Several thermophilic bacteria species have been reported to catalyze distinct reactions of steroid structural modifications, while sterol degradation by thermophilic microorganisms has not been studied so far. As shown in this research, the strain of S. hirsuta fully utilized cholesterol (Fig. 1), and the degradation pathway was predicted (Fig. 2) based on the time-courses of the intermediates (Fig. 1) and genome-wide bioinformatics analysis (Fig. 3).
The set and the order of the genes putatively involved in steroid catabolism in the clusters of S. hirsuta genome are similar to the clusters described for the reference actinobacteria: M. tuberculosis H37Rv and R. jostii RHA1 (Olivera and Luengo 2019) (Fig. 3). In general, cluster 1 in S. hirsuta is similar to the cluster of the sterol catabolic pathway in M. tuberculosis H37Rv (Fig. 3). The differences are that hsa genes in S. hirsuta are represented as a “complete” operon hsaBCDAFGE, while in M. tuberculosis H37Rv (as well as in R. jostii RHA1) the genes of this block are divided into two parts hsaFGE and hsaBCDA within the cluster genes (Fig. 3). The orthologs of kshAB in cluster 1 in the S. hirsuta genome locate close to each other (Fig. 3, Table S3); there is only one gene of an undefined function between them, in contrast to kshA and kshB from M. tuberculosis H37Rv, which are very far from each other in mycobacterial cluster of cholesterol catabolism (Fig. 3).
Together, clusters 2 and 3 of the S. hirsuta contain the vast majority of orthologs of the cholate cluster genes from R. jostii RHA1 participating in the bile acids rings A/B degradation, but not all orthologs of the genes from R. jostii RHA1 coding for the enzymes of the bile acids side chain degradation (Fig. 3, Table S3).
ChOs and 3-HSDs
In many actinobacteria, the sterol degradation pathway is known to begin with the modification of 3β-hydroxy-5-ene into the 3-keto-4-ene structure of the ring A by the action of cholesterol oxidases (ChOs) or 3β-hydroxydehydrogenases (3-HSDs) (Donova 2007; Yam et al. 2009), while cytochrome P450-mediated hydroxylation at C26(27) of the cholesterol side chain was reported to be initial reaction in the sterol degradation in Rhodococcus strains (Rosłoniec et al. 2009; Kreit 2017). As evidenced from the time course of cholesterol conversion by S. hirsuta the initial reactions of cholesterol degradation, i.e. modification of 3β-ol-5-ene-moiety of the steroid A-ring and hydroxylation of the sterol side chain at C26(27), occurred independently (Fig. 2). During first 24 h cholesterol (I) mainly transformed to cholestenone (II) and in a lesser extent – to 26-hydroxycholesterol (VII). Cholestenone (II) further subjected to the side chain hydroxylation at C26(27) to form the corresponding 3-keto-4-en-26-alcohol (IV). In turn, the 3β-ol-5-ene moiety in 26-hydroxycholesterol (VII) modified to the corresponding 3-keto-4-ene also resulting in the compound (IV).
Cholesterol oxidases are most likely involved in the 3β-hydroxyl group dehydrogenation and ∆5→∆4-isomerization (I→II, VII→IV) in S. hirsuta since no candidate genes responsible for the 3β-hydroxysteroid dehydrogenase (3-HSD) were found in the Ac-666T genome (Lobastova et al. 2020). Two candidate cho genes: choD F1721_14655 and choE F1721_09795 encoding the cholesterol oxidases were revealed in the genome and both are situated out of the clusters related to cholesterol catabolism in Ac-666T. It corresponds to the literature data evidencing that in most cases the genes coding for cholesterol oxidases in actinobacteria located out of the clusters of the genes that involved in cholesterol catabolism (Kreit 2017).
No orthologs of 3β-HSD were identified also in the genome of N. simplex VKM Ac-2033D, and highly efficient oxidation of different sterols (cholesterol, sitosterol, stigmasterol, campesterol) to the corresponding stenones (cholestenone, sitostenone, stigmastenone, campestenone) is attributed to the activity of cholesterol oxidase in this strain (Shtratnikova et al. 2021).
On the other hand, the presence of both cholesterol oxidases and 3β-HSDs have been identified in the genomes of mycobacterial strains (Uhia et al. 2012; Bragin et al. 2013). The functionality and role of the enzymes can differ in different actinobacteria. As shown for Mycolicibacterium neoaurum ATCC 25795 (syn. Mycobacterium neoaurum), two cholesterol oxidases (ChoM1 and ChoM2) are accounting for the 3β-ol-5-ene to 3-keto-4-ene modification and essential for the cell growth on cholesterol (Yao et al. 2013). In contrast, the knock-outs of the cholesterol oxidases ChoD (orthologous to ChoM1) in other relative mycobacteria (M. neoaurum VKM Ac-1815D, M. smegmatis mc2 155) did not block cholesterol to cholestenone oxidation (Uhia et al. 2011; Ivashina et al. 2012) thus evidencing the presence of other enzymes involved in 3β-ol-5-ene to 3-keto-4-ene modification such as 3β-HSDs.
Most of the actinobacterial ChOs are of the dual functions: catalyze both oxidation of the 3β-hydroxy group and Δ5→4 isomerization of 3β-hydroxy-5-ene steroids. A separate Δ5-3-ketosteroid isomerase, encoded by the gene ksi in Comamonas testosteroni was shown to be responsible for the Δ5→4 isomerization (Horinouchi et al. 2012). The genome of S. hirsuta contains two candidate genes ksi (ksdI). KsdI F1721_32675 is located between cyp125 F1721_32680 and ltp3-ltp4 F1721_32665-F1721_32660 in cluster 1, and another ksdI F1721_00740 is situated among the genes putatively involved in the A/B-rings oxidation: kshB3 F1721_00735 and hsaB3 F1721_00745 in cluster 2 (Fig. 3, Table S3). In the genome of N. simplex, only one of two ksdI genes, - KR76_23530 with an unclear function was up-regulated in the presence of phytosterol (Shtratnikova et al. 2021).
CYP 125
The cleavage of the cholesterol/cholestenone side chain by actinobacteria begins with hydroxylation of the terminal methyl group catalyzed by steroid 26(27)-monooxygenase to form the corresponding 26(27)-alcohols (Kreit 2017). As shown for R. jostii RHA1, the same enzyme is accounting for further oxidation to the corresponding carboxylic C26-oic acids (Rosłoniec et al. 2009). Cytochrome P450 monooxygenases encoded by cyp125 had been isolated and characterized from M. tuberculosis (Capyk et al. 2009b) and R. jostii RHA1 (Rosłoniec et al. 2009). Cyp125 from the clinical strain M. tuberculosis CDC1551 was shown to be involved in the oxidation of 26-hydroxycholest-4-en-3-one to cholest-4-en-3-one-26-oic acid (Ouellet et al. 2010). The genes cyp125, cyp142, cyp124 were reported to encode the enzymes performing terminal hydroxylation at C(26)27 (Kreit 2017).
The candidate gene cyp125 F1721_32680 was identified in the S. hirsuta genome (Figs. 3, 4). The gene product could be responsible for the modifications to 26-hydroxycholestenone, 3-oxocholest-4-ene-26-oic acid, 3-oxocholesta-1,4-diene-26-oic acid, 26-hydroxycholesterol, and 3β-hydroxycholest-5-ene-26-oic acid (Fig. 2) that have been identified among the intermediates/metabolites from cholesterol. Probably, this strain possesses a specific steroid 26(27)-monooxygenase capable of oxidizing the C27-sterol side chain regardless 3β-hydroxy-5-ene- or 3-oxo-4-ene- (3-oxo-1,4-diene-) structure of the ring A. No orthologs of cyp124, or cyp142 were identified in S. hirsuta genome.
Side chain degradation
As well-established for many actinobacteria, the aliphatic side chain of sterols is degraded through a cascade of reactions similar to the β-oxidation of fatty acids. The C26-oic acid is activated by the coenzyme A (CoA) and then the side chain is shortened with release of two propionyl-CoAs and one acetyl-CoA (Fig. 4).
The genes chsE4 (fadE26) and chsE5 (fadE27) were shown to encode acyl-CoA dehydrogenases forming a complex that oxidizes 3-oxo-cholest-4-ene-26-oyl-CoA (Yang et al. 2015). ChsE4-ChsE5 complex of M. tuberculosis H37Rv was shown to take on the function of the blocked ChsE1-ChsE2 enzymes. The strain also synthesized ChsE3 specifically catalyzing the oxidation of 3-oxochol-4-ene-24-oil-CoA in the second round of β-oxidation of the cholesterol side chain (Yang et al. 2015). The orthologous genes, namely, chsE1 (F1721_33645), chsE2 (F1721_32785), chsE3 (F1721_28750), chsE4 (F1721_32605) and chsE5 (F1721_32610) were found in the genome of S. hirsuta (Figs. 3, 4).
The steroid metabolite with the C5 carbon side chain is ligated further by the steroid-24-oyl-coenzyme A synthetase. The phylogenetic analysis of more than 70 acyl-CoA synthetases aimed on the elucidation of their physiological role revealed four different types of the acyl-CoA synthetases from R. jostii RHA1 and M. tuberculosis H37Rv which were specific to the chain length of steroids (Casabon et al. 2014). FadD19 from M. tuberculosis H37Rv activated cholesterol metabolites with the C8 steroid side chain, whilst FadD17 from M. tuberculosis H37Rv ˗ the C5- or longer side chain, and CasG from R. jostii RHA1 ˗ the C5 cholate side chain. The metabolites with the C3 side chain accumulated during the cholate oxidation by R. jostii RHA1 were activated by the steroid-22-oyl-CoA synthetase CasI (Casabon et al. 2014). The orthologs of fadD19 (F1721_32635), fadD17 (F1721_32615), сasG (F1721_02405) and сasI (F1721_28770) encoding isofunctional acyl-coenzyme A synthases were revealed in the genome of S. hirsuta (Fig. 3). Probably, the presence of the homologous genes encoding various acyl-coenzyme A synthases of cholesterol and cholate catabolic pathways in S. hirsuta contributes to the adaptation of the thermophile microorganism in nature.
A special function of acyl-CoA synthetase FadD19 was reported to consist in its participation in the degradation of C24-branched sterols (sitosterol, stigmasterol, and others) as it was shown for R. rhodochrous (Wilbrink et al. 2011). The same function can be assumed for the fadD19 ortholog in the S. hirsuta (Fig. 4).
As shown for R. rhodochrous RG32, the oxidative decomposition of the phytosterols with the branched β-sitosterol-like side chain is mediated by the aldol lyases encoded by ltp3 and ltp4 (Wilbrink et al. 2012). The candidate genes ltp3 (F1721_32665) and ltp4 (F1721_32660) putatively involved in the degradation of sterols with the C24-branched side chain have been identified also in S. hirsuta (Figs. 3, 4).
The enoyl-coenzyme A hydratase encoded by echA19 gene acts at the early stage of sterol side chain degradation (Van der Geize et al. 2007). The Hsd4A might act as a dehydrogenase at the early stage of the degradation of the unsubstituted sterol side chain, and as a 17β-hydroxysteroid dehydrogenase at the last step of the side chain cleavage, or as a D-3-hydroxyacyl coenzyme A dehydrogenase at the branched fatty acids degradation (Van der Geize et al. 2007). The candidate gene echA19 (F1721_32640) and two distant homologs of hsd4A: F1721_32600 and F1721_33680 were revealed in the S. hirsuta genome (Fig. 3). Interestingly, both hsd4A genes were detected in cluster 1; and each hsd4A gene is located at the beginning or at the end of a group of genes encoding the sterol aliphatic side chain cleavage in the cluster. Due to distant homology of two Hsd4As (coding by F1721_32600 and F1721_33680) with a.a. identity of the proteins about 49%, one can assume different substrate specificity for these S. hirsuta isoenzymes towards the steroids with distinct side chain length.
The role of thiolase FadA5 at the last cycle of the cholesterol side chain β-oxidation was demonstrated for M. tuberculosis H37Rv (Schaefer et al. 2015). Orthologous fadA5 (F1721_32685) is present in the genome of S. hirsuta (Fig. 3).
Steroid/lipid transport system
In M. tuberculosis strains the operon that contained the genes for a putative lipid transfer protein (ltp2/Rv3540c), 2 MaoC-like hydratases (chsH1/Rv3541c and chsH2/Rv3542c), 2 acyl-CoA dehydrogenases (fadE29/chsE2/Rv3543c and fadE28/chsE1/Rv3544c), and cytochrome P450 (cyp125/Rv3545c) had been reported to be essential for virulence (Thomas et al. 2011). The orthologous genes ltp2 (F1721_32770), fadE28/chsE1 (F1721_33645), fadE29/chsE2 (F1721_32785), chsH1 (F1721_32775) and chsH2 (F1721_32780) were also found in the genome of S. hirsuta (Fig. 3). Unlike other actinobacteria, a transposon element (contig VWPH01000033.1 with length 1457 bp) is most likely located between the fadE28/chsE1 (F1721_33645) and fadE29/chsE2 (F1721_32785) in the thermophile Ac-666T.
Steroid nucleus degradation
The key reactions in the only known 9(10)-seco pathway of steroid core degradation are 1(2)-dehydrogenation and 9α-hydroxylation (e.g. Donova and Egorova 2012). 1(2)-Dehydrogenation is known to be carried out by 3-ketosteroid Δ1-dehydrogenases (KstDs) (Itagaki et al. 1990). The presence of several KstDs with distinct activities have been reported for actinobacteria species (Bragin et al. 2013; Zhang et al. 2015; Shtratnikova et al. 2016; Guevara et.al 2017; Zhang et al. 2018). In the genome of the Ac-666T strain three putative KstDs were identified (Fig. 3, Table S3). The candidate kstD3 gene is situated in the cluster 1 (Fig. 3). Two other candidate kstDs, namely, kstD2 and kstD1 are located side by side in the cluster 2 (Fig. 3). As reported earlier, S. hirsuta effectively transformed androst-4-ene-3,17-dione (AD), 3β-hydroxy-5-en-17-one (DHEA) and 3β,7(α/β)-dihydroxy-5-ene-D-homo-lactones into the corresponding 1(2)-dehydrogenated derivatives thus evidencing high KstD activity (Lobastova et al. 2019).
In the present study, detection of the intermediates with a 3-keto-1,4-diene structure such as cholesta-1,4-dien-3-one (III) and 3-oxo-cholesta-1,4-diene-26-oic acid (VI) evidenced that the 1(2)-dehydrogenation can take place at the early stages of sterol catabolism in S. hirsuta (Fig. 2). As shown for M. neoaurum DSM 1381, KstD1, KstD2, KstD3 catalyze 1(2)-dehydrogenation of various steroid substrates at different stages of sterol degradation. The kstD1 was highly up-regulated in response to phytosterol, while recombinant KstD2 exhibited a higher enzymatic activity towards the substrates without, or with a short side chain such as AD, or 22-hydroxy-23,24-bisnorchol-4-en-3-one (Zhang et al. 2018). Probably, the presence of several KstDs might provide 1(2)-dehydrogenation of various steroids in S. hirsuta.
The S. hirsuta KstDs showed higher identity with the KstDs in N. simplex (syn. Pimelobacter simplex) as compared with the KstDs from mycolic-acid rich actinobacteria, - M. tuberculosis and R. jostii (Fig. 5). Highest level of the activity towards C21-3-keto-steroids has been demonstrated earlier for N. simplex KstD2 encoded by KR76_27125 (Shtratnikova et al. 2021) that is in close identity with S. hirsuta KstD2.
The 1(2)-dehydrogenase of M. tuberculosis, which is related to the pathway of cholesterol catabolism, is in the same clade with KstD3 from S. hirsuta, while S. hirsuta KstD1 is more similar to the corresponding enzymes in N. simplex.
The KstD4 from S. hirsuta locates in the same clade with KstD4 (TesI) from N. simplex.
It is well-known that the 9α-hydroxylation is carried out by 3-ketosteroid 9α-hydroxylase KshAB consisting of the oxygenase component (KshA), and the reductase component (KshB) (Capyk et al. 2009a). Five different paralogous genes were reported to encode the KshA subunits in Mycolicibacterium fortuitum VKM Ac-1817D (syn. Mycobacterium sp. VKM Ac-1817D) (Bragin et al. 2013), thus providing 9α-hydroxylation of steroid metabolites at the different stages of sitosterol catabolism (Bragin et al. 2019). Several KshAs with different substrate specificity have also been found in R. rhodochrous DSM 43269: KshA1 was shown to participate only in the cholic acid catabolism while KshA5 could hydroxylate several substrates (Petrusma et al. 2011). The number of the genes coding for the kshA or kshB subunits depends on the strain species (Bragin et al. 2013; Shtratnikova et al. 2016). Herein, two orthologs of kshA and two orthologs of kshB were revealed in the genome of S. hirsuta (Fig. 3). One pair of the genes (kshA F1721_32745, kshB F1721_32755) is located far from the second pair (kshA F1721_00725, kshB F1721_00735). Most likely, the two KshABs might differ on their substrate specificity in S. hirsuta.
It should be noted that no any C19-steroid intermediates such as AD, ADD, testosterone, or 1(2)-dehydrotestosterone were detected during the cholesterol transformation by S. hirsuta. It could be explained either by their rapid degradation when their concentrations are below the level of detection, or ring A/B disruption of the “earlier” intermediate steroids (with preserved side chain at C17). For instance, at the bile acid transformation with Rhodococcus strains, the 9,10-seco-steroid intermediates with the partially degraded side chains were formed, evidencing that side chain degradation and opening of the ring B occurred simultaneously (Costa et al. 2013a, b). The order of 9α-hydroxylation and 1(2)-dehydrogenation of 3-oxo-4-ene-steroids resulting in the formation of the unstable 9α-hydroxy-3-oxo-1,4-diene-intermediates depends on the substrate specificity of the corresponding enzymes.
Steroid core degradation
The next step of steroid core destruction is hydroxylation at C4 in the A ring of 3-hydroxy-9,10-seco-androst-1,3,5(10)-triene-9,17-dione (3βHSA) by the flavin-dependent monooxygenase (HsaAB) resulting in the 3,4-dihydroxy-derivative (3,4-DHSA) (García et al. 2012). The detailed characterization of HsaAB was performed for the monooxygenase from M. tuberculosis (Dresen et al. 2010). The catabolic operon hsaBCDAFGE F1721_32700-F1721_32730 presumably involved in further degradation of the fragments of the ring A was identified in the genome of S. hirsuta. In addition, the candidate genes, namely, hsaA3 (F1721_00755), hsaB3 (F1721_00745), hsaC3 (F1721_00760) and hsaD3 (F1721_00695) orthologous to the R. jostii RHA1 hsaA3B3C3D3 genes were found in the strain Ac-666T (Fig. 3, Table S3). The candidate genes hsaF and hsaG have been revealed encoding HsaF and HsaG that putatively participate in the final stages of the ring A remnants degradation with formation of pyruvate and propionate (Figs. 3, 4).
Degradation of the C/D rings begins with the action of FadD3 which physiological role has been studied in M. tuberculosis (Casabon et al. 2013). Unlike mycolic acid rich M. tuberculosis H37Rv and R. jostii RHA1, as well as not-containing mycolic acids N. simplex, in which genomes fadD3 encoding the HIP-CoA synthetase lies in the corresponding cluster, ortholog of fadD3 in the S. hirsuta genome is located out of clusters (Fig. 3).
The fadE30 gene encoding the acyl-CoA dehydrogenase was shown to be involved in dehydrogenation at C4 of 5-OH-HIP (Van der Geize et al. 2011). The intermediate with the intact rings C and D: 5-OH-HIC-CoA is produced by two reactions catalyzed by IpdC which introduces a double bond in the C ring and IpdF that oxidizes the 5-OH group in M. tuberculosis. Crotonase Ech20 is responsible for the hydrolytic cleavage of the ring C to give HIEC-CoA. IpdAB encodes an enzyme that hydrolytically cleaves the C ring in the substrate COCHEA-CoA (Crowe et al. 2017). The candidate genes: ipdAB F1721_33690-F1721_33695, ipdC F1721_33700, fadE30 F1721_33715, and echA20 F1721_33685 found in the cluster 1 of the S. hirsuta genome represent the genes presumably involved in the degradation of the rings C and D (Fig. 3).
The product of the opening of both the C and D rings is subjected to the action of a putative thiolase FadA6 resulting in acetyl-CoA and 4-methyl-5-oxo-octanedioyl-CoA (Fig. 3) (Olivera and Luengo 2019). The last intermediate undergoes a β-oxidation with an acyl-CoA dehydrogenase FadE32, or by the Fad31-FadE32 complex in Mycobacterium (Crowe et al. 2017). Finally, the products of β-oxidation, an acetyl-CoA and 2-methyl-β-ketoadipyl-CoA would be released, followed by the formation of propionyl-CoA and succinyl-CoA (Fig. 4). The orthologs of fadE31 (F1721_33725), fadE32 (F1721_33730), fadE33 (F1721_33735) were detected in S. hirsuta genome (Fig. 3, Table S3).
Search for the key genes of steroid catabolism in the genomes of thermophilic/thermotolerant bacteria
In order to find out whether steroid degraders are widespread among the thermophile bacteria, the BLAST search for the key genes of the steroid catabolic 9,10-seco-pathway, – kstD and kshAB has been performed using 52 publicly available genomes of the thermophilic/thermotolerant strains. Only seven actinobacterial strains were proposed to be steroid degraders (Supplementary Table S4). The rest thermophilic/thermotolerant strains do not contain enzymes similar to the reference ones (KstD and KshAB of M. tuberculosis H37Rv) by more than 35% and, most likely do not degrade steroids.
The thermophilic strains of G. kaustophilus and P. thermoglucosidasius were reported to provide separate reactions of steroid modifications (Sideso et al. 1998; Al-Tamimi et al. 2010). The BLAST search of more than 20 enzymes related to steroid catabolism in these bacteria discovered putative proteins that are 47% and 45% similar to the reference FadA5, respectively, as well as enzymes of P. thermoglucosidasius that are similar to HsaF and HsaE of M. tuberculosis H37Rv by 48% and 41%, respectively (Supplementary Table S5). Taking into consideration that FadA5 is known to be involved also in fatty acid β-oxidation, the corresponding proteins in G. kaustophilus and P. thermoglucosidasius may not be intended for steroid catabolism. Products of hsaE and hsaF participate in the oxidation of a fragment of the steroid nucleus, which is a hydroxydiene-derivative of hexanoic acid, that means that the similar genes do not necessarily participate in the catabolism of steroid compounds. Taking into account the absence of other genes coding for steroid oxidation enzymes, most likely, these enzymes of G. kaustophilus and P. thermoglucosidasius are not associated with steroid catabolism, and the oxidation/reduction and hydroxylation reactions performed by the strains (Sideso et al. 1998; Al-Tamimi et al. 2010) are catalyzing by the enzymes that are non-specific towards steroids.