Hydrogen peroxide-independent production of α-alkenes by OleT JE P450 fatty acid decarboxylase

12 Pages·2014·0.96 MB·English

Liuetal.BiotechnologyforBiofuels2014,7:28
http://www.biotechnologyforbiofuels.com/content/7/1/28
RESEARCH Open Access
Hydrogen peroxide-independent production of
α
-alkenes by OleT P450 fatty acid decarboxylase
JE
Yi Liu1,2, Cong Wang1, Jinyong Yan1, Wei Zhang1, Wenna Guan1, Xuefeng Lu1 and Shengying Li1*
Abstract
Background: Cytochrome P450 OleT from Jeotgalicoccus sp. ATCC 8456, a new member of the CYP152
JE
peroxygenase family, was recently found to catalyze the unusual decarboxylation of long-chain fatty acids to form
α-alkenes using H O as thesole electron and oxygen donor. Because aliphatic α-alkenes are important chemicals
2 2
that can be used as biofuels to replace fossil fuels, or for making lubricants, polymers and detergents, studies on
OleT fatty acid decarboxylase are significant and may lead to commercial production of biogenic α-alkenes in the
JE
future, which are renewable and more environmentallyfriendlythanpetroleum-derived equivalents.
Results: We report the H O -independent activity of OleT for the firsttime.Inthepresence of NADPH and O , this
2 2 JE 2
P450 enzyme efficiently decarboxylates long-chain fattyacids (C to C ) in vitro when partnering with either the
12 20
fused P450 reductase domain RhFRED from Rhodococcus sp. or the separate flavodoxin/flavodoxin reductase from
Escherichia coli.In vivo,expression ofOleT orOleT -RhFRED indifferent E.colistrainsoverproducingfreefattyacids
JE JE
resulted in production of variant levels ofmultiple α-alkenes, with a highest total hydrocarbon titer of 97.6 mg·l-1.
Conclusions: The discovery of the H O -independent activityof OleT not onlyraises a number offundamental
2 2 JE
questions on themonooxygenase-like mechanism of this peroxygenase, but also will direct thefuture metabolic
engineering work toward improvementof O /redox partner(s)/NADPH for overproduction of α-alkenes by OleT .
2 JE
Keywords: Alkenes, Biofuels, Monooxygenase,P450 fatty acid decarboxylase, Peroxygenase
Background alkanes or alkenes, because they highly mimic the chem-
The urgency to develop sustainable fossil fuel alterna- ical composition and physical characteristics of petrol-
tives is driven by rapidly increasing global consumption, eum-based fuels [8-10]. Thus, the biosynthetic pathways
irreversibly diminishing reserves, unpredictable geopolit- for aliphatic hydrocarbons from diverse organisms have
ical factors, fluctuating price of crude oil, growing been attracting great attentions in recent years [11-13].
concerns about national energy security, and serious en- Particularly,thescalableandcost-effectivemicrobialbio-
vironmental concerns surrounding immense greenhouse synthesis of fatty alkanes or alkenes is considered as one
gas emissions mainly resulting from combustion of fossil ofthe mostpromising ways to produce‘drop-in compat-
fuels [1-3]. Biofuels produced from biological resources ible’ biofuels [5,8].
represent a compelling alternative to fossil fuels because To date, four microbial biosynthetic pathways that
they are renewable and more environmentally friendly convert free fatty acids or fatty acid thioesters into
[4-7]. Among various biofuels, bioethanol and biodiesel bio-hydrocarbons have been identified, including the
are dominating the current global market. However, it cyanobacterialpathwaysconsistingofanacyl-acyl carrier
is widely accepted that the ideal biofuels are bio- protein (acyl-ACP) reductase and an aldehyde decarbo-
hydrocarbons,especiallythemedium-tolong-chainfatty nylase, which together convert fatty acyl-ACPs into
alkanes [14]; a three-gene cluster responsible for gener-
ating alkenes with internal double bonds through the
*Correspondence:[email protected]
1KeyLaboratoryofBiofuels,ShandongProvincialKeyLaboratoryofEnergy head-to-head condensation of two fatty acyl-coenzyme
Genetics,QingdaoInstituteofBioenergyandBioprocessTechnology, A (acyl-CoA) molecules in Micrococcus luteus [15]; a
ChineseAcademyofSciences,No.189SonglingRoad,Qingdao,Shandong unique P450 decarboxylase OleT from Jeotgalicoccus
JE
266101,China
sp. ATCC 8456, which directly decarboxylates long-
Fulllistofauthorinformationisavailableattheendofthearticle
©2014Liuetal.;licenseeBioMedCentralLtd.ThisisanOpenAccessarticledistributedunderthetermsoftheCreative
CommonsAttributionLicense(http://creativecommons.org/licenses/by/2.0),whichpermitsunrestricteduse,distribution,and
reproductioninanymedium,providedtheoriginalworkisproperlycredited.TheCreativeCommonsPublicDomain
Dedicationwaiver(http://creativecommons.org/publicdomain/zero/1.0/)appliestothedatamadeavailableinthisarticle,
unlessotherwisestated.
Liuetal.BiotechnologyforBiofuels2014,7:28 Page2of12
http://www.biotechnologyforbiofuels.com/content/7/1/28
chain fatty acids to form α-olefins in presence of H O NADPH. We also found that E. coli flavodoxin (Fld) and
2 2
[16,17]; and a type I polyketide synthase from Synecho- flavodoxin reductase (FdR) are capable of supporting
coccus sp. PCC 7002 [18,19], which is capable of trans- OleT activity as well. Guided by these new findings, our
JE
forming fatty acyl-ACPs into α-olefins via sequential initialmetabolicengineeringeffortsbasedontheH O -in-
2 2
polyketide synthase chain elongation, keto reduction, dependentdecarboxylationoffattyacidsbyOleT invivo
JE
sulfonation mediated by its sulfotransferase domain, and led to a group of α-alkene overproducers with the best
the coupled hydrolysis and decarboxylation catalyzed by oneproducing97.6mg·l-1totalα-alkenes.
thethioesterasedomain.
Among these pathways, the single-step decarboxyl- Results
ation of fatty acids catalyzed by OleT P450 enzyme InvitrofattyaciddecarboxylationbyOleT and
JE JE
[16] apparently represents the simplest one. Moreover, it OleT -RhFRED
JE
directly uses free fatty acids instead of fatty acid thioe- The unusual activity of the first P450 fatty acid decarb-
sters as substrates, which is believed to be advantageous oxylase OleT was determined by reconstituted in vitro
JE
for metabolic engineering because fatty acids are more reactions [16]. Specifically, OleT was able to decarb-
JE
abundant and their abundance and composition can be oxylate stearic acid and palmitic acid to generate
well manipulated in E. coli [20-22], one of the most de- 1-heptadecene and 1-pentadecene, respectively, in the
veloped microbial cell factories. Thus, this P450 fatty presence of H O (Figure 1A). It also catalyzed the
2 2
acid decarboxylative machinery may hold great potential α- and β-hydroxylation of fatty acids as side reactions.
to be engineered into a biological α-alkene-producing Using the purified P450 enzymes (Additional file 1:
system. Notably, in addition to being biofuel molecules, Figure S1 and Additional file 2: Figure S2), we first con-
α-alkenes are also used broadly for making lubricants, firmed these three types of activity of OleT using myr-
JE
polymersanddetergents[17,23,24]. istic acid (C ) as substrate. Consistent with the previous
14
The superfamily of cytochrome P450 enzymes (CYPs) report [16], fatty acid decarboxylation was the dominant
are considered among the most versatile biocatalysts reaction, while the α- and β-hydroxymyristic acid only
in nature [25]. Typically, P450 enzymes employ one or accounted for 0.2% and 6.1% of total products, respect-
more redox partner proteins to transfer two electrons ively (Figure 1B).
from NAD(P)H to the heme iron reactive center for According to previous amino acid sequence analysis,
dioxygen activation, and then insert one atom of O into OleT was assigned to the CYP152 family, with the
2 JE
theirsubstrates[25-27].Therefore,theseoxidativebioca- well-studied P450BSβ [29,34,35] from Bacillus subtilis
talysts are often termed as P450 monooxygenases. How- and P450SPα [28,36] from Sphingomonas paucimobilis as
ever, the CYP152 family members such as P450SPα [28], family members. Essentially, this family of P450 enzymes
P450BSβ [29] and OleTJE [16] have been identified to ex- utilizes H2O2 as the sole electron and oxygen donor to
clusively use H O as the sole electron and oxygen oxidize their substrates, in contrast to most other P450
2 2
donor,andarethusclassifiedintoauniqueP450peroxy- monooxygenases, whose catalytic activity depends on
genase category. dioxygen, NAD(P)H and redox partner protein(s) [37].
Practically, the peroxygenase activity of P450 enzymes Thus, these P450 enzymes are functionally referred to as
is often treated as an advantageous feature because peroxygenases [38]. Indeed, it was reported that P450
H O is much cheaper than NADPH and redox proteins reductase systems such as ferredoxin and ferredoxin
2 2
in many P450 applications such as biocatalysts for reductase did not support the activity of P450BSβ and
in vitro synthetic reactions and enzyme additives in P450SPα[36,39].Thisisconsistentwiththesubstitutionof
laundry detergents [30,31]. However, the large-scale pro- Pro246 and Arg245 (numbering in OleT ) in CYP152 per-
JE
duction of low-cost α-alkene biofuels by OleT P450 oxygenases for the highly conserved threonine and an
JE
decarboxylase cannot rely on the H O -dependent en- acidic residue (for example, Glu or Asp) in most P450
2 2
zymatic system because the use of large amounts of per- monooxygenases (Additional file 3: Figure S3), which are
oxide is cost prohibitive, and high concentration of involved in a specificH-bondnetworkresponsible for de-
H O canquicklydeactivate biocatalysts[32]. liveryofprotonsduringtheP450catalyticcycle[34,40].
2 2
Therefore,theH O -independentactivityofOleT ,ifit However, in the study of another CYP152-member
2 2 JE
exists, is preferred for cost-effective microbial production P450 , from the anaerobic microorganism Clostridium
CLA
of α-alkenes. In the present study, we demonstrated such acetobutylicum [41], it was demonstrated that both
activity for the first time. Through engineering a self- P450CLA and P450BSβ functioned when provided with O2,
sufficient version of OleT by fusing it to the NADPH and flavodoxin/flavodoxin reductase from E. coli
JE
Rhodococcus fusion reductase (RhFRED) domain from orthediflavinreductasedomainofP450 fromBacillus
BM3
Rhodococcussp.NCIMB9784[33],thecatalyticactivityof megaterium, suggesting an alternative monooxygenase-
the resultant fusion P450 enzyme can be solely driven by likemechanismoftheseP450peroxygenases.
Liuetal.BiotechnologyforBiofuels2014,7:28 Page3of12
http://www.biotechnologyforbiofuels.com/content/7/1/28
Figure1DecarboxylationandhydroxylationreactionscatalyzedbyOleT (A)Thedecarboxylationandhydroxylationoffattyacids(C toC )
JE. 12 20
catalyzedbyOleT .(B)ProductdistributionofthedecarboxylationandhydroxylationreactionscatalyzedbypurifiedOleT andOleT -RhFRED,
JE JE JE
respectively.Myristicacidwasusedassubstrate.1-TE,1-tridecene;α-OHMA,α-hydroxymyristicacid;β-OHMA,β-hydroxymyristicacid.
To investigate whether OleT peroxygenase can also myristicacidto1-tridecene(Figure3).Thisclearlyindicates
JE
function as a monooxygenase like P450CLA and P450BSβ, that the decarboxylation activity of OleTJE-RhFRED can be
the RhFRED reductase domain from Rhodococcus sp. solelysupportedbyNADPH.
NCIMB 9784 [33] was fused to the C-terminus of OleT .
JE
In the presence of the electron donor NADPH, the fusion SubstratespecificityofOleT andOleT -RhFRED
JE JE
protein OleT -RhFRED successfully converted lauric acid The substrate preference of OleT and OleT -RhFRED
JE JE JE
(C ) into 1-undecene with the conversion ratio of 51.1 provides useful information to help understand the in vivo
12
±0.3%, approximately half that of OleT plus H O (93.0 behaviorofthesefattyaciddecarboxylasesandtoguidethe
JE 2 2
±4.3%conversion; Figure2).Notably,theproduct distribu- metabolic engineering of fatty acid biosynthesis. Thus, we
tion (decarboxylation versus hydroxylation) of the OleT - determined their substrate specificity (Figure 4) using a
JE
RhFRED catalyzed reaction was similar to that of the number of straight-chain saturated fatty acids with even-
OleT reaction (Figure 1B), indicating the fused reductase numberedchainlengthrangingfromC toC .Inthecase
JE 8 20
domain hasno significant impacton the catalytic mechan- of OleT (Figure 4A), myristic acid (C ) turned out to be
JE 14
ismofP450domain.Ascontrol,OleT cannotdirectlyuse the best substrate for olefin production among the tested
JE
NADPHaselectrondonor.OleT -RhFREDwasnotactive fatty acids with the conversion ratio of 97.0 ± 2.8%. Lauric
JE
by itself. Interestingly, OleT -RhFRED retained its ability acid(C )andpalmiticacid(C )weresub-optimal.OleT
JE 12 16 JE
to use H O as a cofactor, but there was not a significant wasonlyabletoconvertaminorityofstearicacid(C )and
2 2 18
additive effect between NADPH and H O for supporting arachidicacid(C ),butcouldnotdecarboxylatecapricacid
2 2 20
theactivityofOleT -RhFRED(Figure2). (C )orcaprylicacid(C ).
JE 10 8
To prevent the spontaneous generation of H O during In comparison, the substrate preference spectrum of
2 2
theOleT -RhFREDreaction,whichcouldcomplicateinter- OleT -RhFRED slightly shifted toward shorter fatty acids
JE JE
pretation of the results, dithiothreitol (DTT: the reducing (Figure 4B), which suggests that the P450-reductase inter-
agentoftenusedinthestorageandreactionbufferofP450 action might induce small conformational change of the
proteins) was omitted from all buffers used during protein OleT active site. Specifically, OleT -RhFRED showed the
JE JE
purification, storage and reaction because it has previously highest activity against lauric acid (83.8 ±0.1% conversion),
been reported that DTT, dioxygen and the heme iron cen- but was inactive toward arachidic acid. Similar to OleT ,
JE
ter of P450 can react to generate H O [16,42]. To further OleT -RhFRED could not decarboxylate capric acid or
2 2 JE
exclude the possibility that H O could be produced from caprylic acid. It is also worth noting that the system of
2 2
the peroxide shunt pathway during the monooxygenase OleT -RhFRED plus NADPH plus O was less active than
JE 2
catalytic cycle [27], bovine liver catalase that is able to effi- OleT plusH O whenreactingwithmosttestedfattyacids
JE 2 2
ciently eliminate H O was added to the fatty acid decarb- exceptforlauricacid,suggestingOleT mighthaveevolved
2 2 JE
oxylation reactions (in the DTT-free buffer) catalyzed by tobecomeabetterperoxygenasethanamonooxygenase.
OleT and OleT -RhFRED. In the OleT plus H O
JE JE JE 2 2
system, pre-addition of catalase completely abolished the Invivoproductionofα-alkenesinEscherichiacoli
reaction(Figure3).Bycontrast,intheOleT -RhFREDplus Motivated by the efficient in vitro conversion of free fatty
JE
NADPH reaction, pre-added catalase did not significantly acids (C ) to corresponding α-alkenes (C ) by two re-
n n-1
change (indeed, it slightly improved) the conversion of lated but mechanistically distinct decarboxylation systems
Liuetal.BiotechnologyforBiofuels2014,7:28 Page4of12
http://www.biotechnologyforbiofuels.com/content/7/1/28
4×105 U 4×106
3×105 IS 3×106
2×105 LA 2×106 A
1×105 1×106
0 // 0
LA
4×105 IS 4×106
3×105 3×106
2×105 2×106 B
1×105 1×106
0 // 0
4×105 4×106
3×105 U LA IS 3×106
2×105 2×106 C
1×105 1×106
//
y 0 0
sit
n
k Inte 432×××111000555 U LA IS 432×××111000666 D
Pea 1×1005 // 01×106
4×105 4×106
IS
3×105 U LA 3×106
2×105 2×106 E
1×105 1×106
0 // 0
LA
4×105 IS 4×106
3×105 3×106
2×105 2×106 F
1×105 1×106
0 // 0
LA
12×105 IS 4×106
9×105 U 3×106
6×105 2×106 G
3×105 1×106
0 // 0
7.0 8.0 9.0 22.0 23.0 24.0 25.0 26.0 27.0 28.0
Retention Time (min)
Figure2Gaschromatography-massspectroscopyanalysisofdecarboxylationreactions(10min)catalyzedbyOleT orOleT -RhFRED
JE JE
underdifferentreactionsystems.(A)OleT +HO;(B)OleT +NADPH;(C)OleT -RhFRED+NADPH;(D)OleT -RhFRED+HO (E)OleT -
JE 2 2 JE JE JE 2 2; JE
RhFRED+NADPH+HO (F)OleT -RhFREDinabsenceofNADPH;(G)authenticstandardsof1-undecene(U)lauricacid(LA)andheptadecanoic
2 2; JE
acid(IS:internalstandard).
including OleT plus H O and OleT -RhFRED plus As expected, all these engineered strains successfully
JE 2 2 JE
NADPH plus O , we sought to engineer hydrocarbon- accumulated multiple α-alkenes, including 1-tridecene,
2
producing strains of E. coli. To guarantee sufficient trideca-1,6-diene, 1-pentadecene, pentadecene-1,8-diene
substrate supply of free fatty acids for OleT P450 and heptadeca-1,10-diene (Figure 5A and Additional file
JE
decarboxylase, two previously engineered fatty acid- 4: Figure S4), which correspond to the decarboxylation
overproducing strains including XL100 (BL21:ΔfadD) and products of the 14:0, 14:1, 16:0, 16:1 and 18:1 fatty acids,
XL100/(pMSD8 + pMSD15) [43] were used as hosts for respectively. This fatty acid composition is consistent
the decarboxylases. Transformation of these two strains with previous reports [43]. In all cases, the major alkene
withpET28b-oleT orpET28b-oleT -RhFREDresultedin product was heptadeca-1,10-diene with the highest yield
JE JE
YL5 (XL100 with pET28b-oleT ), YL6 (XL100 with of 7.7 mg·l-1 in YL7. Since XL100/(pMSD8 + pMSD15)
JE
pET28b-oleT -RhFRED), YL7 (XL100 with pET28b- is able to produce more free fatty acids than XL100 [43],
JE
oleT , pMSD8 and pMSD15), and YL8 (XL100 with the total alkene titers of YL7 and YL8 were significantly
JE
pET28b-oleT -RhFRED,pMSD8andpMSD15)(Table1). greater than those of YL5 and YL6. Notably, expression
JE
Liuetal.BiotechnologyforBiofuels2014,7:28 Page5of12
http://www.biotechnologyforbiofuels.com/content/7/1/28
Figure3EffectsofcatalaseontheinvitrodecarboxylationactivityofthreeOleT (1μM)reactionsystems.SystemscomprisedOleT
JE JE
plusH O ,OleT -RhFREDplusNADPH,andOleT plusflavodoxinandflavodoxinreductaseplusNADPH.Thepercentageconversionof
2 2 JE JE
myristicacidisshownbesideeachbar.
of P450 decarboxylases did not significantly affect the native redox system of E. coli to drive catalysis, such as
cellgrowth ofallstrains(Additional file5:FigureS5). P450CLAandP450BSβ[41].
According to a previous report [44], H O levels in To test this hypothesis, we expressed and purified the
2 2
growing E. coli are less than 20 nM, and 2 μM H O E. coli Fld and FdR. In the in vitro reaction containing
2 2
could cause substantial growth inhibition. Therefore, the OleT , Fld, FdR and NADPH, myristic acid was sub-
JE
physiological concentration of H O is far below the stantially converted into 1-tridecene (Figure 3). There-
2 2
optimal H O concentration (200 to 500 μM) [16] for fore, Fld and FdR probably better serve OleT than the
2 2 JE
supporting OleT catalysis. Thus, we had predicted fused RhFRED in vivo, explaining why YL7 was a better
JE
that OleT -RhFRED using endogenous NADPH, whose alkene producer than YL8. However, the supportive role
JE
physiological concentration is around 120 μM [45], ofH O cannotbeentirelyneglected.
2 2
might catalyze the decarboxylation more efficiently than Inspired by these results, we attempted to enhance the
OleT using H O as a cofactor. Unexpectedly, the level of dissolving dioxygen for improving OleT cataly-
JE 2 2 JE
highest titer (11.8 mg·l-1) and productivity (4.5 mg·g sis during the cultivation of the best alkene producer,
dcw-1) of total alkenes was achieved by YL7 with OleT YL7, byapplyingahigherrotationrateof250rpm.More-
JE
instead of YL8 with OleT -RhFRED. This contradiction over,thedefinedmineralmedium[14]containing3%glu-
JE
has led to a hypothesis that OleT is able to employ the cose was used to standardize the culture condition for
JE
Figure4Substratepreferencespectrumof(A)OleT and(B)OleT -RhFRED.Thesubstratepreferencewasdeterminedbycalculatingthe
JE JE
percentageconversionofeachfattyacidsubstrateintocorrespondingα-alkeneproduct.Intheseassays,0.2μMenzymeswereused.
Liuetal.BiotechnologyforBiofuels2014,7:28 Page6of12
http://www.biotechnologyforbiofuels.com/content/7/1/28
Table1Strainsandplasmidsusedinthisstudy expression was evaluated. Octadec-11-enoic acid was
Strainor Relevantcharacteristics Sourceor the most abundant fatty acid with a titer of 96.1 mg·l-1
plasmid reference (Additional file 6: Figure S6). This well explains why its
E.colistrains decarboxylated product, heptadeca-1,10-diene, had the
BL21(DE3) F-ompTgaldcmlonhsdSB Novagen highestyieldamongproducedalkenes(Figure5).Withre-
(rB- mB-)λ(DE3) spect to fatty acid conversion, octadec-11-enoic acid had
DH5α F-ξ80lacZΔM15Δ(ΔlacZYA-argF) Invitrogen the highest conversion ratio (85.9%) followed by hexadec-
U169recA1endA1hsdR17(rk-,mk+) 9-enoic acid (85.7%), myristic acid (75.7%), palmitic acid
phoAsupE44thi-1gyrA96relA1λ-
(74.0%)andlauricacid(24.4%).Thelowconversionoflau-
XL100 BL21:ΔfadDa [43] ric acid seems to be inconsistent with the in vitro result
YL1 BL21(DE3)withpET28b-oleTJE Thisstudy (Figure 4A), which is likely due to the low intracellular
YL2 BL21(DE3)withpET28b-oleTJE-RhFRED Thisstudy concentrationoftheC12fattyacid.
YL3 BL21(DE3)withpACYCDuet-fdR Thisstudy
Discussion
YL4 BL21(DE3)withpCDEDuet-fld Thisstudy
The cytochrome P450 enzymes are a superfamily of
YL5 XL100withpET28b-oleT Thisstudy
JE b-type heme proteins, capable of catalyzing more than
YL6 XL100withpET28b-oleTJE-RhFRED Thisstudy 20 different types of reactions [25,46,47]. The H O -
2 2
YL7 XL100withpET28b-oleTJE,pMSD8b, Thisstudy assisted decarboxylation of long-chain fatty acids cata-
pMSD15c
lyzed by P450 OleT represents a novel activity of this
JE
YL8 XL100withpET28b-oleTJE-RhFRED, Thisstudy highly versatile superfamily. This activity may be mech-
pMSD8,pMSD15
anistically similar to that of P450 (CYP53B) from the
Rm
Jeotgalicoccus Wildtype ATCC yeast Rhodotorula minuta [48], which decarboxylates
sp.ATCC8456
isovaleratetoform isobutene.
Plasmids
According to protein sequence alignment (Additional
pET28b Kmr,T7promoter,pBR322origin Novagen file 3: Figure S3), OleT belongs to the CYP152 peroxy-
JE
pET28b-oleTJE Kmr,pET28bderivativecontaining Thisstudy genase family together with the well-studied P450BSβ
oleTJEgene and P450SPα [28,29], and other members. Functionally,
pET28b-oleTJE- Kmr,pET28bderivativecontaining Thisstudy these P450 peroxygenases were previously thought to
RhFRED oleT andRhFREDgene
JE have no ability of utilizing dioxygen to drive catalysis as
pACYCDuet-1 Cmr,T7promoter,P15Aorigin Novagen typical P450 monooxygenases do [39]. However, we have
pACYCDuet-fdR Cmr,pACYCDuet-1derivative Thisstudy unambiguously demonstrated that OleT can perform
JE
containingfdRgene H O -independent catalysis when partnering with the
2 2
pCDFDuet-1 Strr,T7promoter,CloDF13origin Novagen fused RhFRED reductase or the E. coli Fld/FdR system
pCDEDuet-fld Strr,pCDFDuet-1derivative Thisstudy to transfer electrons from NADPH to the heme iron
containingfldgene reactive center. This strongly suggests that OleT
JE
pMSD8 Ampr,P :accB,C,D,A, [43] can undergo the monooxygenase catalytic cycle to
T7
pSC101origin generatethehighly reactiveferryl-oxocation radical spe-
pMSD15 Cmr,PBAD:tesA,p15aorigin [43] cies (Compound I) for catalysis. In the well-accepted
aThefadDgeneencodestheacyl-CoAsynthetase,whichisthefirstenzyme mechanism for dioxygen activation in P450 monooxy-
involvedinfattyaciddegradation.bpMSD8:theexpressionvectorforE.coli genases (Figure 6), two protons (and two electrons from
acyl-CoAcarboxylase.cpMSD15:theplasmidusedforoverexpressionoftheE.
colithioesterasegenetesA.Amp,ampicillin;Cm,chloramphenicol;Km, NADPH) are required for generation of Compound I
kanamycin;Str,streptomycin. [27]. However,theconservedthreonineandanacidicresi-
dueinvolvedinprotondeliveryinnormalP450monooxy-
the future metabolic engineering, and the culture time genases [34,40] are replaced by Pro and Arg, respectively,
wasextendedto40h.Significantly,thetotalalkenetiterof which are absolutely conserved in OleT and all other
JE
YL7 under these conditions was 97.6 mg·l-1 (Figure 5B), known CYP152 family members (Additional file 3: Figure
which is almost seven-fold higher than that in lysogeny S3). This strongly suggests an unknown proton transfer
broth (LB) medium at 220 rpm for 20 h (Figure 5A). The pathway. To elucidate this hypothetical pathway, we are
productivity of total alkenes was also improved from 4.5 currentlyseekingtosolvethecrystalstructureofOleT .
JE
mg·g dcw-1 (in LB over 20 h) to 24.9 mg·g dcw-1 (in the Evolutionarily, because the early Earth’s environment
defined mineral medium over 40 h, data not shown). probably had more H O and peroxygenated organic
2 2
Again,heptadeca-1,10-dienewasthemajoralkeneproduct chemicals than O , P450 peroxygenases are presumed to
2
(41.4mg·l-1).As control,theproductionoffreefattyacids have emerged ahead of P450 monooxygenases [38].
by the strain XL100/(pMSD8 + pMSD15) without OleT Thus, the ability of most P450 monooxygenases to use
JE
Liuetal.BiotechnologyforBiofuels2014,7:28 Page7of12
http://www.biotechnologyforbiofuels.com/content/7/1/28
Figure5Productionofα-alkenesbyheterologousexpressionofOleT orOleT -RhFREDinfattyacid-overproducingEscherichiacoli
JE JE
strains.(A)Productionofα-alkenesbyOleT andOleT -RhFREDintwofattyacid-overproducingE.colistrains.Thelinechartdenotesthealkene
JE JE
productivity.(B)Theα-alkeneproductionandfattyacidconversionprofilesofstrainYL7inthedefinedmineralmedium.Thehistogramsdemonstrate
thetiterofeachdetectedα-alkenetiters.Thelinechartdemonstratestheconversionratioofcorrespondingfattyacids.FFA,freefattyacid.
H O as a surrogate for the O /2e-/2H+ system (perox- [30,31]. However, the peroxygenase nature is not a good
2 2 2
ide shunt pathway, Figure 6) could be understood as a feature if attempting to construct a biofuel-producing
remnant function inherited from their peroxygenase an- microorganism by taking advantage of the fatty acid
cestors, whereas the OleT P450 peroxygenase bearing decarboxylation activity of OleT . Essentially, the intra-
JE JE
the monooxygenase property might represent a transi- cellular level of H O cannot be increased to the con-
2 2
tion species during the evolving process from peroxy- centration (101 to 102 μM range) required for efficiently
genasetomonooxygenase. supporting OleT because at this level, H O is toxic or
JE 2 2
Under certain circumstances, such as in vitro synthetic fatal to all organisms including E. coli. Thus, that the ac-
reactions using purified P450s and application of P450 tivityofP450OleT canbesupportedbyO /redoxpart-
JE 2
enzymes in laundry detergents, the peroxygenase activity ner(s)/NADPH besides H O is a significant discovery
2 2
supported by H O is advantageous because expensive because the former three factors are more viable targets
2 2
redox partner proteins and NAD(P)H are not needed for metabolic engineering [49-51]. Being aware of this,
Figure6ProposedtwoalternativecatalyticmechanismsofOleT with(monooxygenase-like)orwithout(peroxygenase-like)redoxsystems.
JE
TheArg245isproposedtoberequiredforsubstrateanchoringviathestrongelectrostaticinteractionswiththecarboxylgroupoffattyacidsubstrate.
Thesupplyoftwoprotonsintheputativemonooxygenasecatalyticcycleremainsunclear.Thedashedarrowindicatestheperoxideshuntpathway.
Liuetal.BiotechnologyforBiofuels2014,7:28 Page8of12
http://www.biotechnologyforbiofuels.com/content/7/1/28
future metabolic engineering work aiming to improve proximately 2 to 5 mg·l-1) [20]; and the ATP required
the in vivo productivity of α-alkenes by OleT probably decarboxylationof3-hydroxy-3-methylbutyrate catalyzed
JE
should be directed to improvement of the intracellular by the R74H mutant of mevalonate diphosphate decarb-
level of dioxygen, redox partner protein(s) and/or oxylasefromSaccharomycescerevisiae(productivity:5,888
NADPH. For example, among numerous metabolic pmol·h-1·gcells-1)[57].Itisonlylowerthanthe300mg·l-1
strategies for in vivo up-regulation of NADPH [49], the of total alkane titer produced when the cyanobacterial
NADPH regenerating system could be used for main- pathway consisting of the acyl-ACP reductase Orf1594
taining the high intracellular level of this reducing cofac- from Synechococcus elongates PCC7942 and the aldehyde
tor to better support the activity of OleT and the decarbonylase from Nostoc punctiforme PCC73102 was
JE
overproduction of fatty acids [52,53], both of which re- heterologously expressed in E. coli [14]. However, it is
quirea sufficient supplyofNADPH. expected that more metabolic engineering efforts and
InthemutantE.colistrainswithup-regulatedfattyacid optimizationoffermentationwillfurtherincreasethetotal
biosynthesis (XL100 and XL100/(pMSD8 + pMSD15)), alkene titers of the OleT /E. coli system, which is cur-
JE
both OleT and OleT -RhFRED were functional, there- rentlyongoinginourlaboratory.
JE JE
fore significantly converting fatty acids into α-alkenes.
Consistent with the previous report [16], the common Conclusions
major product of four engineered alkene producers (YL5- The H O independence of OleT described in this
2 2 JE
8) turned out to be heptadeca-1,10-diene, but the second work not only raises a number of fundamental questions
mostabundantalkenevaried(Figure5A).Thisislikelybe- regarding its monooxygenase-like mechanism, but also
cause octadec-11-enoic acid (the precursor of heptadeca- could direct future metabolic engineering work toward
1,10-diene) was the predominant component of the fatty improvement of O /redox partner(s)/NADPH for opti-
2
acid pool of the tested strains (Additional file 6: Figure mal activity of OleT in vivo. Considering its high con-
JE
S6). However, there might be an additional reason that version rate in vitro and H O -independent functionality
2 2
octadec-11-enoic acid is a preferred substrate in vivo. in vivo, it is of great potential for engineering a hyper-
Interestingly, in previous studies [43,54,55], palmitic acid producer ofα-alkenesonthe basis ofOleT .
JE
rather than octadec-11-enoic acid was the major compo-
nent in the fatty acid pool of E. coli. This inconsistency Methods
couldbeduetodifferentcultureconditions. Materials
After 20 h cultures of YL7 in LB medium, significant Fatty acid substrates, terminal alkene authentic standards,
amounts of fatty acids remained unreacted with OleT and derivatizing reagents were purchased from TCI
JE
(Additional file 4:Figure S4). Bycontrast,themajorityof (Shanghai, China). Antibiotics were obtained from Solar-
fatty acids in YL7 were consumed by OleT over the Bio (Beijing, China). Other chemicals were from Sigma
JE
longer and more oxygenated cultivation in the defined Aldrich (St. Louis, MO, USA) or Ameresco (Solon, OH,
mineral medium (Figure 5B). This suggests that, at the USA). Oligonucleotides were synthesized by Sangon Bio-
current stage, the yield of fatty acids might be the major tech (Shanghai, China), and their sequences are shown in
limiting factor for further improvement of α-alkene Additionalfile7:TableS1.ThePfuDNApolymerasesand
titers. Thus, the overproduction of fatty acids needs to all restriction endonucleases were obtained from Fermen-
be significantly optimized prior to other engineering ef- tas(Vilnius,Lithuania)orTakara(Dalian,China).Thekits
forts on up-regulating the level of OleT , redox part- used for molecular cloning were from OMEGA Bio-Tek
JE
ners,O andNADPH. (Jinan, China) or Promega (Madison, WI, USA). Protein
2
In this report, the best bio-hydrocarbon-producing purification used Qiagen Ni-NTA resin (Valencia, CA,
strain YL7 accumulated 97.6 mg·l-1 of total alkenes. This USA),Millipore AmiconUltracentrifugalfliters(Billerica,
yield is comparable to or better than a majority of engi- MA,USA)andPD-10desaltingcolumnsfromGEHealth-
neered alkane and alkene biosynthetic pathways with care (Piscataway, NJ, USA). Bovine liver catalase was pur-
reported yields. These include the artificial alkane bio- chasedfromSigmaAldrich.
synthetic pathway in E. coli consisting of the carboxylic
acid reductase from Mycobacterium marinum that cata- Molecularcloning
lyzes formation of fatty aldehydes directly from fatty Strains and plasmids constructed and used in this study
acids, and the aldehyde decarbonylase from Synechocys- are listed in Table 1. The gene of oleT was amplified
JE
tis sp. PCC 6803 to produce alkanes (yield: approxi- from the genomic DNA of Jeotgalicoccus sp. ATCC
mately 2 mg·l-1) [56]; the hybrid system using the fatty 8456 using the primer pair of OleT-NdeI/OleT-HindIII
acid reductase complex (LuxC, LuxE and LuxD) to (Table 1). The gel-cleaned PCR fragment was double
provide fatty aldehyde as the substrate for downstream digested by NdeI and HindIII and subsequently ligated
aldehyde decarbonylase to generate alkanes (yield: ap- into the NdeI/HindIII pre-treated pET28b to afford
Liuetal.BiotechnologyforBiofuels2014,7:28 Page9of12
http://www.biotechnologyforbiofuels.com/content/7/1/28
pET28b-oleT (Additional file 8: Figure S7A). For imidazole) until no proteins were detectable in flow-
JE
pET28b-oleT -RhFRED, the genes of oleT and through. The bound target proteins were eluted with
JE JE
RhFRED were first fused by overlap extension PCR [58]. elutionbuffer(pH8.0,50mMNaH PO ,300mMNaCl,
2 4
Briefly,thegene encodingRhFRED reductase wasampli- 10% glycerol and 250 mM imidazole). The eluent was
fied from the previously constructed pET28b-pikC- concentrated with an Amicon Ultra centrifugal filter,
RhFRED [59,60] with a pair of primers including and buffer exchanged on a PD-10 desalting column. Fi-
RhFRED-F and RhFRED-R. The oleT gene was ampli- nally, the desalted purified proteins (Additional file 1:
JE
fied from pET28b-oleT , using the primers OleT-F and Figure S1) in storage buffer (pH 7.4, 50 mM NaH PO ,
JE 2 4
OleT-RhFRED-OE. Then the two PCR fragments with 10% glycerol) were flash-frozen by liquid nitrogen and
overlapsequencewere mixed,annealed, extendedandfi- stored at −80°Cfor later use.
nally amplified with the OleT-F/RhFRED-R primer pair,
giving rise to the fused gene of oleT -RhFRED. This fu- Determinationofproteinconcentration
JE
sion product was digested by NdeI/HindIII and inserted Following the method described by Omura and Sato [62],
into the NdeI/HindIII-digested pET28b to generate the CO-bound reduced difference spectrum (Additional
pET28b-oleT -RhFRED(Additionalfile 8:FigureS7B). file 2: Figure S2) of each P450 enzyme was recorded on a
JE
Using the genomic DNA of E. coli DH5α as template, UV-visible spectrophotometer DU 800 (Beckman Coulter,
thetwo genesencodingFldandFdRwereamplifiedwith Fullerton, CA, USA). The functional P450 concentration
the primer pairs of Fld-BamHI-F/Fld-SalI-R and FdR- was calculated using the extinction coefficient (ϵ450–490)
BamHI-F/FdR-SalI-R, respectively. Next, the fld and fdR of 91,000 M-1·cm-1. The concentration of Fld and FdR
genes were sub-cloned into pACYCDuet-1 and from E. coli was determined using ϵ = 4,570 M-1·cm-1
579
pCDFDuet-1 respectively, resulting in pACYCDuet-fld [63]andϵ =7,100M-1·cm-1[64],respectively.
456
and pCDFDuet-fdR. All sub-cloned sequences were
confirmed by DNA sequencing at Sangon Biotech, Invitroenzymaticassayswithpurifiedproteins
Shanghai. The fatty acid decarboxylation assays containing 0.2 to
1.0 μM OleT or OleT -RhFRED, 200 μM fatty acid
JE JE
Proteinoverexpressionandpurification substrate (from C to C ), 500 μM H O (for OleT ) or
8 20 2 2 JE
The E. coli BL21(DE3) cells carrying the recombinant 500 μM NADPH (for OleT -RhFRED) in 200 μl of stor-
JE
expression vector were grown at 37°C for 16 to 20 h age buffer were carried out at 28°C for 2 h. In the assay
in LB medium containing certain selective antibiotics to test whether E. coli Fld and FdR are able to support
(50 μg·ml-1 kanamycin, 34 μg·ml-1 chloramphenicol, or the activity of OleT , 5 μM Fld and 5 μM FdR were
JE
50 μg·ml-1 streptomycin), which were used to inoculate mixed with 200 μM myristic acid, 500 μM NADPH and
(1:100 ratio) Terrific Broth medium containing corre- 1 μM OleT in 200 μl of storage buffer. To remove
JE
sponding antibiotics, thiamine (1 mM), 10% glycerol and spontaneously generated H O , bovine liver catalase was
2 2
a rare salt solution [61]. Cells were grown at 37°C for added tothefinal concentration of20U·ml-1.
3 to 4 h until the optical density at 600 nm (OD ) All above described reactions were quenched by addi-
600
reached 0.6 to 0.8, at which isopropyl-β- -thiogalactopy- tionof20μlof10MHCl.Heptadecanoicacidwasadded
D
ranoside (IPTG, 0.2 mM as final concentration) and asinternalstandardandthemixturewasextractedby200
δ-aminolevulinic acid (0.5 mM, only for P450 expression) μl ethyl acetate. Samples were then analyzed by gas
wereadded,followedby18hofcultivationat18°C. chromatography-massspectroscopy(GC-MS;seebelow).
Protein purification was carried out as described else-
where [61] with slight modifications. Specifically, the Invivoproductionofα-alkenes
cell pellets harvested by centrifugation (5,000 × g, 4°C, The E. coli fadD deletion mutant strain XL100 [43] that
15 min) were stored at −80°C and melted at ambient overproducesfatty acids wasselected asthestartinghost
temperature immediately before use. Then the cell for construction of alkene-producing strains. Plasmids
pellets were resuspended in 40 ml of pre-chilled lysis pMSD8 and pMSD15 were gifts from Dr. John Cronan
buffer (pH 8.0, 50 mM NaH PO , 300 mM NaCl, 10% and were used in the strain XL100 to construct fatty
2 4
glycerol and 10 mM imidazole) through vortexing. acid-overproducing strains. Bacterial cells transformed
Following a sonication step, the cell lysate was centri- with certain plasmid(s) (Table 1) were either grown in
fuged at 12,000 × g for 30 min to remove the insoluble 50 ml of LB medium or in 50 ml of defined mineral
fraction. To the supernatant, 1 ml of Ni-NTA resin medium [14] containing 3% glucose as the carbon
was added and gently mixed at 4°C for 1 h. The slurry source, supplemented with appropriate selective antibi-
was loaded onto an empty column, and washed with otics,thiamine(1mM)andararesalt solution. Allculti-
approximately 100 ml of wash buffer (pH 8.0, 50 mM vations were performed at 37°C and induced at an
NaH PO , 300 mM NaCl, 10% glycerol and 20 mM OD of 0.9 to 1.0 with 0.4% arabinose (for pMSD15)
2 4 600
Liuetal.BiotechnologyforBiofuels2014,7:28 Page10of12
http://www.biotechnologyforbiofuels.com/content/7/1/28
followed by 0.2 mM IPTG (for pMSD8, and pET28b- Quantification was achieved by comparison of integrated
oleT or pET28b-oleT -RhFRED) after 0.5 h. Next, cells peak areas with calibration curves of authentic standards.
JE JE
were grown at 28°C (the optimal temperature for OleT The conversion percentages of free fatty acids to corre-
JE
expression) for an additional 20 h at 220 rpm (in LB sponding α-alkenes were estimated using the equation:
medium) or 40 h at 250 rpm (in the defined mineral [totalalkenes]/([totalalkenes]+[totalfreefattyacids]).
medium). For analysis of hydrocarbon production, 20 ml
of culture with 10 μl heptadecanoic acid added as
Additional files
internal standard was sonicated for 10 min and then
thoroughly mixed with an equal volume of chloroform-
Additionalfile1:FigureS1.SDS-PAGEanalysisofpurified(A)OleT
methanol (2:1, vol/vol). The aqueous-organic mixture JE
and(B)OleT -RhFRED.M,proteinmarker.
JE
was centrifuged (8,000 × g for 15 min) for phase separ-
Additionalfile2:FigureS2.CO-boundreducedspectraofpurified(A)
ation. The organic phase was transferred into a clean OleT and(B)OleT -RhFRED.
JE JE
tube, evaporated under a nitrogen flow, and re-dissolved Additionalfile3:FigureS3.ProteinsequencealignmentofCYP152
in 500 μl of n-hexane as the testing sample. Prior to familymembersandthreeselectedP450monooxygenases.Protein
GC-MS analysis, 5 μl eicosane was added as calibration sequenceswereobtainedfromNCBIproteindatabases.CYP152A1
(P450BSβfromBacillussubtilis);CYP152A2(P450CLAfromClostridium
standard. All experiments were repeated two to four acetobutylicum);CYP152B1(P450SPαfromSphingomonaspaucimobilis);
times. CYP152B2(fromAzotobactervinelandii);CYP152C1(fromRhodobacter
sphaeroides);CYP152C2(fromRhodobactersphaeroides);CYP152D1(from
Streptomycesscabies);CYP152E1(fromCyanothecesp.CCY0110);
Analyticalmethods CYP107L1(P450 fromStreptomycesvenezuelae);CYP102A1(P450
PikC BM3
The GC-MS analytical method for hydrocarbon and fromBacillusmegaterium);CYP101A1(P450camfromPseudomonasputida).
Conservedaminoacidresiduesareshaded.TheArgandProabsolutely
fatty acid samples was adapted from Guan et al. [65].
conservedinCYP152familyaremarkedbyasterisks.
The analyses were performed on an Agilent 7890A gas
Additionalfile4:FigureS4.GC-MSanalysisoftheorganicextractof
chromatograph equipped with a capillary column HP- theYL7cultureinLBbroth.Eicosaneandheptadecanoicacidareserved
INNOWAX (Agilent Technologies, Santa Clara, CA, ascalibrationstandard(CS)andinternalstandard(IS),respectively.
USA; cross-linked polyethylene glycerol, i.d. 0.25 μm Additionalfile5:FigureS5.ThedrycellweightoftheYL5-8cultures
usingLBbroth.
film thickness, 30 m by 0.25 mm) coupled to an Agilent
Additionalfile6:FigureS6.Productionprofileoffreefattyacidsby
5975C MSD single quadrupole mass spectrometer oper-
thestrainXL100/(pMSD8+pMSD15).
atedunderelectronionizationmodeat70eVinthescan
Additionalfile7:TableS1.Primersusedinthisstudy.
range of 50 to 500 m/z. The helium flow rate was set to
Additionalfile8:FigureS7.Plasmidmapsfor(A)pET28b-oleT and
1 ml·min-1. The oven temperature was controlled ini- (B)pET28b-oleT -RhFREDexpressionvectors. JE
JE
tially at 40°C for 4 min, then increased at the rate of Additionalfile9:FigureS8.GC-MSanalysisoftheα-andβ-hydroxy
10°C per min to 250°C, and held for 15 min. The inject- myristicacidauthenticstandardsonHP-INNOWAXcapillarycolumn.
(A)Withoutderivatization,α-hydroxymyristicacidwasunseenduetoits
ing temperature was set to 280°C with the injection vol-
highboilingpoint.Importantly,nothermallydegradedterminalolefin
ume of 1 μl under splitless injection conditions. Under (1-tridecene)wasobservedinthedashedbox.(B)Withoutderivatization,
these conditions, the previously reported thermal degra- β-hydroxymyristicacidwasunseenduetoitshighboilingpoint.Again,
dations of α- and β-hydroxy fatty acids to form alkenes nothermallydegradedterminalolefin(1-tridecene)wasobservedin
thedashedbox.(C)Thedecarboxylationreactionofmyristicacid(2h)
(the minor products of OleTJE catalyzed reaction) in catalyzedbyOleTJE.Thepeakshowninthedashedboxcorresponds
the GC inlet [16] were not observed (Additional file 9: to1-tridecene.
Figure S8), making the silylating protection of the
hydroxyl group unnecessary if only caring about the
α-alkene products. To detect α- and β-hydroxy fatty Abbreviations
ACP:acylcarrierprotein;CoA:coenzymeA;CYP:cytochromeP450enzyme;
acids, samples were derivatized with an equal volume of DTT:dithiothreitol;FdR:flavodoxinreductase;Fld:flavodoxin;GC-MS:gas
N, O-bis(trimethylsilyl)trifluoroacetamide with 1% tri- chromatography-massspectrometry;IPTG:Isopropyl-β-D-thiogalactopyranoside;
LB:lysogenybroth;OD:opticaldensity;PCR:polymerasechainreaction;
methylchlorosilane at 70°C for 15 min. GC-MS analysis
RhFRED:Rhodococcusfusionreductase.
followed the previous protocol developed by Rude et al.
[16] except for using the Agilent J&W DB-5 MS column
(i.d. 0.25 μm film thickness, 50 m by 0.25 mm). During Competinginterests
Theauthorsdeclarethattheyhavenocompetinginterests.
GC-MS analysis, peak identity was determined by com-
parison of retention time and fragmentation pattern
Authors’contributions
with authentic standard compounds where available
SLconceivedofthestudy.YL,CW,XLandSLdesignedtheexperiments.YL,
and to the National Institute of Standards and Technol- JYandWZperformedtheexperimentsincludingplasmidconstruction,
ogy, USA mass spectral database. The location of double proteinoverexpression,purification,characterizationandenzymaticassays.
bond(s) in α-olefins was deduced by derivatization YL,CWandWGcarriedoutGC-MSanalysis.YL,CWandSLdraftedthe
manuscript.JY,WZ,WGandXLhelpedtorevisethemanuscript.Allauthors
with dimethyl disulfide as described previously [66]. readandapprovedthefinalmanuscript.