Direct contact of fermented rice bran beds promotes food-to-hand transmission of lactic acid bacteria

Abstract The skin microbiome, which varies widely between individuals, plays a crucial role in human health. It also interacts with the environment in various ways, including during the preparation of fermented food. Nukadoko is a pickle and traditional fermented food in Japan that utilizes lactic acid bacteria to ferment vegetables. When preparing or maintaining Nukadoko, it is mixed with bare hands. Despite the known interaction between Nukadoko and human skin, no studies have explored its impact on Nukadoko quality or skin microbiome changes. This study examines these effects during Nukadoko maintenance. Three participants were asked to stir commercially available late-stage Nukadoko for 14 days and not stir it for the remaining 14 days to examine microbial settlement and shedding. Microbiome analysis was performed on human skin and Nukadoko. We found that microorganisms from rice bran beds can temporarily settle on human skin but are shed quickly. Stirring rice bran beds by hand may have short-term effects on the skin microbiome. This study provides insights into the communication between human and food microbiomes in traditional Japanese fermented foods.


Introduction
Fermentation is a phenomenon used for food pr eserv ation carried out by micr oor ganisms .Among fermented foods , pickles , fla vored by fermenting v egetables, ar e pr oduced in food industries worldwide .F ermentation impro ves food preservation and aids in the de v elopment of its aroma, flavor, and texture.Lactic acid bacteria play a primary role in fermentation, specifically in the homofermentation and heterofermentation of lactic acid.Bacteria that ar e gener all y undesir able for food pr eserv ation, suc h as gr ampositiv e bacteria, ar e vulner able to low pH.These bacteria carry out fermentativ e pr oduction under anaer obic conditions to induce the growth of lactic acid bacteria and production of lactic acid (Voidarou et al. 2020 ).This type of fermentation is commonly observed in dairy products and in fermented vegetables (Ashaolu and Reale 2020 ).
Nukadoko, a traditional Japanese fermented food (Nakayama et al. 2007 ), is a rice bran bed that ripens pickles ( Nukazuke ).The traditional and predominantly manual method of preparing Nukadoko is to add salt water to the rice bran, knead it well, and then add vegetables to the rice bran bed for natural fermentation in the presence of lactic acid bacteria (Sakamoto et al. 2011 ;Ono et al. 2014 ).Nukadoko produced in this manner has a variety of micr oor ganisms containing Gr am-positiv e bacteria, Gr amnegative bacteria, and yeast (Ono et al. 2014 ).Recently, the addition of fermentation starters, such as long-aged or commercially av ailable Nukadoko , whic h allows easier and mor e stable pr epa-ration and maintenance of the bed, has become the mainstream method (Sakamoto et al. 2011 ).
The microbial composition of Nukadoko and surfaces of pickled vegetables has been investigated through massively parallel sequencing to identify 16S ribosomal RN A (16S rRN A) amplicon sequences (Nakayama et al. 2007 ;Sakamoto et al. 2011 ;Ono et al. 2014Ono et al. , 2015 ; ;Sawada et al. 2021 ).Pyr osequencing-based anal ysis r ev ealed the micr obial dynamics of Nukadoko created in the laboratory with 16 different long-term aged bran beds as fermentation starters (Sakamoto et al. 2011 ).Nukadoko of different origins, in combination with fermentation starters, sho w ed a variety of microbial compositions.Another study sho w ed that the microbial diversity of Nukadoko with added spices, such as Japanese peppers and r ed peppers, differ ed because of the effect of secondary metabolites in spices (Ono et al. 2015 ).Nukadoko from different manufacturers has also been reported to contain different microbiomes (Ono et al. 2014 ;Sawada et al. 2021 ).Furthermor e, div ersity in organic and amino acids, which is influenced by microbiome variations in pickled vegetables (Sawada et al. 2021 ), significantly affects fla vors .
Ho w e v er, for maintaining optimal micr obial comm unities in Nukadoko , the rice bran bed r equir es stirring with bar e hands either daily or every few da ys .T he human skin is inhabited by v arious micr oor ganisms that can affect fermentation (Byrd, Belkaid and Segre 2018 ).Previous studies have identified human skin-associated Staphylococcus in Nukadoko at an early stage of pr epar ation (Ishizaki et al. 2001 ).Ho w e v er, no studies hav e examined how the skin micr obiome, whic h v aries widel y fr om indi vidual to indi vidual, affects the quality of Nukadoko .Conv ersel y, Nukadoko can contain micr oor ganisms that may benefit the human skin.The effects of continued exposure to Nukadoko on the microbial composition of the human skin have never been thoroughl y e v aluated.
We used 16S rRNA amplicon sequencing to e v aluate the effects of interaction between Nukadoko and the human skin microbiome during Nukadoko maintenance .T hr ee anon ymous participants maintained a commercially available Nukadoko at a late stage for 30 da ys .Shar ed amplicon sequencing v ariants (ASVs) wer e computed to identify micr oor ganisms tr ansmitted fr om Nukadoko to human skin.This study sheds light on the human-food microbiome interaction in traditional Japanese fermented foods.

Ethics
The study protocol was approved by the local ethics r esearc h committee of Waseda University (Ethics Review Procedures Concerning Research with Human Participants; application number: 2021-423; a ppr ov ed on February 7, 2022).All pr ocedur es wer e conducted according to the ethics committee's guidelines and regulations .All participants pro vided written informed consent before participating in the study.

Nukadoko maintenance and sample collection
The study participants were healthy volunteers recruited from acquaintances (N = 3); all were Japanese nationals, of which two were female, and one was male .T he study was conducted in Tok y o, J apan, in February and March, 2022.The participants were giv en commerciall y av ailable Nukadoko at the late sta ge and wer e asked to stir it for 14 days and not stir it for the remaining 14 days to examine microbial settlement and shedding on the skin.P articipants wer e asked to turn ov er Nukadoko fr om the bottom, mixing it thor oughl y.This mixing was repeated 2 or 3 times, with the whole session lasting about 3 min.We did not restrict participants from using of soap or ethanol in this study after mixing sessions .Nukadoko samples were collected on days 0, 3, 6, 9, 12, and 14 using individually wrapped disposable plastic spoons.Skin microbiome samples were collected on days 0, 3, 6, 9, 12, 14, 15, 18, 21, 24, 27, and 29 by swabbing the palm for 3 min using a sterile cotton-tipped swab (ESwab™; Copan Dia gnostics, Br escia, Italy).Sw abs w ere stored in tubes with Liquid Amies Medium solution.(Copan Dia gnostics, Br escia, Ital y).Both sample types were immediatel y fr ozen and stor ed at −20 • C until DNA extraction.The study w orkflo w is illustrated in Fig. 1 .The sampling duration for each Nukadoko and skin microbiome sample was at least 6 h.

Total DNA extraction and high-throughput sequencing
Samples wer e tr eated with 750 μL of l ysis buffer fr om the Gen-Find V2 DNA extraction kit (Beckman Coulter, Indianapolis, IN, USA).The suspension was vortexed for 10 min, heat-treated at 100 • C for 10 min, and centrifuged for 5 min at 20 000 g.The supernatant was then mixed with EZ beads (AMR, Tok y o, J apan), and DN A w as fr a gmented using the MM-400 unit (Retsc h, Haan, Germany) at a maximum speed for 3 min.The remaining DN A purification steps w er e performed using the abov ementioned GenFind V2 DNA extraction kit (Beckman Coulter), according to the manufacturer's protocol.DN A w as eluted with 80 μL of nuclease-free water; using the KAPA HiFi Hot-Start ReadyMix (Roche, Basel, Switzerland) (Caporaso et al. 2011 ;Klindworth et al. 2013 ) and specific primers (341F: 5 -TCGTCGGC AGCGTC AGA TGTGT A T AAGAGAGACACCT ACGGGNGG CWGCA G-3 ) and 806R (5 -GTCTCGTGGGCTCGGGAGA TGTGT A T A A GA GA CA GGA CTA CHV GGGTATCT AATCC-3 ), the V3-V4 region of the 16S rRNA gene was amplified.The thermal conditions were 95 • C for 3 min, follo w ed b y 32 c ycles at 95 • C for 30 s, 55 • C for 30 s, and 72 • C for 30 s, with a final extension at 72 • C for 5 min.DNA samples, libr ary pr epar ation, and amplicon sequencing were performed using 300-bp paired-end sequencing on the Illumina MiSeq platform (Illumina Inc., San Diego, CA, USA) at GenomeRead Inc. (Kagawa, Japan).

Microbiome analysis
Micr obiome anal ysis was performed as pr e viousl y r eported (Ito et al. 2023 ).Briefly, r aw FASTQ files wer e imported into the QI-IME2 platform (2022.8)as qza files (Bolyen et al. 2019 ).Denoising and read quality control were performed using the QIIME dada2 denoise-paired function, and reads were classified into ASVs (Callahan et al. 2016 ).We used 269 nt for-p-trunc-len-f and 255 nt for -p-trunc-len-r .The SIL VA database's SSU 138 ( https: // www.arb-silva.de/documentation/ release-138/ ) was used with the QIIME feature-classifier classification scikit-learn package for taxonomic assignment (Quast et al. 2012 ;Bokulich et al. 2018 ).ASVs classified as c hlor oplast, mitoc hondria, or unassigned wer e excluded from subsequent statistical analysis.Subsampling is a common method for inferring micr obiome differ ences between samples and is a suitable analytical approach for analyzing new datasets.To e v aluate the effect of sequence r ead counts on micr obiome diversity assessment, we plotted changes in the Shannon index over a range of read counts from 0 to 10 000, using rarefaction curves.

Calculation of shared ASVs
We defined shared ASVs as ASVs shared by > 1% of both datasets ( Nukadoko and skin) in this study.When Nukadoko was touched for the first two weeks, data from Nukadoko and skin from the same day were used as pairs; when the bran was not touched for the next two weeks, data from Nukadoko from the last day and each skin microbiome data were used as pairs .T he calculation was conducted using R (version 4.

Nukadoko formed a conserv a ti v e microbiome
First, 16S rRN A amplicon sequencing w as performed to investigate the extent to which the skin microbiome affected the rice bran beds.After removal of mitochondrial and chloroplastderiv ed r eads, we obtained 18 114 reads at maxim um, 13 053 r eads at minimum, and 15 937 reads at the median for Nukadoko samples and 21 706, 41 949, and 32 996 reads for skin samples.Details of the r eads gener ated fr om D AD A2 ar e pr esented in Table S1 .We did not observe substantial changes in the microbiome composition of Nukadoko over two weeks compared with that on day 1.Specifically, the Loigolactibacillus genus was predominantly abundant among all three participants and accounted for approximately 69%-79% of the relative abundance throughout the 14 days (Fig. 2 A).Pantoea was the second most common genus, accounting for 5%-10% of the total.Xanthomonas and Staphylococcus were also identified on all the da ys .Howe v er, the tr end of Loigonolactobacillus comprising m uc h of the microbiome composition did not c hange.Shannon div ersity index, as an alpha div ersity indicator, and the observed features did not show any substantial variation.Shannon diversity index was a ppr oximatel y 6 consistentl y, and the observed features were approximately 50, showing daily and participant-specific variation, both slight (Fig. 2 B).

The skin microbiome varies from participant to participant
P articipants stirr ed the br an and collected micr obiomes fr om their palms using the swab method 6 h later by themselves.In contrast to the Nukadoko microbiome, the skin microbiome varied from participant to participant (Fig. 3 A).Across the participants, Cutibacterium , Pseudomonas , Staph ylococcus , and Acinetobacter w ere the most common genera.Acinetobacter was more abundant in Participant 1, while Cutibacterium was more consistently identified in Participant 2, and K ocuria w as particularly identified in Participant 3 than in the other two participants.Participants spent two weeks maintaining Nukadoko with monitoring and were further observed for two weeks without contact with it (Fig. 3 A, B).
We computed the shared ASVs to determine the extent to whic h micr oor ganisms wer e tr ansferr ed between Nukadoko and the participants' palm skin.The abundances of the shared ASVs in skin samples are shown as a bar graph (Fig. 3 C).Shared ASVs were not detected before stirring Nukadoko (Pr e-inter action on Da y 0: Da y 0) except for subject 3 (1.77%).We confirmed that shared ASVs were positively detected on the skin throughout the first tw o w eeks including post-interaction on Day 0 (Day 0').The range of the shared ASVs was from 1.08% to 22.9%.The identified and shared ASVs were derived from Loigolactibacillus, Allorhizobium-Neorhizobium-Pararhizobium-Rhizobium, or Unassigned (Fig. 3 C) .Loigolactibacillus which had become the dominant genus in Nukadoko, was detected as dominant in the shared ASVs of the skin microbiome of all participants .T he shared ASVs were detected on subjects 1 and 3 on day 15.No further observations of the shared ASVs were made after day 15.

Discussion
This study r e v ealed that Nukadoko, at the late stage, formed an extr emel y conserv ativ e and stable micr obial comm unity.Loigolactibacillus , the dominant species of Nukadoko , was briefly transferred to the skin microbiome.
Out of the three stages of Nukadoko , the initial stage (before day 10), middle stage (day 10-30), and late stage (after day 30), the late stage was investigated in this study (Ono et al. 2014 ).Pr e vious studies have investigated the stable expansion of the microbiome in rice bran beds by inoculating plain rice bran with a fermentation starter and maintaining the transition of the microbiome thr ough the thr ee sta ges (Sakamoto et al. 2011 ).Ho w e v er, most customers buy commercially available matured Nukadoko and ferment vegetables by soaking them.No r esearc h has yet been conducted on microbiome variation during the maintenance of this fermented food at the late stage.To the best of our knowledge, this is the first study to address this issue .T he most important characteristic of Nukadoko is that it r equir es car eful stirring with bar e hands by caretakers.Because of this, Nukadoko is always at risk of the easy introduction of foreign and undesirable microbes .T he skin microbiome can also harbor bacteria that cause food poisoning, such as Staphylococcus (Kadariya, Smith and Thapaliya 2014 ).
In this study, thr ee differ ent participants maintained Nukadoko at home, and the microbiome hardly fluctuated in any of the batc hes ov er tw o w eeks .T he genus Loigolactibacillus was the priority species for the Nukadoko inv estigated (Fig. 2 A).Alternativ el y, Nukadoko and pickled vegetables with Lactiplantibacillus plantarum as the dominant species and extr emel y div erse micr oor ganisms was pr e viousl y r eported (Ono et al. 2014 ;Sawada et al. 2021 ).One of the problems in this comparison is the reclassification of the Lactobacillus genus in 2020 (Zheng et al. 2020 ).A r eanal ysis of past studies is r equir ed to coordinate gr oups in Nukadoko based on their micr obiome c har acteristics.Although absent in Nukadoko used in this study, Halomonas spp.found in the final product of pickled vegetables has been reported to contribute to the ele v ation of glutamate concentrations (Sawada et al. 2021 ).Microorganisms in Nukadoko may contribute to the formation of flavors, and the role of eac h micr oor ganism should be thor oughl y inv estigated in future studies.Nukadoko is a fermented food that is customizable and r equir es consider ation of numer ous par ameters to identify its chemistry, including the ingredients to be utilized, the location of the fermentation, and the people who will produce it.T hus , developing a microbiome database of the fermented food can lead to safer and more flavorful fermentation.
Se v er al studies hav e used shar ed ASVs, including bacterial ASV tr ansmission anal ysis, to determine the extent to whic h micr oorganisms are shared between mothers and infants (Maqsood et al. 2019 ) and a survey on microorganisms in milk collected from sever al r egions and seasons in China (Liang et al. 2022 ).In our study, Nukadoko was collected before stirring, and skin samples were collected 6 h after stirring.Ther efor e, Nukadoko and skin samples were used for pairwise shared ASVs analysis, allowing us to confirm the sharing rate on each day (Fig. 3 C).
16S rRNA amplicon sequencing is becoming an incr easingl y useful and affordable technique for microbiome screening.Howe v er, it has become clear that the r esults v ary depending on the DNA extraction method, type of universal primer utilized, and method of analysis (Keisam et al. 2016 ).Similarly, some studies hav e r eported that sampling methods also affect the alpha diversity of skin micr obiome (Bjerr e et al. 2019 ).Ther efor e, to allow for variations owing to technical problems, the threshold for shared ASVs was set to 1% in this study.Shared ASVs are a valid calculation for identifying the microbial source of fermented foods but is limited by the shortcomings of 16S rRNA amplicon sequencing.To clarify the extent to whic h micr oor ganisms hav e been tr ansferred, it is necessary to detect cells at the single-cell level and compr ehensiv el y compar e the r esults, using meta genomics.Another technical limitation of 16S rRNA amplicon sequencing is the inability to distinguish between live and dead bacterial cells.To assess the impact of bacteria more accurately, it is necessary to employ culturing or staining-based methods that provide higher r esolution.Also, 16S rRNA amplicon sequencing onl y detects bacteria, whereas yeast has been reported in rice bran.Yeast plays an important role in the flavor of bran as it is responsible for ethanol fermentation.It is necessary to investigate the amount of yeast present in bran beds by ITS amplicon sequencing or metagenomic analysis.
Studies of skin microbiome transfer have been reported in the past that considered the results of microbiome transfer from different donors to participants over a 24-h timescale (Perin et al. 2019 ).This study suggests that the microbiome implanted in the donor is present for 24 h.Our data are consistent with this, as Loigolactibacillus was identified on day 15, e v en after the participant stopped touching the bran on day 14.The micr oor ganisms ma y ha v e differ ent effects on the host in terms of the settlement, but e v en touc hing the br an bed may cause attac hment for a short period.
We hav e pr e viousl y inv estigated the emotional r elationships between human maintaining Nukadoko and its microbiome using an inter activ e Nukadoko r obot or Nukabot (Chen et al. 2021 ).In the context of human-computer interaction, we evaluated the process of participants gaining awareness of native microorganisms through vocal conversation.When Nukadoko was stirred daily at incr easing r ates, mor e conv ersation took place, and a higher sense of emotional care was generated among the participants.In this study, we r e v ealed a bacterial tr ansfer fr om Nukadoko to the human skin microbiome.Ho w ever, bacterial communication can occur in bidir ectional, fr om Nukadoko to human, and human to Nukadoko .Although ther e hav e been studies on the production of Nukadoko , none have addressed how much of the microbiome in Nukadoko comes or tr ansfers fr om humans during its initial stages.T hus , m uc h r emains to be discov er ed about micr obial-le v el interaction between humans and Nukadoko and human-food communication.

F igure 1 .
Study w orkflo w.Monitoring w as performed for tw o w eeks, and Nukadoko and skin samples w ere collected as a pair.Nukadoko samples w ere collected with plastic spoons in the first 2 weeks.Palm swabbing for skin microbes was done in parallel.A sampling of skin microbiome on days 15 to 29 was done without stirring Nukadoko to measure the speed of bacterial shedding.

Figure 2 .
Figure 2. Changes in the microbial composition of the rice-bran beds.(A) Relative composition of the microbiome of rice-bran beds maintained by each participant over time .T he top 15 genera are presented by their names, and the rest are grouped as the remainder .(B) Time-course changes in observ ed featur es and Shannon div ersity index in micr obial comm unities of rice-br an beds.

Figure 3 .
Figure 3. Changes in the microbial composition of the skin with the interaction of the rice-bran beds.(A) Relative composition of the skin microbiome of each participant over time from day 0 to day 14.The top 15 genera are presented by their names, and the rest are grouped as the remainder .(B) Relative composition of the skin microbiome without interaction with Nukadoko from each participant over time from day 15 to day 29.(C) The abundance of shared ASVs and their compositions .T he X-axis shows days and the Y-axis indicates the proportion of shared ASVs in skin.'Day 0' is befor e inter action, and 'Day 0" is 6 h after interaction.