Self-resistance in Streptomyces, with Special Reference to β-Lactam Antibiotics

Antibiotic resistance is one of the most serious public health problems. Among bacterial resistance, β-lactam antibiotic resistance is the most prevailing and threatening area. Antibiotic resistance is thought to originate in antibiotic-producing bacteria such as Streptomyces. In this review, β-lactamases and penicillin-binding proteins (PBPs) in Streptomyces are explored mainly by phylogenetic analyses from the viewpoint of self-resistance. Although PBPs are more important than β-lactamases in self-resistance, phylogenetically diverse β-lactamases exist in Streptomyces. While class A β-lactamases are mostly detected in their enzyme activity, over two to five times more classes B and C β-lactamase genes are identified at the whole genomic level. These genes can subsequently be transferred to pathogenic bacteria. As for PBPs, two pairs of low affinity PBPs protect Streptomyces from the attack of self-producing and other environmental β-lactam antibiotics. PBPs with PASTA domains are detectable only in class A PBPs in Actinobacteria with the exception of Streptomyces. None of the Streptomyces has PBPs with PASTA domains. However, one of class B PBPs without PASTA domain and a serine/threonine protein kinase with four PASTA domains are located in adjacent positions in most Streptomyces. These class B type PBPs are involved in the spore wall synthesizing complex and probably in self-resistance. Lastly, this paper emphasizes that the resistance mechanisms in Streptomyces are very hard to deal with, despite great efforts in finding new antibiotics.


Introduction
The first antibiotic, penicillin, was discovered in 1929 by Fleming [1] and rediscovered in 1940 by Chain et al. [2] and in 1941 by Abraham et al. [3]. Surprisingly, even before the rediscovery of penicillin, a bacterial penicillinase (β-lactamase) was reported [4] and only few years later, penicillin-resistant Staphylococcus aureus [5,6] and Streptococcus pneumoniae [7] were already described, suggesting strongly that the resistant trait was intrinsically present in their genomes of these bacteria and, in addition, penicillin-inactivating substances (β-lactamases?) were speculated to be involved in the resistance [8]. Gram-negative bacteria such as Neisseria gonorrhoeae [9], Neisseria meningitides [10] and Escherichia coli [11] were intrinsically resistant. The same phenomenon was also observed with streptomycin [12]. That the resistance traits were inherently present in the genomes of Streptomyces was supported by the fact that the percentage production of β-lactamases in newly isolated Streptomyces strains and those isolated thirty years ago from soil was not so different with each other at the time in 1978 [13]. If the selective pressure of the use of β-lactam antibiotics would affect the resistant trait, the relative frequencies of β-lactamase producing Streptomyces strains newly isolated from soil would be much more, indicating that the selective pressure at that time is not so strong as at the present time. Recently, however, D'Costa et al. demonstrated that genes conferring resistance be highly resistant to benzyl penicillin, although they are Gram-positive bacteria, so it is interesting to know the self-resistance mechanisms in Streptomyces and the relationships between PBPs, β-lactams and β-lactamases [19,37]. This review discusses the self-resistance in Streptomyces, focusing primarily on β-lactam antibiotics.

Self-resistance mechanisms
As described above, the self-resistance in antibiotic-producing bacteria such as Streptomyces is at least one of the origins of multi-drug resistance prevailing and threatening in various environments at the present time. Major biochemical mechanisms of drug resistance are detoxication or inactivation of the drug, changing the target site, blocking the transport of the drug into the cell and the efflux pump system [19,38]. These mechanisms are true with the self-resistance mechanisms or, bacteria use very similar or the same mechanisms for self-resistance in the antibiotic-producing bacteria and antibiotic resistance in pathogenic bacteria. While the major resistance mechanism in pathogenic bacteria is β-lactamases, that in Streptomyces is supposed to be penicillin-binding proteins [19]. A number of excellent review papers and databases have been published on antibacterial multidrug resistance and self-resistance [19,[39][40][41][42][43].
β-Lactamases are grouped into four classes on the basis of their amino acid sequences [65][66][67][68][69][70][71][72]. This classification reflects more or less the substrate specificity: class A (penicillin-hydrolyzing), class C (cephalosporin-hydrolyzing) and class D (oxacillin-hydrolyzing) β-lactamases are active site serine enzymes and class B β-lactamases require one or two zinc ions for their activity and are called as metallo-β-lactamases [69,[73][74][75][76]. A previous phylogenetic study indicated that class A β-lactamases could be divided into three subgroups [77]. β-Lactamases are not essential bacterial proteins in themselves, but are speculated to have evolved from the essential PBPs in some β-lactam-producing bacteria such as Streptomyces or related bacteria, because these bacteria have to have some self-protective strategies against β-lactam antibiotics [19,48]. PBPs are involved in peptidoglycan biosynthesis in prokaryotic microorganisms. Therefore, it is interesting to know the relationships of β-lactam antibiotics, β-lactamases and PBPs in Streptomyces, β-lactam-producing microorganisms. In addition, the relationship of β-lactamases in Streptomyces and those in pathogenic bacteria should be clarified. In this section, the phylogenetic relationships were analyzed among β-lactamases in Streptomyces and a couple of pathogenic bacteria, whose genome sequences were wholly clarified.
Previously, I investigated the characteristics of 59 β-lactamase candidate sequences among 529 putative β-lactamase and PBP sequences in 19 Actinobacteria [78,79], which were collected from the NCBI database [80]. There were some characteristic features in the 59 candidate sequences. First, β-lactamase candidates were not distributed uniformly in the strains: while Saccharomonospora viridis, Streptomyces avermitilis, Streptomyces scabiei and Streptosporangium roseum possess as many as five, twelve, five and five candidate genes, respectively, Mycobacterium leprae, Nocardioides sp. JS614 and Rhodococcus erythropolis had no candidate genes. In general, Streptomyces species produced β-lactamase candidates at higher rates. However, the numbers of candidate sequences were not necessarily related to the taxonomic classification. For example, although Streptosporangium and Nocardioides were closely related taxonomically [81], the numbers of β-lactamase candidates in Streptosporangium were five but those in Nocardioides were zero. Even in the same genus, M. tuberculosis produced four but M. leprae produced none. β-Lactamases detected by their enzymatic activity also distributed independently of actinobacterial taxonomy [82]. These results indicate strongly that β-lactamases evolve and are produced independent of taxonomy.
Second, most of the β-lactamases detected so far by their enzymatic activity in Gram-positive bacteria belong to class A and a few belong to class B. However, this fact was not the case with whole genomic level.
Third, it is intriguing that Thermomonospora curvata, a high G + C Gram-positive soil thermophilic Actinobacteria, produced two class B and one class D β-lactamase candidates but no class A and class C enzymes. Usually, Gram-positive bacteria produce class A and class B β-lactamases, but not class C and class D. In addition, T. curvata DSM43183 Tcur2040 β-lactamase is closely related to Gram-negative class D β-lactamases, especially OXA-5 and OXA-27 type β-lactamase [79], although T. curvata is a Gram positive bacteria. Hereafter, genetic analyses are extended to β-lactamases of other 12 Streptomyces species listed in Table 1.

Class A β-lactamases
The amino acid sequences of putative 25 class A β-lactamases together with a reference sequence (TEM-1, Klebsiella penumoniae plasmid, YP_005351445) were aligned by using the Muscle and T-coffee programs [83,84]. The results showed that all the sequences in Table 2 possessed main residues involved in the catalytic reaction (Table 3), except SCAT_1418, SCAT_4581, SRIM_07318, and F750_2387 (Table S1). These four sequences are defective in S 70 XXK 73 and S 130 DN 132 . Furthermore, SCAT_1418, SCAT_4581 and F750_2387 do not carry K 234 T/SG 236 , indicating that these candidates do not function as β-lactamases (Table 2).
Although some other sequences are also deficient or possess incompletely identical signature amino acid sequences, they are supposed to operate as β-lactamases even if the catalytic efficiency is not perfect. GWG 166  In addition blast analysis using SCAT_1418 as a query, metallo-dependent hydrolase of Streptomyces violaceusniger (WP_014060235) and metallo-dependent hydrolase of Streptomyces himastatinicus (WP_00971837) were identified. Actually, SCAT_1418 possesses some class B β-lactamase signature residues. Likewise, SCAT_4581 is highly similar to Zn-dependent hydrolases from Actinobacteria, so that they may be class B β-lactamases. SRIM_07318 is too short to function as a β-lactamase, but may be a part of class A β-lactamase, as its amino acid sequence of the C-terminal part is very similar to that of class A β-lactamases from Streptomyces albus (WP_060729068) and Rhodococcus rhodnii (WP_037260801). F750_2387 is a large membrane protein and belongs to metallo-β-lactamase superfamily. The fact that SCAT_1418, SCAT_4581 and F750_2387 belong to metallo-β-lactamases but not to class A reflects on large molecular distances between these β-lactamases and others such as TEM-1 and SHV-1 (Table S2). Interestingly, the amino acid sequence of SRIM_10531 is completely identical with that of Streptomyces cellulosae (D12653). This is one example of the horizontal gene transfer. Another interesting thing is that the amino acid sequences of B446_02285 and B446_33010 in class A β-lactamases, and B446_02210 and B446_33085, and B446_02000 and B446_33290 in class B β-lactamases, and B446_01720 and B446_33570 in β-lactamase-like sequences and B446_01315 and B446_33975 in class C β-lactamases are also completely identical, indicating that about 7.7 Mb region including these genes were duplicated in the chromosome in the past. In fact, two transposases exist at both sides in reverse direction from 7,984,885-7,985,573n and 287,352-288,040n in Streptomyces collinus genome, respectively, suggesting that these transposases acted in the transposition or the duplication.
A phylogenetic tree of putative 20 class A β-lactamases constructed with those of S. aureus (M15526), P. aeruginosa (Q03170), K. pneumoniae (TEM-1, YP_005351445), K. pneumoniae (SHV-11, X98101), and E. coli (SHV-1, AF148850) as reference sequences disclosed that although these reference sequences are originated from quite different species, they form only one cluster (cluster C in Figure 1). In contrast, β-lactamases of Streptomyces are scattered in various branches, indicating that only a few β-lactamase genes of Streptomyces species have been transferred to pathogenic bacteria at the present time.
: presence; X: absence; : probable; other symbols are for amino acid residues.  The amino acid numbering follows Ambler [65] for classes A and D, Galleni et al. [85] for class B, and Lobkovsky et al. [86] for class C β-lactamases, and Goffin and Ghuysen [87] for PBPs.
Molecules 2016, 21, 605 7 of 31 The amino acid numbering follows Ambler [65] for classes A and D, Galleni et al. [85] for class B, and Lobkovsky et al. [86] for class C β-lactamases, and Goffin and Ghuysen [87] for PBPs.  as reference sequences. The tree was constructed by using ClustalX 2 [88] and E. coli AmpC (ABM69263) as outgroup. Clusters A and B show blue-dextran/NADP + -non-binding and binding β-lactamases from Streptomyces, respectively and, cluster C shows those from pathogenic bacteria used as reference sequences. The bootstrap probabilities are shown at branching nodes.
However, it is possible that other β-lactamase genes of Streptomyces species may transfer to pathogenic bacteria in the near future, although many threatening class A genes may be evolved through mutation but not transferred to pathogenic bacteria, as it is seen in TEM-type and SHV-type β-lactamases [89]. Actually, molecular distances determined by the protdist program [90] between most class A β-lactamases of Streptomyces and those of pathogenic bacteria (Table S2) are not so large as in class B and class C β-lactamases, indicating that these class A β-lactamases are related closely with each other and with those of pathogenic bacteria than class B and class C enzymes. It is speculated, therefore, that the horizontal gene transfer of class A β-lactamase genes is easier than for class B and class C β-lactamases. The β-lactamases of Streptomyces can be classified into blue-dextran/NADP + -binding and non-binding groups [91,92]. Blue-dextran/NADP + -non-binding β-lactamases belong to cluster A and those of binding β-lactamases go to cluster B in the phylogenetic tree ( Figure 1). However, the role of NADP+-binding of β-lactamases remains to be clarified.
As for the relationship between class A β-lactamases and the β-lactam biosynthetic gene cluster, SSCG_00130 (corresponding to SCLAV_4216, partial stop and SCAT_5692) is located in the terminal region of the cephamycin biosynthetic gene cluster and SSCG_00160 (corresponding to SCLAV_4187) is located within the clavulanic acid biosynthetic gene cluster in S. clavuligerus ( Figure S1). These two clusters exist in the adjacent position. SSCG_01467 (corresponding to SCLAV_2436) is not related to β-lactam biosynthetic genes [78]. Similarly, SCAT_5692 is present in the terminal region of the cephamycin biosynthetic gene cluster in S. cattleya ( Figure S2). However, a gene corresponding to SSCG_00160 in S. clavuligerus is missing in S. cattleya in accord with the absence of clavulanic acid biosynthesis. Therefore, it is suggested strongly that class A β-lactamases and β-lactam biosynthesis are closely related with each other. However, it remains to be clarified whether these β-lactamases are involved in the self-resistance in the Streptomyces species. SSCG_00160 (SCLAV_4187) is also thought to be involved in clavulanic acid biosynthesis [93,94].

Class B β-lactamases
On the basis of phylogenetic and sequences analyses, 63 class B β-lactamase candidates were selected in 12 Streptomyces species. The amino acid sequence alignment using the Muscle and T-coffee programs [83,84] revealed that all the sequences contain the conserved active site residues (HXHXD 120 ) except SSCG_05130, SHJG_8335, SCAT_4145, and SRIM_16845 (Table 4 and Table S3; In the case of AAA22276_Bacillus cereus reference sequence, the active site residues are shifted to downstream in Table S3).
: presence; X: absence; : probable; other symbols are for amino acid residues.
Moreover, comparing to Class A and class C β-lactamases, molecular distances calculated by using the protdist program [90] are mostly quite large between those of Streptomyces species with each other and those with pathogenic bacteria (Table S4), suggesting that the class B β-lactamase enzymes are assemblies of highly diverged proteins. Recently, new type of metallo-β-lactamases such as New Delhi metallo-β-lactamase (NDM) have emerged and swiftly spread worldwide. It is possible that various metallo-β-lactamases of Streptomyces are also transferred to pathogenic bacteria and pose a similar public health threat.

Penicillin-Binding Proteins (PBPs)
The bacterial cell wall peptidoglycan is a three-dimensional, covalently closed, net-like mesh called sacculus in which glycan strands are cross-linked by peptide chains. It maintains cell shape and provides mechanical strength to resist osmotic pressure [96,97]. The peptidoglycan chains are composed of alternating β-1,4-linked N-acetylglucosamine and N-acetylmuramic acid residues. The carboxyl groups of the N-acetylmuramic acid residues are involved in amide bond to terminal L-alanine residues of the peptide units L-alanyl-γ-D-glutamyl-L-diaminoacyl-D-alanine. The glycan chain is biosynthesized by catalysis of glycosyltransferases and the cross-linkage of the peptide chains between two adjacent glycan chains is catalyzed by transpeptidases [87,98]. The transpeptidases, also called penicillin-binding proteins (PBPs), were initially identified as their ability to bind penicillins [99][100][101]. Depending on the structure and the catalytic activity of their N-terminal domain, they are classified into class A, class B and class C PBPs [87,96,101,102]. Both class A and class B PBPs have the transpeptidase activity in the C-terminal region. However, the N-terminal domain in class A PBPs is responsible for their glycosyltransferase activity, while the glycosyltransferase domain is lacking in class B PBPs. Instead, the PBP dimer domain (Pfam03717) is present at the N-terminal region in class B PBPs. However the detailed functional role of the dimer domain is yet to be fully explicated. Class C PBPs are also called low molecular weight PBPs and, having the carboxypeptidase activity, are responsible for the maturation and recycling of the peptidoglycan [96], but they are not essential and are excluded from further study. There are many excellent review paper on PBPs [87,[96][97][98][100][101][102], and the comprehensive review on the PBPs in Actinobacteria in general was already published [103], so this review describes the PBPs in Streptomyces, especially emphasizing their roles in self-resistance.

Class A PBPs
The numbers and types of putative PBPs in 24 Streptomyces are summarized in Table 6. The amino acid sequences of putative 100 class A PBPs in 24 Streptomyces species together with a reference sequence (Q04704/CVP62105, S. pneumoniae) were aligned by the Muscle and T-coffee programs (Table  S7). All sequences analyzed possess active site S 465 XXK 468 residues, excluding SAV_4423, SBI_09068, STRVI_3845, SCAB_64431, SZN_18682, DC74_7647, SSQG_02328, SSFG_02387, SAV_7219, SHJG_3853, SRIM_08328, SBI_03076, and STRVI_2314. However, active site residues S 524 XN 546 , and K 716 TG 718 are mostly conserved. In fact, blast analysis of STRVI_3845 as a query showed that glycosyltransferases of S. iranensis (WP_044567878), S. hygroscopicus (WP_060948597) and S. bingchenggensis (WP_014181633) are identified and, that of SSQG_02328 as a query, PBPs of S. afghaniensis (WP_037667886), S. iakyrus (WP_033306427), and S. caeruleatus (KUN93289) are identified, and that of SHJG_3853 as a query, PBPs of S. antibioticus (WP_053211715), S. reticuli (CUW28219), and S. achromogenes (WP_030604457) are identified. In addition, the glycosyltransferase and transpeptidase domains are preserved in these sequences, indicating that these sequences can function as PBPs. A phylogenetic tree was constructed by using SCO4049 (putative penicillin acylase) as outgroup (Figure 4). The class A PBPs are clearly divided to three groups: A, B and C. SAV_4423 is a lone sequence. The PBPs in these groups are obviously corresponding to those in clusters A, B and C in the amino acid alignment (Table S7).   (Table S8). The PBPs in groups A, B, and C are corresponding to those in clusters A, B, and C in the amino acid alignment shown in Table S7. The bootstrap probabilities are shown at branching nodes.
According to the penicillin-binding cores of PBPs, class A PBPs are classified into two subclasses in Gram-negative bacteria and three subclasses in Gram-positive bacteria [104]: A1 (whose prototype is Escherichia coli PBP1A, PBPA_ECOLI), A2 (whose prototype is E. coli PBP1B, PBPB_ECOLI), A3 (whose prototype is Streptococcus pneumoniae 1A, PBPA_STRPN), A4 (whose prototype is S. pneumoniae 2A, AJ002292), and A5 (whose prototype is S. pneumoniae 1B, AJ002291). Interesting enough, all class A PBPs from Streptomyces analyzed in this paper form a completely different cluster in the phylogenetic tree from these five subclasses (Figure 4), where E values (number of alignments expected by chance) between PBPs in subclasses A1 to A5 and those in Streptomyces are in the range of 3.4ˆ10 -16 to 4.9ˆ10 -31 , indicating low similarities. These results suggests strongly that the gene transfer and/or gene conversion occurred rarely between PBPs in Streptomyces and those in Gram-positive and Gram-negative bacteria. However, the molecular distances are not so large except PBP of Helicobacter pylori and PBPs of Streptomyces, indicating that these PBPs are not so different as speculated by the phylogenetic tree with each other (Table S8), and it is possible that the horizontal gene transfer of these genes may take place from Streptomyces to pathogenic bacteria near future.

Class B PBPs
One hundred sixty one class B PBPs in 24 Streptomyces listed in Table 6 were analyzed. Intriguingly, more than half of Streptomyces species carry two successive class B PBPs in tandem. For example, XNR_2096 and XNR_2097 (E value is 5.4ˆ10 -58 , the same hereafter), SAV_3603 and SAV_3604 (3ˆ10 -68 ), SCLAV_4178 and SCLAV_4180 (1.5ˆ10 -11 ), SMCF_7795 and SMCF_7796 (3.3ˆ10 -60 ), SCO3156 and SCO3157 (2.8ˆ10 -47 ), BN159_5121 and BN159_5122 (1.4ˆ10 -62 ), SSFG_04216 and SSFG_04217 (1.3ˆ10 -56 ), SSRG_03705 and SSRG_03706 (2.7ˆ10 -62 ), SHJG_4627 and SHJG_4628 (2.8ˆ10 -69 ), SLIV_21910 and SLIV_21915 (4.3ˆ10 -58 ), SCAB_53611 and SCAB_53621 (6.8ˆ10 -53 ), SSEG_00010 and SSEG_00011 (3.8ˆ10 -63 ), and SSQG_03242 and SSQG_03243 (5.5ˆ10 -65 ). These amino acid sequences are not only very similar to each other, but also all the sequences belong to the same cluster in a phylogenetic tree constructed by using SCO4049 as outgroup (cluster C in Figure 5). In fact, the amino-acid sequence identity and similarity of PBPs in cluster C in the phylogenetic tree are in the range of 49.2%-51.8% and 71.8%-77.8%, respectively. Moreover, the nucleotide sequences of each pair are arrayed in the same direction, indicating that they were duplicated and transferred to each other very recently. Interestingly enough, the pair of a cephamycin and clavulanic acid producer, that is, S. clavuligerus SCLAV_4179 and SCLAV_4180 is an exception. The tree was constructed by using ClustalX 2 [88] and S. coelicolor putative penicillin acylase (SCO4049, CAC32320) as outgroup. The sequences marked with yellow boxes were used for the calculation of molecular distances (Table S9). The bootstrap probabilities are shown at branching nodes. The phylogenetic tree were arbitrary divided to three groups, A, B, and C. SCLAV_4179 belong to cluster B and SCLAV_4180 pertain to cluster A in the phylogenetic tree and the similarity of the amino-acid sequences is very low (E-value is 1.5ˆ10 -11 ). This character may be related to β-lactam production and these two PBPs seem to behave independently, because these two sequences are aligned in the reverse direction, i.e., SCLAV_4180 (pbpA) acts together with the clavulanic acid biosynthetic gene cluster, while SCLAV_4179 (pbp2) goes on with SCLAV_4178 (cytochrome P450, partial start) and SCLAV_4177 (RNA polymerase) or independently [78]. Moreover, PBP SCLAV_4179 in S. clavuligerus is reported to have a low affinity to β-lactam antibiotics and is essential to the growth [105]. Therefore, it is assumed that PBP SCLAV_4179 is involved in the self-resistance. SCLAV_4198 (class BPBP, pcbR) is located in the border between cephamycin and clavulanic acid biosynthetic gene clusters but behaves independently from these clusters, considering the gene direction. In S. cattleya which produces only cephamycin but not clavulanic acid, genes corresponding to SCLAV_4180 and SCLAV_4198 are deleted, but a gene (SCAT_5676) corresponding to SCLAV_4179 and responsible for self-resistance is retained [78]. This fact supports the above ideas ( Figure S2).
Related to self-resistance, Ogawara and Horikawa reported nearly 40 years ago that β-lactamproducing Streptomyces species such as S. clavuligerus possessed PBPs of very low affinity to benzylpenicillin [106]. Later, two PBPs in S. clavuligerus, that is, SCLAV_4179 and SCLAV_4198 were reported to have low affinity to penicillins [105,107]. A mutant disrupted in SCLAV_4198 gene exhibited a significant decrease in its resistance to benzylpenicillin and cephalosporins [107]. Moreover, a probe containing SCLAV_4198 hybridized to genomic DNAs from β-lactam producers, S. jumonjinensis NRRL 5741, S. griseus NRRL 3851 and S. lipmanii NRRL 3584, suggesting that SCLAV_4198-like sequences and SCLAV_4198-mediated resistance mechanisms are likely to be present in these β-lactam-producing species. Likewise, most Streptomyces species have two PBPs in SCLAV_4198-belonging cluster C in the tree.
These characters are moderately similar with each other: SRIM_04191 and SRIM_06646 (1.4ˆ10 -59 ), SCLAV_4198 and SCLAV_2276 (8.1ˆ10 -27 ), F750_6320 and F750_2998 (5.6ˆ10 -29 ), STRVI_1135 and STRVI_3190 (1.2ˆ10 -61 ), and SBI_04376 and SBI_06233 (7.7ˆ10 -36 ) are examples. Together with the fact that tandem two genes in most Streptomyces species are associated with cluster C as described above, these results clearly indicate that most Streptomyces species are firmly defended by the presence of two low-affinity PBPs. Moreover, cluster B PBPs, to which SCLAV_4179 belong, reconfirm the self-resistance [103], that is, two pairs of low affinity PBPs (e.g., SCLAV_4198 and SCLAV_2276 in cluster C in Figure 5, and SCLAV_4179 and SCLAV_1774 in cluster B in Figure 5) guard Streptomyces from the attack of β-lactam antibiotics.

PASTA Domains: Connector between Penicillin-Binding Proteins and Protein Kinases
Protein phosphorylation was first described as a major regulatory mechanism in eukaryotes [108,109]. Later, the regulation by phosphorylation with serine/threonine protein kinases (STPKs) in particular has been expanded to prokaryotes [110][111][112][113]. While investigating the S. coelicolor homolog of PknB, a STPK, Yeats et al. identified a definitive domain in its C-terminus. This domain is also found in C-terminus of high molecular weight PBPs, so that it is termed as the PASTA domain (penicillin-binding protein and serine/threonine kinase-associated domain) [114]. PASTA domains are not found in eukaryotes and are restricted to firmicutes and Actinobacteria. Interestingly, some PASTA domains in PBPs such as PBP2x of S. pneumoniae bind peptidoglycan and β-lactam antibiotics [115], but others such as PonA2 of M. tuberculosis do not [116], indicating that PASTA domains in PBPs operate in a species-specific manner. On the other hand, PASTA domains in STPKs bind peptidoglycan and β-lactam antibiotics, and STPKs with PASTA domains are reported to play important roles in the regulation of bacterial cell division, and morphogenesis [111,112,114,117,118]. Previously, I described the distribution of PASTA domains in Actinobacteria [119], so in this review I discuss the interaction of PASTA domains in PBPs and STPK from the point of view of self-resistance.
In sharp contrast to other bacteria such as B. subtilis, Clostridium perfringens and S. pneumonia, PBPs with PASTA domains are detected only in class A PBPs but not in class B PBPs of Actinobacteria. In addition, none of Streptomyces species has PBPs with PASTA domain. Class A PBPs have both transglycosylase and transpeptidase domains, whereas class B PBPs have only transpeptidase domain. Therefore, it is interesting to know the interaction between transglycosylase and PASTA domains in Actinobacteria. However, it is not interpreted yet. Related to this fact, one of class B PBPs without PASTA domain and a STPK with four PASTA domains are located in adjacent position in most Streptomyces species (Table 6). For example, DC74_4185 and DC74_4186, XNR_3038 and XNR_3037, SAV_4339 and SAV_4338, SBI_05407 and SBI_05406, SCO3847 and SCO3848, and others. These PBPs belong to cluster Ac in a phylogenetic tree ( Figure 5). The genomic structures in PBP-STPK regions of S. coelicolor, S. clavuligerus, and S. griseus are very similar and suggest that STPK, PBP and FtsW/RodA family protein form an operon (Table 7). In S. coelicolor, this family of PBP (i.e., SCO3847) is reported to form the Streptomyces spore wall synthesizing complex (SSSC) [120,121]. Although they are supposed to be transported to peptidoglycan biosynthesis machinery through the PASTA domains of STPK (i.e., SCO3848) and be involved in the peptidoglycan biosynthesis and/or morphogenesis coupled with PBPs and FtsW/RodA family proteins, Jones et al. reported that SCO3848 (PknB) regulates the timing of development and TCA cycle favoring antibiotic production together with two forkhead-associated proteins (FHA, SCO3843 and SCO3844) [122]. In addition, SCO3848 (PknB) are supposed to play coordinately with the protein phosphatase (i.e., SCO3845) inside the operon. It is very interesting to know how self-resistance is implicated in this reaction, because the identity and the similarity of PBPs of this group (cluster Ac in Figure 5) to SCLAV_4180 are in the range of 40%-91%, and 70%-95%, respectively and those of SCLAV_2947 (a low affinity PBP) and SCLAV_4180 are 53.3% and 80.2% in 497 amino acid residues, respectively (E value is 1.7ˆ10 -102 ). The indication that clusters Aa, Ab, and Ac PBPs in Figure 5 are close together with each other is confirmed by the molecular distance determination (Table S9). Cluster A PBPs in Figure 5 are also closely related with PBPs of pathogenic bacteria (Table S9). Unfortunately, however, the role of cluster A PBPs in self-resistance remains to be elucidated.

Relationship between β-Lactam Biosynthetic Gene, β-Lactamase and PBP: Additional Comments
The relationship between β-lactam biosynthetic gene clusters, β-lactamases, and PBPs of S. clavuligerus and S. cattleya was shown in Figure S1 [33,[123][124][125][126][127]. The genetic organization of clavulanic acid biosynthetic gene cluster of S. clavuligerus was published [93,128]. Comparison of cephamycin gene clusters in S. clavuligerus and S. cattleya indicates clearly that clavulanic acid gene cluster from SCLAV_4180 to SCLAC_4198 in S. clavuligerus was inserted between SCAT_5676 and SCAT_5678 of S. cattleya ( Figure S2). SCAT_5677 is missing. Interestingly, both sides of the clavulanic acid gene cluster are occupied by PBPs (SCLAV_4180 and SCLAV_4198), indicating that these PBPs behave together with clavulanic acid/cephamycin gene cluster and are involved in the self-resistance of S. clavuligerus. Similar gene constructs for cephamycin biosynthesis are observed in Nocardia lactamdurans, Lysobacter lactamgenus, Penicillium chrysogenum, and Acremonium chrysogenum [129][130][131][132]. However, β-lactamase and PBP genes can be detected neither in β-lactam biosynthetic gene clusters nor in whole genomes of P. chrysogenum, and A. chrysogenum. On the other hand, the organizations of clavulanic acid gene clusters of three Streptomyces species, S. clavuligerus, S. flavogirseus, and S. viridis are also similar with each other. Unfortunately, however, relationship between β-lactam biosynthesis, β-lactamases, and PBPs remains to be clarified, although β-lactamase was reported to be expressed during the active growth phase, prior to the formation of β-lactam antibiotics [129]. Recently, an interesting paper on taxonomy, physiology, and natural products of Actinobacteria was published [133].

Conclusions
Although antibiotics still play a key role for the prevention of microbial infections as remaining treasures from the twentieth century, antibiotic resistance is prevailing and putting us in a critical situation. Furthermore, it is said that the post-antibiotic era is coming soon [134]. As described in this paper, the β-lactamases in antibiotic-producing Streptomyces are diverse in their characteristics, and the PBPs are multiplexed in their guard systems. In addition, as antibiotic resistance mechanisms are supposed to originate and evolve in their producing microbes, and be transferred to pathogenic bacteria by transformation, transduction, transfection and/or conjugation, the public health crisis will be getting worse and worse. β-Lactamases and PBPs are two major resistance mechanisms in pathogenic bacteria against β-lactam antibiotics, the most frequently used antibiotics for infectious diseases at the present time. Moreover, from about 40 years ago, the rate of discovery of new antibiotics has declined rapidly and many large pharmaceutical companies have abandoned research and development on antibiotics [135]. To avoid this situation, therefore, it is urgently needed to make the public and the government recognize the situation. In addition, new antibiotics should be screened by using various technologies such as activation of cryptic gene clusters for antibiotic biosynthesis, metagenomics mining, combinatorial biosynthesis, target-directed computer-aided chemical synthesis, systems biology, genetic manipulation, omics, and so on [136][137][138][139][140][141]. Furthermore, combination therapies with monoclonal antibodies and vaccines are also required. When these requirements are attained, "New Golden Age" is possible [141]. However, it is worth mentioning again that the resistance mechanisms in Streptomyces species are very hard to deal with as described in this paper.