Common Features and Intra-Species Variation of Cutibacterium modestum Strains, and Emended Description of the Species

Cutibacterium modestum is a new species coined in 2020 as the fifth species of genus Cutibacterium, which includes Cutibacterium acnes. The species is predicted as a minor but common member of skin microbiome and includes a group tentatively named as “Propionibacterium humerusii”. The description of the species has been provided only with a single strain. To establish the characteristics of C. modestum and search for possible disease-related subtypes, we investigated the biochemical characteristics of eight live strains and performed in silico comparison of nine genomes. The common features, which included the morphology of Gram-stain positive short rods, the negativity of phenylalanine arylamidase, and several unique MALDI-TOF MS spectral peaks, were considered useful in laboratory identification. Pairwise comparisons of the genomes by in silico DNA–DNA hybridization showed similarity values of 98.1% or larger, which were far higher than the subspecies cutoff of 79–80%. The 16S rRNA gene sequences of thirteen isolates and genomes were identical. Their recA gene sequences were identical except for two strains, HM-510 (HL037PA2) and Marseille-P5998, which showed unique one-nucleotide polymorphisms. The biochemical features using API kits were slightly different among the isolates but far closer than those of the nearest other species, C. acnes and Cutibacterium namnetense. Spectra of MALDI-TOF mass spectrometry showed slight differences in the presence of m/z 10,512 (10 kD chaperonin GroS) and three other peaks, further clustering the eight isolates into three subtypes. These results indicated that these isolates did not separate to form subspecies-level clusters, but subtyping is possible by using recA gene sequences or MALDI-TOF mass spectrometry spectra. Moreover, this work has confirmed that a group “P. humerusii” is included in C. modestum.


Introduction
Genus Cutibacterium was established in 2016 as a result of the reclassification of genus Propionibacterium into four genera, which are Propionibacterium, Cutibacterium, Acidipro-pionibacterium, and Pseudopropionibacterium (later corrected to Arachnia), based on their whole-genome similarity and niche [1,2]. Cutibacterium turned out to be a group of four species that reside on the human skin as their major habitat. Cutibacterium modestum was proposed in 2020 as the fifth species of the genus, with strain M12 T = JCM 33380 T = DSM 109769 T as the type strain [3]. The strains and genome entries of this new species are accumulating. A recent database search in July 2021 shows fourteen 16S rRNA gene sequences of the isolates, with >99% similarities to that of the type strain M12 T , were deposited to the DDBJ/EMBL/GenBank database [4]. The origins of these isolates are the meibomian gland, skin, bone infection, blood, and a cardiac pacemaker device. These isolates include those described with a tentative species name 'Propionibacterium humerusii', which can be considered to be included in C. modestum. In the first description of 'P. humerusii' by Butler-Wu et al. [5], the bacterial group is described as consisting of 3% of the human skin microbiome. This triggers speculation that this species can have significance in human health and diseases.
However, little description of C. modestum is provided yet, and this causes difficulty on the accurate identification in clinical laboratories. This is not only because this species is not yet widely known, but also because there is an ambiguity in the identification procedure as no concrete description of the species exists to rely on. The currently available description is provided based on the analysis of only one strain [3] and there is no evidence that the description covers most of the strains to be further isolated and analyzed. Therefore, there is a need to establish the species description for clinical identification. The accumulation of the correct species-level identifications in laboratories should link to revealing the species' role in human health and diseases. To establish the common characteristics of this new species, we collected the existing isolates and investigated their morphological, genomic, biochemical, and matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry (MS) features. In the collection process, we used the registered 16S rRNA gene and genome sequences in the public nucleotide database as a clue, as these strains are registered with different species names. As a result, we could collect four isolates from Japan and the US. Furthermore, one of the institutions provided three more isolates that were unregistered. On the other hand, a couple of the registered isolates were reported as lost, and a couple of the contacted institutions did not reply.
The recent investigation into Cutibacterium acnes, the most famous and well-investigated species of the genus, resulted in a deep understanding of its pathogenicity. The species shows deep intraspecies clustering that forms three groups/subspecies, all of which are unique in cellular shape, genome, biochemical properties, proteome, and pathological characteristics, deserving subspecies definitions [6,7]. For C. modestum, it is yet unclear whether this new species forms similar distinct groups that may be related to pathogenicity. Therefore, our study also examines whether such clusterings are present in C. modestum by using the live strains and database genomes.

Selection of Bacterial Strains
Strain M12 T (= JCM 33380 T = DSM 109769 T ) was obtained from the human meibomian gland and previously described as the type strain of C. modestum [3]. C. modestum strains KB17-24694, KB10376, KB11371, and KB24935 were obtained from Kobe University Hospital. A C. modestum strain JK19.3 is one of the strains in C. acnes biofilm analysis [8] and was shared from the Jikei University, Department of Bacteriology. Two C. modestum strains, HM-510 (HL037PA2) and HM-515 (HL044PA1), with registrations as Propionibacterium acnes, were obtained from BEI Resources (Manassas, VA, USA). Strains JCM 6425 T (C. acnes subsp. acnes), JCM 6473 T (C. acnes subsp. defendens), and JCM 18919 T (C. acnes subsp. elongatum) were purchased from RIKEN BioResource Research Center Microbe Division/Japan Collection of Microorganisms (Tsukuba, Ibaraki, Japan). A C. namnetense strain DSM 29427 T was purchased from the Leibniz Institute DSMZ-German Collection of Microorganisms and Cell Cultures (Braunschweig, Germany). The details of these strains are listed in Table 1.

Morphological Assessments
The strains used in this study were grown on trypticase soy agar supplemented with 5% sheep blood (BD, Franklin Lakes, NJ, USA) for 5 days at 37 • C under anaerobic conditions created in an anaerobic jar with an AnaeroPack sachet (Mitsubishi Gas Chemical, Tokyo, Japan). The colonies were observed, and Gram stain was performed using Gram Stain Kit (Stabilized) (BD).
In addition, full genome analysis was performed for strain KB17-24694. First, the genomic DNA of the strain was extracted and purified by using Genomic-tip 100/G kit (Qiagen, Hilden, Germany). The concentration and quality of the extracted DNA was checked by using Synergy LX (BioTek, Winooski, VT, USA), QuantiFluor dsDNA System (Promega, Madison, WI, USA), and 5200 Fragment Analyzer System with Agilent HS Genomic DNA 50 kb Kit (Agilent Technologies, Santa Clara, CA, USA), and the material passed for further analysis. For short reads, a library of circularized DNA was created by using MGIEasy FS DNA Library Prep Set and MGIEasy Circularization Kit (MGI Tech, Shenzhen, China). Then, sequencing was performed by using DNBSEQ-G400 (MGI Tech) with a parameter of paired-end 200 bp. For long reads, a library was created by using Ligation Sequence Kit (Oxford Nanopore Technologies, Oxford, UK). Then, sequencing was performed by using GridION (Oxford Nanopore Technologies) with standard procedures. The short reads were removed their adapter sequences and reads with smaller than 127 bp or below Q20 were discarded by using the Cutadapt and Sickle softwares. On the other hand, the long reads were removed their adapter sequences and reads with smaller than 1000 bp were discarded by using the Porechop and Filtlong softwares. The filtered short and long reads were assembled by using Unicycler ver. 0.4.7 to create one chromosome sequence and one plasmid sequence.
The 16S rRNA gene sequences were prepared as a single text file in FASTA format. The sequence file was pasted in a DNA sequence alignment mode of MEGA X ver.10.2.5 software [17] and aligned by MUSCLE alignment mode. The recA gene sequences were also aligned in the same manner.

Biochemical Profiling
C. modestum strains KB17-24694, KB10376, KB11371, KB24935, JK19.3, and C. namnetense strain NTS 31307302 T were cultured anaerobically for five days as described in Section 2.2. Using API kits (API Coryne and API Rapid ID 20A, bioMérieux, Marcy-l'Étoile, France), biochemical analysis was performed according to the manufacturer's instructions for these strains. In addition, catalase tests were performed by using API Coryne kit for C. modestum strain M12 T and all three C. acnes strains.

MALDI-TOF Mass Spectrometry
MALDI-TOF mass spectrometry profiles of the strains were obtained as reported by Dekio et al. [18] with the following modifications. The spectra were obtained with a MALDI-TOF MS Autoflex Speed Biotyper (Bruker, Billerica, MA, USA) with parameters as follows: mass range, 2000-20,000; ion source 1, 19.50 kV and 2, 18.50 kV; lens, 6 kV; detector gain voltage, 2600 V (linear base). The spectra were obtained at least in triplicate and the representative spectrum with the clearest and richest pattern was selected for each strain. The raw spectral data was analysed with flexAnalysis v3.4 software (Bruker). The raw data in mzXML format was deposited to jPOST Repository, with accession number JPST001214 (PXID: PXD026667) (URL: https://repository.jpostdb.org/, accessed on 21 October 2021).

Morphology
After five days of anaerobic culture, all eight strains showed very similar white, opaque, and lenticular colonies with diameters of 0.9-1.0 mm. Gram stain revealed all strains being Gram-positive short rods ( Figure 1).
were created as described previously [18].

Morphology
After five days of anaerobic culture, all eight strains showed very similar white, opaque, and lenticular colonies with diameters of 0.9-1.0 mm. Gram stain revealed all strains being Gram-positive short rods ( Figure 1).

Genome Analysis
The dDDH values for the pairwise comparisons of C. modestum isolates were presented in Table 2. These values ranged between 98.1% and 100%, far above those not only of the species cutoff of 70% [19] but also of the subspecies cutoff of 79-80% [20] (green highlight, Table 2). It was a striking contrast to the pairwise comparisons of the three C. acnes genomes, which showed the values between 75.5% and 73.1%, reflecting their subspecies-level differences (yellow highlight, Table 2). On the other hand, the values between C. modestum genomes and those of other Cutibacterium species were between 22.3% and 32.3% with all the other type strains, which were below the species cutoff of 70%.
The value between the genomes of HM-510 (HL037PA2) and HL037PA3 was 100%, which was the only 100% among the pairwise comparisons. Considering the very similar characteristics of these two genomes (both 2.61 Mb and G + C content 59.94%), and the similarity of these strain names, we speculated these were identical. Based on these findings, we excluded HL037PA3 for further analysis.
A phylogenetic tree by using these dDDH values was successfully created (Figure 2). The three subspecies of C. acnes branched deep, while genomes of C. modestum did not.

16S rRNA Gene and recA Gene Analyses
Nearly full 16S rRNA gene sequences of four KB isolates were successfully determined. The comparison across the twelve 16S rRNA sequences (Table 1, excluding HL037PA3) revealed that all these sequences were identical.
Also, nearly full recA gene sequences of the five KB and JK isolates were successfully determined. The comparison across the twelve recA gene sequences (Table 1, excluding HL037PA3) revealed that two polymorphisms exist, which were G57A for strain HM-510 (HL037PA2) and T66C for strain Marseille-P5998 ( Figure 3). We additionally determined the gene sequence for strain HM-510 (HL037PA2) to confirm this polymorphism. These polymorphisms in the coding region did not affect the translation to amino acids, which are glutamic acid (E) for frame 55-57 (GAG and GAA) and histidine (H) for frame 64-66 (CAT and CAC).

16S rRNA Gene and recA Gene Analyses
Nearly full 16S rRNA gene sequences of four KB isolates were successfully determined. The comparison across the twelve 16S rRNA sequences (Table 1, excluding HL037PA3) revealed that all these sequences were identical.
Also, nearly full recA gene sequences of the five KB and JK isolates were successfully determined. The comparison across the twelve recA gene sequences (Table 1, excluding HL037PA3) revealed that two polymorphisms exist, which were G57A for strain HM-510 (HL037PA2) and T66C for strain Marseille-P5998 ( Figure 3). We additionally determined the gene sequence for strain HM-510 (HL037PA2) to confirm this polymorphism. These polymorphisms in the coding region did not affect the translation to amino acids, which are glutamic acid (E) for frame 55-57 (GAG and GAA) and histidine (H) for frame 64-66 (CAT and CAC).
The characteristic feature of C. modestum strains was a marked negativity among various tests, including α-glucosidase, phenylalanine arylamidase, leucine arylamidase, alanine arylamidase, histidine arylamidase, serine arylamidase, pyrazinamidase, β-galactosidase, gelatin hydrolysis, and ribose fermentation tests. These results indicated relatively lower reactivity of C. modestum strains compared with the type strains of three C. acnes subspecies and C. namnetense. Especially, the negativity of phenylalanine arylamidase was considered to discriminate C. modestum from the other two species. N-Acetyl-β-glucosaminidase and indole tests were negative for C. modestum type strain M12 T but were positive for most other C. modestum strains.
The characteristic feature of C. modestum strains was a marked negativity among various tests, including α-glucosidase, phenylalanine arylamidase, leucine arylamidase, alanine arylamidase, histidine arylamidase, serine arylamidase, pyrazinamidase, β-galactosidase, gelatin hydrolysis, and ribose fermentation tests. These results indicated relatively lower reactivity of C. modestum strains compared with the type strains of three C. acnes subspecies and C. namnetense. Especially, the negativity of phenylalanine arylamidase was considered to discriminate C. modestum from the other two species. N-Acetyl-β-glucosaminidase and indole tests were negative for C. modestum type strain M12 T but were positive for most other C. modestum strains.

MALDI-TOF Mass Spectrometry
MALDI-TOF MS spectra of all eight strains of C. modestum and type strains of C. acnes and C. namnetense were successfully obtained (Figure 4). The spectra of C. modestum   -515 (HL044PA1)). The y-axis is different between 4000-9000 m/z and 9000-12,000 m/z ranges. The peak assignments to proteins were obtained from a previous report [18]. We tried to identify molecules X, Y, and Z by comparing the genomes of strains of Groups A (strain M12 T : BJEN01), B (strain KB17-24694: AP024747), and C (strain HM-515: ADZU01). However, it was not possible for these molecules because there were no gene sequence in Group B and C genomes that corresponded to proteins with the appropriate molecular weights and matching their observed mass to charge ratio (m/z). Following this result, possible post-translational modifications were considered. The observed m/z of the target molecule should increase 14, 42, or 56 by methylation, acetylation, and propionylation, respectively. As a result, the candidate proteins with modifications for these ions were suggested as follows: the candidates for molecule X include C25_01851 hypothetical protein (molecular weight = 10,026. 17 (HM-515 (HL044PA1)). The y-axis is different between 4000-9000 m/z and 9000-12,000 m/z ranges. The peak assignments to proteins were obtained from a previous report [18]. We tried to identify molecules X, Y, and Z by comparing the genomes of strains of Groups A (strain M12 T : BJEN01), B (strain KB17-24694: AP024747), and C (strain HM-515: ADZU01). However, it was not possible for these molecules because there were no gene sequence in Group B and C genomes that corresponded to proteins with the appropriate molecular weights and matching their observed mass to charge ratio (m/z). Following this result, possible post-translational modifications were considered. The observed m/z of the target molecule should increase 14, 42, or 56 by methylation, acetylation, and propionylation, respectively. As a result, the candidate proteins with modifications for these ions were suggested as follows: the candidates for molecule X include C25_01851 hypothetical protein (molecular weight = 10,026. 17

Discussion
Our phenotypic, genomic, biochemical, and MALDI-TOF MS analyses revealed the relative uniformity among C. modestum strains. These strains and genomes did not form clusters at the subspecies level but exhibited differences at the lower level in recA gene, biochemical, and MALDI-TOF MS spectral characteristics, suggesting the feasibility of strain-level typings.

Common Characteristics of Cutibacterium modestum
Our results outlined the description in the previous new species proposal [3], and these did not contradict the description of genus Cutibacterium [1] nor that of previous genus Propionibacterium [21]. Furthermore, the results indicate the taxonomical position of the type strain M12 T being a typical strain among C. modestum strains. The results of this strain include the genome being located very close to other strains in the phylogenetic tree (Figure 2), the identical recA sequence to most other strains (Figure 3), the typical biochemical behavior except negativity in N-acetyl-β-glucosaminidase and indole tests (Table 3), and a common Group A MALDI-TOF MS spectrum (Figure 4).
In the biochemical analysis by using API kits, the numbers of the positive reactions by C. modestum strains were the smallest among the three Cutibacterium species (Table 3), verifying the modest nature of this species, as in its epithet. The results of C. modestum were the most similar to those of C. acnes subsp. elongatum. Among the 32 reactions that consist of the kits, the negativity of phenylalanine arylamidase is specific to distinguish C. modestum from C. acnes or C. namnetense. Moreover, the negativity of α-glucosidase, leucine arylamidase, β-galactosidase, gelatin hydrolysis, and ribose fermentation are key reactions of distinguishing C. modestum.
On the other hand, MALDI-TOF MS spectra of C. modestum showed distinct patterns that can be discriminated from those of the nearest two species as shown in Figure 4 [18]. The majority of these peaks were identical among C. modestum strains, reflecting the uniformity of corresponding coding genes. This fact is strikingly different from C. acnes subtypes, which exhibit different characteristic spectra [18]. The comparative proteogenomics also raised the possibility of the post-translational modification of the observed proteins in the spectra.
The above results of the biochemical and MALDI-TOF MS analysis should be useful for species-level identification in clinical laboratories.

Variation of Cutibacterium modestum
We sought for subspecies-level discrimination at first, in analogy with C. acnes, which includes three subspecies that are different in various aspects, that is in morphology, genome, 16S rRNA gene, recA gene, biochemical tests, and pathogenicity [4]. However, the C. modestum isolates did not show differences in this level. According to our pairwise genome comparisons (dDDH values), the variation range of C. modestum isolates was relatively small to repel subspecies discrimination ( Table 2). The lack of heterogeneity in 16S rRNA gene sequences is consistent with this observation. Following this, we established a method for recA gene typing to investigate the possibility of housekeeping gene typing of this relatively homogeneous population. As a result, we were able to observe three groups based on one-nucleotide polymorphisms: the first group including strain M12 T and the other two groups with mutations G57A and T66C. Although the polymorphisms we observed were within a single nucleotide, we consider this to be significant. This is because a similar one-base difference can distinguish C. acnes type IA and IB, which are different in pathogenicity [22]. In a case of C. acnes, recA gene typing [22] has advanced to multilocus sequence typing (MLST) [23] and to subspecies proposal [6,7], both of which link to the pathogenicity of distinguished groups. Our recA gene typing may become the first step in the further investigation to understand the niche and the role of this new species.
The biochemical analysis revealed slight differences among C. modestum strains. The prominent differences were the negative results for arginine dihydrolase in strains M12 T and KB24935, for N-acetyl-β-glucosaminidase in strain M12 T , for proline arylamidase in strain KB17-24694, and for glucose fermentation in strain HM-510 (HL037PA2). The glucose fermentation negativity in strain HM-510 (HL037PA2) suggests the link of this result to recA gene polymorphism G57A, but further accumulation of isolates is needed for verification.
Mixed results were observed for the presence of indole, which indicates that indole detection by using API Rapid ID 32A kit is not useful for C. modestum identification. This is also the case for C. acnes; the product datasheet of this kit states that the positive rate of indole reaction for C. acnes is only 62%, although Bergey's manual states all three subspecies of C. acnes are indole positive [21]. Goldenberger et al. emphasizes the importance of the indole test along with the catalase test with regard to species delineation. In the article, the need for clarification of strain M12 T , which was negative for indole, was hinted with reference to their isolate 602588-20-USB and strain P08, which were positive [24]. Our results of eight C. modestum strains were weakly positive in six strains and negative in two strains, and no strong positives were observed. However, our genome analysis indicated that strain P08, the indole positive strain, is included in C. modestum (Table 2, Figure 2). Moreover, both positive and negative strains share the same recA gene sequence ( Figure 3) and MALDI-TOF MS spectra (Group A in Figure 4). To investigate whether the difference among strains is derived from the presence/absence of tryptophanase gene, we checked the presence of tryptophanase gene in the four C. modestum genomes of the isolates used in our biochemical analysis (strain M12 T , BJEN01; strain KB17-24694, AP024747; strain HM-510, ADYH01; strain HM-515, ADZU01). All genomes possessed one each of the gene, with 46-49% similarity from Escherichia coli tryptophanase gene (tnaA) and 87% similarity from the correspondent gene in C. acnes. Considering all these results, we speculate that negative, weak positive, and clear positive results for indole can be obtained from C. modestum strains, although the tryptophanase gene is universal.
MALDI-TOF MS spectra of C. modestum showed three spectral patterns, the finding of which proves to be reliable to perform easy and cost-effective type analysis. Unfortunately, the MALDI-TOF MS spectra typings were not congruent with any of the recA gene or biochemical results. More analysis is needed to establish a meaningful typing scheme explaining their differences in niche and pathogenesis.

"Propionibacterium humerusii" Is Included in Cutibacterium modestum
The presence of C. modestum was initially suggested with a tentative name as "Propionibacterium humerusii" in 2011 with genomic analysis of an isolate P08 [5]. Unfortunately, this work did not result in the formal proposal of the species. Our proposal of C. modestum followed, and at that stage, we were aware of the 16S rRNA gene sequence of strain P08 being identical to the type strain of C. modestum [3]. Very recently, Goldenberger et al. stated that C. modestum and "P. humerusii" represent the same species [24]. However, we consider it differently. In the strict sense, the range that "P. humerusii" covers has not been defined, and thus it is unclear whether "P. humerusii" covers all strains of C. modestum. Our conclusion based on this history is that C. modestum is not equal to "P. humerusii", but the former includes the latter.

Conclusions
As described above, our morphological, biochemical, and MALDI-TOF MS analyses outlined the general features of C. modestum, which are useful in species-level diagnostic identification. All C. modestum isolates were Gram-stain positive short rods. Negativity of phenylalanine arylamidase should be a key to distinguish C. modestum from C. acnes and C. namnetense. MALDI-TOF MS spectra of C. modestum is unique and useful for species identification once the spectral data is registered in the identification system.
Moreover, multidimensional analysis of the strains and genomes of C. modestum revealed that strain-level (below subspecies-level) typings of C. modestum are possible by using recA gene sequences, biochemical reactions, or MALDI-TOF MS spectra. The G57A polymorphism of recA gene is especially worth attention, as this is linked to a negative result in the glucose fermentation test in one strain. On the other hand, these strains and genomes did not form subspecies-level clusters. To note, however, this study does not exclude the possibility of future discovery of the subspecies-level clusterings by further accumulation of isolates.