 17018–25; Nacalai Tesque Inc.) on the growth of strain #13 in Tw20 or SL liquid medium was also tested. The inoculated media were cultured at 30°C for 3 days in a shaking incubator at 250 rpm. The OD₆₀₀ values of the cultures were measured using a cell density meter (Ultrospec10; General Electric). Because the SL and pectin solutions were respectively blown and cloudy before inoculation, non-inoculated media for both were used as references for the OD₆₀₀ measurements. M9 medium was used as the reference for the other cultures.

We prepared M9 agar (1.5%) plates containing 1.0% of each carbon source (Tw20, pectin, SL, and AL) without either casamino acids or glycerol. For the AL plate, AL was mixed with hot M9 medium by vortexing to produce a homogeneous solution that was then immediately poured into empty petri dishes. AL floated to the surface during solidification and covered the surfaces of the prepared agar plates. Approximately 5 μL (for OD₆₀₀ = 8.0, adjusted accordingly; e.g., 5.71 μL if the original culture was OD₆₀₀ = 7.0) or non-adjusted volume 5 μL of fresh pre-cultures of strain #13 and E. coli TOP10 (control) were streaked onto the plates containing each carbon source, and the plates were incubated at 30°C for a few days to 3 weeks.

Tw20 detection

Tw20 in the liquid cultures was detected using lipophilic carbocyanine fluorescent dye DiIC18(3) (041–33423; FUJI FILM Wako Pure Chemical Industries Ltd.) as previously described (Martos et al. 2020). Three pre-cultures were prepared individually, and 5 μL (for OD₆₀₀ = 8.0) of each pre-culture was inoculated into 2.5 mL of M9 medium containing 1.0% Tw20 (no glycerol or casamino acids) and incubated at 30°C for 3 days in a shaking incubator at 250 rpm. On each day, 10 μL of culture from each tube was sampled for Tw20 measurements (white solids in the cultures were avoided). The OD₆₀₀ values of the cultures were also measured.

To measure the Tw20 levels in the cultures, 2 mM DiIC18(3) was dissolved in ethanol and diluted to 20 μM with MilliQ water. Acetate buffer (3 mol/L, pH 5.2, 31138–31; Nacalai Tesque Inc.) was diluted to 100 mM with MilliQ water. Each 10 μL aliquot of bacterial culture was mixed with 90 μL of MilliQ water. The diluted cultures (4 μL) were dispensed into the wells of a black 96-well, flat-bottom plate (Nunclon^TM Delta Surface; Thermo Fisher Scientific) and 196 μL of reaction mixture (172.4 μL of MilliQ water, 3.6 μL of 20 μM DiIC18(3), and 20 μL of 100 mM acetate buffer) was added to each well and carefully mixed by pipetting. The plate was covered with a lid, incubated at 37°C for 30–40 min, and cooled to room temperature. The fluorescence of the samples was measured using an Infinite M1000PRO plate reader (Tecan) with an excitation wavelength of 550 nm and emission wavelength of 615 nm (both with a 5-nm bandwidth). Background noise was reduced by subtracting the value of M9 medium without Tw20 from each reading. Tw20 medium without inoculation was used as a control for each measurement, and the ratios of the values of the samples to this control were calculated.

The effect of strain #13 on the fiber treated by the traditional method

The Bashofu Textile Studio prepared alkali-degummed materials from 3 Itobashou plants. Strain #13 was pre-cultured in TSB liquid medium overnight at 30°C at 250 rpm. The material (50-g) was added to 250 mL of M9 medium with 0.4% glycerol and 0.1% casamino acids inoculated 2.5 mL of the pre-culture (OD₆₀₀ = 2.0) in a glass bottle. A sample without the pre-culture was also prepared. The bottles were incubated at 30°C for 5 days in a shaking incubator at 170 rpm. After incubation, the fibers were separated from debris (unwanted constituents) by hand, thoroughly washed with water, and dried at 60°C for 20–22 h. Randomly picked fibers were fixed with FAA solution (2.5:2.5:45:50 mixture of formalin, acetic acid, ethanol, and water), embedded in Technovit 7100 resin, and cut into 8-µm cross-sectional slices. The sections were stained with 0.1% toluidine blue and observed under an optical microscope (Ni-E; Nikon Co.). The cell wall areas in the images were analyzed using Image J software, and the cross-sectional areas of the fibers were determined from the rims of the cell walls of outermost cells. The minimum threshold methods (Otsu 1979) were used for binarization to measure the areas.

Moreover, dried 15-g of fibers with or without strain #13 treatment were respectively collected and measured cellulose, hemicellulose, and lignin using the detergent method. This measurement was conducted by TOKAI TECHNO Co. Ltd (Mie, Japan).

Breaking test of autoclaved material

Raw materials (Nahagu leaf sheaths, mixed) were harvested from one Itobashou plant in our laboratory and processed according to traditional methods (Kijoka no Bashofu Hozonkai 2009; Nomura et al. 2017). The soft raw material called Nahagu 25-g was placed in 250 mL of pure water in a glass bottle and autoclaved (121°C, 20 min). After cooling, 0.5 mL of pre-cultures (OD₆₀₀ = 8.0) of strain #13 or E. coli TOP10 was added, and the bottles were incubated at 30°C for 1 week in a shaking incubator at 100 rpm. After incubation, the material was rinsed with water and stored at 4°C.

For the breaking test (Fig. S3), the material was randomly picked and cut (typically into 8 mm × 50 mm segments). The cut material was completely soaked in water for at least 4 h prior to the breaking test because artisans use wet material for U-biki. After removing surface water with Kimwipe, a piece of materials was placed on the stage of a creepmeter (RHEONER II; Yamaden Co., Ltd) with the abaxial axis facing up. The breaking test was carried out at room temperature, with the P21 plunger at 20 N at 1 mm/s, and a detection interval of 0.08 s.

For morphological observations, the fibers were separated from the fresh material by laboratory U-biki (Fig. S1C) and dried at room temperature. A few fibers were randomly picked and observed by microscopy (as in above section).

Results and discussion

Selection of bacterial strains

Bashofu raw materials can be easily degraded in soil. Therefore, we used raw Itobashou materials from the first to third-1 selection rounds. Cultures #1, 4, 6, and 9 were cloudy until the second selection round. However, these cultures were not cloudy at the third-1 selection round, and the appearance of the materials in the cultures remained unchanged.

This was likely because the temperature of culture was low for bacterial growth, whereas raw Bashofu material in soil may be partially degraded by fungus, which is known to degrade lignin and other persistent substances (Bumpus et al. 1985; Fernando et al. 2019). As a result, the shaking incubator temperature was increased to 30°C for growth of bacteria, and the cultures in the third-2 selection round were again cloudy. However, the appearance of materials was not dramatically changed after a long-term static incubation period at room temperature after this shaking culture. Then, the fourth and fifth selection rounds were used fresh cultures with alkali-degummed Itobashou materials instead of raw materials. The appearances (degraded) of Itobashou materials in cultures #1 and #6 after the fourth selection round were similar to those after the fifth selection round respectively (Fig. S2E), and we deemed the selection process to have plateaued at the fourth selection round.

From the first to the third selection rounds, the negative controls were not cloudy. However, from the fourth and fifth rounds which used alkali-degummed materials, some precipitation was found in the negative control cultures although the appearance of the materials in the controls remained largely unaltered (blank in Fig. S2E).

Identification of selected bacteria

The culture #1 from the fourth selection round was used for the isolation of colonies. Dominant bacterial genus of the culture #1 were Bacteroides sp. (48.9%) and Azonexus sp. (20.9%). First, we used M9-agar plates containing degummed Itobashou materials for colony isolation, but Azonexus sp. was not identified. Then, we appropriately diluted the culture and plated it on R2A agar plates, which can be used for the isolation of various bacterial strains, including Azonexus sp. (Jangir et al. 2016). Sixteen colonies were picked and a 16S rRNA region amplification was confirmed in 13 colonies out of them by colony PCR. Sanger sequence analysis revealed that four colonies (#4, 9, 12, and 15) were the same sequence. The sequences of three colonies (#5, 13, and 16) were the same except for one or two mutations.

Colonies #2, 10, 11, 12, 13, and 14, which all grew in TSB liquid medium, were analyzed by whole 16S rRNA sequencing. The results of this analysis are summarized in Table 1. The sequence of strain #2 had good agreement with Pseudomonas sp. 38B, which was isolated from soil in Okinawa that can degrade long-chain alkanes (https://www.ncbi.nlm.nih.gov/nuccore/360038752). On the other hand, while the sequence of strain #11 was clean, several candidate species (e.g., Alcaligenaceae sp.) were identified as having a sequence similarity (≧98.6%).

Table 1. Isolated bacteria.

Colonies classified by the same sequence*	An example of similarity of full-length 16S rRNA sequence
2	100% to Pseudomonas sp. 38B
6
4, 9, 12, 15	100% to Delftia acidovorans NBRC 14,950
5, 13,16	99.8% to Stenotrophomonas nitritireducens CCUG 54,932
8
10	99.7% to Achromobacter pulmonis R-16442
11	98.6% to Alcaligenaceae bacterium DS5
14	96.2% to Azoarcus sp. KH32C
1, 3, 7

Bold number colonies (2, 10, 11, 12, 13, 14) were respectively amplified in TSB medium for analysis of the full-length 16S rRNA sequence. Colony 6 was identified as Pseudomonas sp. by analysis of a partial16S rRNA in our laboratory. Colony 1, 3, 7 were not amplified by colony PCR. Colony 8 was amplified by colony PCR, but the full-length 16S rRNA was not tested because growth was very slow in TSB.

*A partial sequence of 16S rRNA.

The strain from colony #14 was Azoarcus sp., which is closely related to Azonexus sp. (Reinhold-Hurek and Hurek 2006). The highest value of similarity for the 16S rRNA sequence of strain #14 was only 96.2%, which was with the reported sequences of Azoarcus sp. (https://www.ncbi.nlm.nih.gov/nuccore/358635055). This low score is a possible indication of a new species (Edgar and Valencia 2018). To confirm this, a more thorough analysis for the whole genome sequence will be required.

The 16S rRNA sequence of strain #13 revealed the isolation of Stenotrophomonas sp. and with 99.8% sequence similarity to that of S. nitritireducens (Table 1), which can hydroxylate unsaturated fatty acid (Yu et al. 2008). Since Stenotrophomonas sp. is related to fatty acid modification, we focused on the characterization of strain #13 as a next step.

Characterization of strain #13

Pectin and lignin degradation

For the separation of Itobashou fibers, pectin degradation is required and is achieved in mild alkali boiling U-daki. Some pectin structures are degraded by β-elimination reactions (Diaz, Anthon, and Barrett 2007; Fuchigami 2014) in mild alkali solutions, and pectin is extracted by autoclaving (Chandel et al. 2016). Plant materials treated with these processes release pectin and become softer. However, excessive degradation of pectin may also cause serious damage to thin Itobashou fibers. In practice, lignin, which gives hardness to plants and fibers, only requires mild degradation, too.

Growth of strain #13 in pectin liquid culture was not reproducible (OD₆₀₀ value of 0.28 and 0.2, no-growth for other individual 4 samples) after incubation at 30°C for 3 days, and strain #13 was not grown on pectin agar plates (Fig. S4A). We grew strain #13 in SL liquid culture (Figure 3a) and on SL agar plates, but decolorization of the plates was not observed (Fig. S4A). Based on these results, we determined strain #13 to have a low ability to degrade for lignin.

Figure 3. Characterization of bacterial strain #13.

a: Three-day incubation in M9 medium containing 1.0% Tween-20 (Tw20), Tween-80 (Tw80), or sodium lignin sulfonate (SL). Positive control is 0.1% casamino acids (casa). Hight of bars indicates average values and error bars indicate maximum and minimum values (n = 3). b: White solid in the culture after 3-day incubation (1.0% Tw20-M9 culture; 2.5 mL). c: Culture streak on amyl laurate (AL)-M9 agar plate (incubated at 30°C for 1 week, no casamino acids). Left: strain #13, right: E. coli TOP10. d: Three-day degradation of Tw20 (n = 3). TSB: 6 uL of pure TSB liquid (instead of overnight culture) was added to 1.0% Tw20 media. Error bars indicate maximum and minimum values. The values in parentheses are OD 600 of strain #13 culture at each day (average values).

Furthermore, we did not detect Remazol Brilliant Blue-xylan degradation (Biely, Mislovičová, and Toman 1988) by strain #13, indicating that this bacterial strain is not able to degrade xylan (polysaccharide of hemicellulose) which also coexists in fibers (plant cell walls) and other parts of plants.

Growth in culture containing fatty acid ester substances

We aerobically cultured strain #13 in media containing fatty acid detergents that are easily dispersed in the media. Strain #13 grew well in Tw20 liquid medium (Figure 3a) and on solid medium (Fig. S4A). White solids were seen in the Tw20 liquid medium containing strain #13 after incubation for 2–3 days (Figure 3b); these solids may have been decomposition byproduct of Tw20, lauric acid, or fatty acid-related compounds following ester degradation. Strain #13 grew slowly in liquid medium containing Tw80 having longer fatty acid than Tw20 (Figure 3a, Fig. S4C).

Although the fatty acid esters AL and BL could not be completely dissolved in M9 liquid culture, strain #13 was able to grow in liquid medium containing AL (OD₆₀₀ ≈0.2 for 3-day-incubation) or BL (OD₆₀₀ ≈0.14 for 6-day-incubation), albeit at a slow rate. Furthermore, we grew the strain on AL-M9 agar plates (Figure 3c). We found that while strain #13 was able to grow on this agar plate, E. coli TOP10 was not.

In contrast, strain #13 did not grow in medium with detergents LAS or SDS which lack ester bonds. Furthermore, this strain did not grow in culture containing only glycerol, and its growth in medium with Tw20 or SL was not affected by the presence of glycerol (Fig. S4B). Glycerol is contained in cutin which is a component of plant cuticle layer (Graça et al. 2002). We expected fatty acid esters in cutin was degraded by strain #13 regardless of presence of glycerol. Structures of these chemicals are depicted in Fig. S4C.

Decrease of Tw20 in culture

The values 0.03, 0.63, 0.89, and 0.77 in Figure 3d are the average OD₆₀₀ values of cultures inoculated with strain #13. After incubation for 1-day, residual Tw20 in the culture decreased to approximately 60%. From day 2, small white particles were seen in the culture; these particles then aggregated to form white solids (Figure 3b), which were not seen in the E. coli TOP10 culture or non-inoculated cultures. The OD₆₀₀ value was affected by these particles/solids, and after 3 days, the concentration of white solids caused the OD₆₀₀ value to decrease from that of day 2. E. coli TOP10 did not grow in the Tw20 culture medium, and residual Tw20 showed no decrease during the incubation period.

The results suggest that strain #13 is able to degrade fatty acid esters and to utilize them for growth.

Effect of strain #13 on the fiber treated by the traditional method

Next, we performed fiber extraction with strain #13 from alkali-degummed materials provided by the Bashofu Textile Studio. The samples were not ideal as they were too soft for U-biki; the materials (Waha) had probably been boiled for a long period in a strong alkali solution. Fibers extracted from such materials are more fragile than usual; nevertheless, we applied strain #13 to these extremely soft materials.

The fibers after culture with strain #13 were compared to those without strain #13 (Figure 4a). Most of the debris was removed during shaking incubation, however, the treatment by strain #13 did not cause fiber decomposition. Following treatment of the materials with strain #13, the weight of the extracted and dried fibers (n = 9) was 1.69 g (standard deviation [STDEV], 0.244) from the 50-g wet starting materials. This was a slight increase from the 1.54 g (STDEV, 0.077) in non-inoculated case (n = 7). We thus concluded that strain #13 did not affect the fiber yield.

Figure 4. Effect of bacterial strain #13 treatment on Waha fibers.

a: Fibers after treatment and washing. Left, without bacterial strain #13 treatment (=Non-inoculated); right, with bacterial strain #13 treatment. b: Fiber cross section images. Scale bars, 100 μm. c: Areas calculated from images of single fiber cross section in B (n≧5, included vascular tissue fiber). The threshold for binarization in Image J was determined using Otsu method. Hight of bars indicates average values and error bars indicate ± STDEV. The values (52.6, 48.0, and 46.6) above the bars of cell wall in the graph indicate the ratio of the cell wall area to cross-sectional area (%).

Next, the content of hemicellulose, cellulose and lignin of fibers before and after treated with strain #13 was measured. Cellulose (untreated 57 g/100 g-dry, treated 56 g/100 g-dry) or lignin (untreated 6.3 g/100 g-dry, treated 6.2 g/100 g-dry) was respectively similar, but hemicellulose was decreased after treatment with strain #13 (untreated 17 g/100 g-dry, treated 14 g/100 g-dry).

Furthermore, morphology of the fibers was observed. Fibers were randomly picked and observed under the microscope. For this observation, lignin in plant cell walls was stained with toluidine blue. Most of fibers were identified as sclerenchyma fiber by the images of those cross sections (Figure 4b). Comparisons of the cross-sectional areas calculated from these images of Figure 4b are shown in Figure 4c. Treatment with strain #13 did not affect the cross sections or cell walls; the ratios of the cell wall area to cross-sectional area for treatment of shaking with strain #13 (46.6%) and non-inoculated (shaking without strain #13, 48.0%) were similar, indicating that the cell walls had not been degraded by strain #13. Untreated was 52.6%, however the images in Figure 4b were consistent with change in lignin content in untreated and treated fiber samples.

Material breaking test

Thickness and mechanical properties of the material

In the next stage of analysis, we first autoclaved the sample (raw) materials. We were concerned about microbial contamination since the alkali-degummed samples from the Bashofu Textile Studio (see section “Effect of strain #13 on the fiber treated by the traditional method”) had been prepared in an open environment.

To test the mechanical strength of the autoclaved samples, we picked 10 at random and analyzed them using a creepmeter, a device typically used for evaluating the texture of foods such as vegetables (Fujimoto et al. 2015; Kanamaru et al. 2021). In the actual U-biki process, materials are not pulled vertically like in a tensile test. Instead, they are first compressed with the edge of the bamboo tool tweezers before the fibers are pulled out. The creepmeter test is akin to the first compression step of U-biki.

All raw data of breaking test (n = 10) were summarized in Table S1, and analyzed. As shown in Table S1, thickness (before loading) and breaking strain length of materials with bacterial treatment (stain #13 or TOP10) or without bacterial treatment (Non-inoc) were measured with a creepmeter (Fig. S3). The thickness and mechanical properties of the samples are shown in Figure 5. The average thickness (before loading) of the materials treated with strain #13 (0.859 mm; Figure 5a, c) was thinner than that of the non-inoculated samples (1.018 mm; t-test, p = .03). Moreover, the coefficient of variation (CV; STDEV/average) of the samples treated with strain #13 decreased to 0.13 from 0.18 (non-inoculated samples), suggesting that the thickness of the materials treated with strain #13 was more uniform (Figure 5c).

Figure 5. Mechanical properties of the materials after biological treatment.

Hight of bars in the graphs (a, b) indicates average values and error bars indicate ± STDEV. Non-inoc is non-inoculated sample. a: Material thickness and breaking strain length. The values (73.9, 78.1, 78.2) in the graph indicate the ratio of the breaking strain length (average value) to material thickness (average value) (%). More detail, see Thickness ratio (%) in Table S1. b: Breaking load. c: Summary of analyzed values (n = 10).

The average breaking strain length with loading on the materials treated with strain #13 (0.672 mm) was smaller than the other tested materials (Figure 5a,c) although t-test analysis between treated and untreated samples did not show a significant difference (non-inoculated to strain #13, p = .24; non-inoculated to TOP10, p = .13). The ratios of the average breaking strain length to the average material thickness (%) were 73.9 (non-inoculated), 78.1 (E. coli TOP10), and 78.2 (strain #13), with the values for the treated samples being higher than the non-inoculated samples. Qualitatively, we found that it was easier for the plunger to dig into the treated materials since they had been softened.

The result of average of breaking load (Figure 5b) was the same tendency as the results of the thickness and the breaking strain length (Figure 5a, TOP10 > Non-inoc > strain #13). The breaking load (Figure 5b,c) of the materials treated with strain #13 (average, 5.338 N) was 80.2% of that of the non-inoculated samples (6.653 N) although t-test analysis between them did not show a significant difference (p = .18). Moreover, the CV (Figure 5c) of the strain #13-treated materials (0.23) was much smaller than that of the non-inoculated materials (0.40). This difference was likely related to the equalized thickness of the treated materials.

Interestingly, the samples treated with E. coli TOP10 were the thickest among all tested samples. One possible reason for this is that the materials treated with E. coli TOP10 were more susceptible to water-swelling because of the degradation of some substances in the materials. The greater thickness also gave rise to an increased breaking load.

That is, the sample treated with TOP10 was soft but thick, the sample treated with strain #13 was soft and thin with a uniform thickness, and the non-inoculated sample was hard and intermediate thickness. We think that the reduction in thickness of the material treated by strain #13 is due to degradation of the plant cuticle layer. However, from this study it is not clear yet to be clarified degradation of which parts (cuticle layer or mesophyll or both) of the materials causes thickness reduction of the materials treated with strain #13. A more in-depth investigation will be needed to ascertain it.

We then separated fibers from randomly picked samples prepared by the laboratory U-biki shown in Fig. S1C. The laboratory U-biki was performed a number of times, and each time we found that it was easier to separate fibers from the materials treated with strain #13 than it was for those treated with E. coli Top10 or without bacterial treatment, both of which required stronger forces in our U-biki for fiber separation and, as a consequence, resulted in greater damage to the materials and breakage of fibers.

Although the number of samples for the breaking test was only 10, we are confident that the quantitative mechanical characterization of materials can facilitate smooth fiber separation in U-biki. We would like to conduct a similar analysis of material hardness on more samples prepared by artisans to determine the best bacterial treatment conditions to produce materials with the desired properties.

Fiber observation of autoclaved samples

Fiber cross sections were compared (Figure 6) in the same method as in section “Effect of strain #13 on the fiber treated by the traditional method”. The average cross sections of the biologically treated samples (strain #13 or E. coli TOP10) were slightly smaller than those of the non-inoculated samples. This was likely because the materials had not been treated in alkali solution, and some substances it contained were subsequently degraded during biological treatment. Calculated ratios of the cell wall area to cross-sectional area for non-inoculated (40.2%) and #13 (42.1%) were similar.

Figure 6. Cross section of fibers after biological treatment.

a: Example images of fiber cross sections. Scale bars, 100 μm. Debris (yellow circle) attached to fiber cross sections was ignored when calculating the area. b: Cross-sectional area and cell wall area (n≧17). The threshold for binarization in Image J was determined using Otsu method. Hight of bars indicates average values and error bars indicate ± STDEV. The values (40.2, 36.3, and 42.1) above the bars of cell wall in the graph indicate the ratio of the cell wall area to cross-sectional area (%).

As the harvested Itobashou plant and pre-treatment steps (boiling in alkali solution or autoclaving) in these tests were different from those in section of “Effect of strain #13 on the fiber treated by the traditional method”, the results from the respective sections cannot be directly compared. However, when taken separately, the results did not show respectively any significant differences in fiber cross-sectional area or cell wall area between samples treated with strain #13, E. coli TOP10, or non-inoculated samples (except for cell wall area for non-inoculated to TOP10 in Figure 6b, t-test, p = .06).

From the results presented in sections “Effect of strain #13 on the fiber treated by the traditional method” and “Material breaking test”, treatment with strain #13 did not negatively affect the fibers of the materials, regardless of the condition of the starting materials. The degradation of unwanted constituents by strain #13 allowed us to obtain fibers with minimal force and without causing undesired fiber breakage. While treatment with strain #13 did not alter fiber thickness (= cross-sectional area) significantly, the overall thickness of the treated material was clearly thinner than the non-inoculated material (Figure 5a).

The possibility of using bacterial strain #13 in traditional bashofu making

The traditional method of Bashofu making is a national cultural heritage of Japan and a unique remnant of ancient Ryukyu culture. In order to preserve this tradition, modern science should not invade the Bashofu method; rather, scientists should investigate the most minor yet effective ways of improving and complementing traditional methods.

To increase the yield of fibers for Bashofu, the application of cutinase to Bashofu makes sense from a simple scientific standpoint. However, enzyme treatment is a modern and expensive technique and one that would require special conditions (optimal pH and temperature) and equipment at traditional Bashofu studios. Therefore, modern enzyme treatment is incompatible with traditional Bashofu making.

Although bio-retting has not been developed for Itobashou fiber extraction, traditional whitening of Bashofu textiles at the finishing step (Kijoka no Bashofu Hozonkai 2009) is related to microbial treatment; the Bashofu textile is soaked in a fermented yam or rice solution called Yunaji. Traditionally, Itobashou leaves from trimmings are biodegraded in soil as agro-waste and used as fertilizer in Itobashou fields. Therefore, a minor modification of the traditional method by using microbes to facilitate U-biki would be favorable for artisans, who may prefer such an approach over adding extra alkali or increasing boiling times.

Bacterial treatment with strain #13 could be implemented into the traditional method after U-daki (boiling in a mild alkali solution) and before U-biki, with the material being soaked in water containing this strain. Indeed, historical records show that after U-daki, the material was sometimes soaked in water (Kanehisa 2014). Moreover, this strain was isolated from a traditional Kijoka Itobashou field. For these reasons, we think that Bashofu artisans are willing to adopt this minor modification of their methods to increase fiber yield during a time of chronic fine fiber shortage.

We already started to discuss with the artisans how to use this bacterial treatment in the old Bashofu making. We brought 7 samples in various conditions after U-daki at Kijoka Bashofu Studio to our laboratory and treated them with strain #13. Fibers were separated from these samples by traditional U-biki at the local studio. We found that careful adjustment of time of both the biological treatment and U-daki was needed because the artisans used indeed various samples. However, in this primary practical attempt, in a few cases fiber separations were easier than usual during U-biki step, and after this biological treatment the artisans were able to separate from fibers hard materials harvested from overgrown Itobashou plants.

Itobashou plant growth control with traditional trimming and U-biki are hard work for elderly artisans. This easy separation of fibers from hard materials is very beneficial to them because it prevents tendonitis in their hands. Furthermore, this will also help to solve the long fiber shortage, which has been the biggest issue in the production of Bashofu. This attempt is just at starting point, and we may have lots of difficulties to use our bacterial treatment practically, however, we would like to continue to adjust the strain #13 treatment in the next harvest season.

Like this study, the scientific and quantitative characterization of traditional materials will contribute to traditional craft production. We hope to elucidate the degradation mechanism of fatty acid esters in the cuticle layer by this strain to more efficiently and easily control fiber extraction, as well as for scientific interest.

Conclusions

The conservation of traditional Ryukyuan craft Bashofu is our primary concern, and the goal of this study was to improve the traditional fiber extraction with as small changes as possible. We isolated Stenotrophomonas sp. from a field of material banana plant, and this strain would degrade unwanted fatty acid esters in Bashofu materials. The obtained materials after treated with the Stenotrophomonas sp. were soft and thinner with uniform thickness, and the fibers could be easily separated mechanically. Microscopic observations of fiber cross section showed the treatment with the Stenotrophomonas sp. did not affect the extracted fibers. This bacterial treatment was considered as an acceptably minimal change to the old Bashofu production process preserved as a national Important Intangible Cultural Property.

However, our study is still primary, and has limitations due to lack of materials. For example, we must investigate how this biological treatment affects wax content and cellulose crystallinity of the extracted fibers, besides of experiments for practical applications. Moreover, more mechanical tests to materials or fibers will be required. We would like to make these measurements issues for future consideration.

Research highlights

• To attempt to adapt bio-retting to the traditional fiber extraction, Stenotrophomonas sp. was isolated from a local banana plant field.

• Treatment using this strain made hard materials soft and allowed easy separation of fibers.

• Our bacterial treatment was considered as an acceptably minimal change to the old Bashofu production process preserved as an Important Intangible Cultural Property.

Acknowledgments

We would like to thank the Bashofu Textile Studio (Okinawa, Japan) and Hitomi Shinzato (OIST) for sample preparation, Takako Kai (a former OIST member) and Mayumi Suzuki (a former OIST member) for Sanger sequencing experiments, Tropical Technology Plus (Okinawa, Japan) for the creepmeter experiments, and Ryohei Yoshida (OIST) for valuable discussion on the experiment of autoclaving sample with the bacterial treatment. We would also like to express our deep gratitude to Yayoi Maehara (OIST) for significant contributions to fiber extractions and mechanical characterization experiments.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15440478.2024.2351166

Additional information

Funding

This research was supported by Okinawa Institute of Science and Technology (OIST) Graduate University research fund and Japan Society for the Promotion of Science (JSPS) KAKENHI Grant Number [20K02354].