A simple procedure for selective hydroxylation of L -proline and l -pipecolic acid with recombinantly expressed proline hydroxylases

Article abstract of DOI:10.1002/adsc.201000863

Due to their diverse regio- and stereoselectivities, proline hydroxylases provide a straightforward access to hydroxprolines and other hydroxylated cylic amino acids, valuable chiral building blocks for chemical synthesis, which are often not available at reasonable expense by classical chemical synthesis. As yet, the application of proline hydroxylases is limited to a sophisticated industrial process for the production of two hydroxyproline isomers. This is mainly due to difficulties in their heterologues expression, their limited in vitro stability and complex product purification procedures. Here we describe a facile method for the production of cis-3-, cis-4- and trans-4-proline hydroxylase, and their application for the regio- and stereoselective hydroxylation of L-proline and its six-membered ring homologue l-pipecolic acid. Since in vitro catalysis with these enzymes is not very efficient and conversions are restricted to the milligram scale, an in vivo procedure was established, which allowed a quantitative conversion of 6 mM l-proline in shake flask cultures. After facile product purification via ion exchange chromatography, hydroxyprolines were isolated in yields of 35-61% (175-305 mg per flask). L-Pipecolic acid was converted with the isolated enzymes to prove the selectivities of the reactions. In transformations with optimized iron(II) concentration, conversions of 17-68% to hydroxylated products were achieved. The regio- and stereochemistry of the products was determined by NMR techniques. To demonstrate the applicability of the preparative in vivo approach for non-physiological substrates, L-pipecolic acid was converted with an E. coli strain producing trans-4-proline hydroxylase to trans-5-hydroxy-L-pipecolic acid in 61% yield. Thus, a synthetically valuable group of biocatalysts was made readily accessible for application in the laboratory without a need for special equipment or considerable development effort.

Full text of DOI:10.1002/adsc.201000863

FULL PAPERS
DOI: 10.1002/adsc.201000863
A Simple Procedure for Selective Hydroxylation of l-Proline
and l-Pipecolic Acid with Recombinantly Expressed Proline
Hydroxylases
a
a,
Christian Klein and Wolfgang Hꢀttel *
a
Institute of Pharmaceutical Sciences, Department of Pharmaceutical and Medicinal Chemistry,
Albert-Ludwigs-Universitꢁt Freiburg, Albertstr. 25, 79104 Freiburg, Germany
Fax : (+49)-(0)761-203-6351; e-mail: wolfgang.huettel@pharmazie.uni-freiburg.de
Received: November 17, 2010; Revised: March 3, 2011; Published online: May 17, 2011
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/adsc.201000863.
Abstract: Due to their diverse regio- and stereoselec- product purification via ion exchange chromatogra-
tivities, proline hydroxylases provide a straightfor- phy, hydroxyprolines were isolated in yields of 35–
ward access to hydroxprolines and other hydroxylat- 61% (175–305 mg per flask). l-Pipecolic acid was
ed cylic amino acids, valuable chiral building blocks converted with the isolated enzymes to prove the se-
for chemical synthesis, which are often not available lectivities of the reactions. In transformations with
at reasonable expense by classical chemical synthesis. optimized iron(II) concentration, conversions of 17–
As yet, the application of proline hydroxylases is lim- 68% to hydroxylated products were achieved. The
ited to a sophisticated industrial process for the pro- regio- and stereochemistry of the products was deter-
duction of two hydroxyproline isomers. This is mined by NMR techniques. To demonstrate the ap-
mainly due to difficulties in their heterologues ex- plicability of the preparative in vivo approach for
pression, their limited in vitro stability and complex non-physiological substrates, l-pipecolic acid was
product purification procedures. Here we describe a converted with an E. coli strain producing trans-4-
facile method for the production of cis-3-, cis-4- and proline hydroxylase to trans-5-hydroxy-l-pipecolic
trans-4-proline hydroxylase, and their application for acid in 61% yield. Thus, a synthetically valuable
the regio- and stereoselective hydroxylation of l-pro- group of biocatalysts was made readily accessible for
line and its six-membered ring homologue l-pipecol- application in the laboratory without a need for spe-
ic acid. Since in vitro catalysis with these enzymes is cial equipment or considerable development effort.
not very efficient and conversions are restricted to
the milligram scale, an in vivo procedure was estab- Keywords: asymmetric catalysis; enzyme catalysis;
lished, which allowed a quantitative conversion of hydroxyprolines; a-ketoglutarate-dependent iron(II)
6
mM l-proline in shake flask cultures. After facile oxygenases; regioselectivity; stereoselectivity
[
2]
Introduction
and antispastic agents. If not producible through col-
lagen hydrolysis, functionalized proline derivatives
Amino acids and their derivatives are valuable build- are not readily available, either as natural products or
ing blocks for organic synthesis. Especially proline by chemical synthesis. A promising approach for se-
with its rigid conformation has strong chirality induc- lective functionalization of amino acids is provided by
ing properties and is therefore widely used for stereo- the enzyme-catalyzed stereoselective hydroxylation.
selective synthesis, both in stoichiometric amounts Especially a-ketoglutarate (a-KG) dependent Fe(II)-
[
1]
and as a chiral catalyst. Thus, proline derivatives, es- hydroxylases catalyze CꢀH activation reactions with
[
3]
pecially those with substitutions at the carbon back- high regio- and stereoselectivity, and in contrast to
bone, represent an interesting substance class for ex- the NAD(P)H-dependent P450 monooxygenases,
tension of the chiral pool. For instance, hydroxypro- their co-substrate a-KG is inexpensive so that a re-
lines are used as building blocks for the synthesis of generation system is not required.
important pharmaceutical drugs such as carbapenem
In the case of l-proline hydroxylation, different a-
antibiotics, angiotensin-converting enzyme inhibitors KG-dependent hydroxylases converting the free
Adv. Synth. Catal. 2011, 353, 1375 – 1383
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1375

Christian Klein and Wolfgang Hꢀttel
FULL PAPERS
Scheme 1. Proline hydroxylases cloned and expressed in E. coli and activities detected in the fungus Glarea lozoyensis.
[
10]
amino acid have been identified in nature, generating the basement membrane. In an alternative synthetic
all four possible regio- and stereoisomers of hydroxy- approach, recombinant E. coli strains producing trans-
proline (Scheme 1). To date, three types of bacterial P4H from Dactylosporangium sp. strain RH1 and cis-
enzymes, cis-3-, cis-4- and trans-4-proline hydroxylase P3H from Streptomyces sp. strain TH1 are applied for
(
cis-P3H, cis-P4H and trans-P4H) have been cloned the enzymatic production of trans-4- and cis-3-hydr-
[
4]
[11]
and overexpressed in E. coli. cis-P3H (type II) from
ACHTUNGTRENNUoGN xyproline, respectively. The process was designed
Streptomyces sp. strain TH1 was also characterized as for the industrial scale using special expression vec-
[
5]
the crystal structure of an iron complex. An a-KG- tors, optimized cultivation conditions and a complex
dependent trans-P3H activity besides a trans-P4H ac- product purification procedure.
tivity was identified in protein crude extracts of the
pneumocandin-producing fungus Glarea lozoyensis.
It was also found that proline hydroxylases are not
strictly substrate specific, but accept a set of proline
[
6]
In contrast to the prolyl hydroxylases involved in congeners as long as the secondary amino acid moiety
[
12]
collagen biosynthesis, microbial proline hydroxylases is untouched. We were intrigued by the potential of
exclusively hydroxylate the free amino acid. The pro- these enzymes for functionalizing non-natural amino
line hydroxylases characterized to date have been acids with diverse regio- and stereoselectivities and a
found in bacteria, where they are associated with the comprehensive system for the production of these
biosynthesis of peptide antibiotics. cis-P3H and trans- compounds on the laboratory scale was anticipated.
P4H were isolated from Actinomycetales including the However, several challenges had to be met: (i) The
producers of telomycin (S. canus ATCC 12647) and enzymes are virtually insoluble and not active when
[
7]
etamycin (S. griseoviridus P8648). cis-P4Hs have expressed in E. coli common expression vectors. (ii)
been identified in two strains of the order Rhizo- They tend to denature rapidly when used for in vitro
biales, Mesorhizobium loti and Sinorhizobium melilo- transformations. (iii) In the case of in vivo conver-
[4]
ti.
sions, an elaborate work-up process to isolate the
The amino acids trans-4- and trans-3-hydroxypro- products from the medium is required.
line are available cost-efficiently by acidic hydrolysis Here we present a generally applicable method in-
of collagen. However, this process includes a complex cluding cloning, protein expression, conversion, ana-
purification procedure, much waste and the need for lytics and product purification, for the enzymatic pro-
[
8]
toxic solvents in the extraction process. cis-4- and duction of three different hydroxyprolines on a prepa-
cis-3-hydroxyproline, which are less common or un- rative scale. The enzymes were also used in vitro to
known in mammalian systems, can be obtained hydroxylate l-pipecolic acid, the six-membered ring
through epimerization reactions or chemical total syn- congener of l-proline, and the products were charac-
[9]
thesis. They are being clinically evaluated as anti- terized. Finally, an in vivo transformation of the non-
cancer drugs and for the treatment of thickening of
1376
asc.wiley-vch.de
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2011, 353, 1375 – 1383

A Simple Procedure for Selective Hydroxylation of l-Proline and l-Pipecolic Acid
natural substrate on a preparative scale is demonstrat- Characterization of Enzyme Reactions
ed.
Conversions of l-proline to hydroxyproline were mea-
sured by HPLC after FMOC-derivatization. The re-
verse-phase assay allows a baseline separation of all
four hydroxyprolines and proline (Figure 2). Concen-
trations from 20–120 mmol·L in the sample can be
quantified by fluorescence detection.
Results
ꢀ
1
Production of Recombinant Proline Hydroxylases
The proline hydroxylase genes were synthesized (in
For in vitro conversions we found that an appropi-
case of the GC-rich sequences cis-P3H and trans-P4H ate concentration of FeSO is crucial for proline hy-
4
codon optimization for E. coli was applied), ligated droxylase activity, more than that of substrate and co-
into several commercially available expression vectors substrate. Thus, the conversion was measured in de-
and tested for protein production. Strong expression pendence on the FeSO concentrations without and in
4
was observed in most cases, however, the proteins presence of l-ascorbate (Figure 3).
were insoluble and no activity was detected. Varia-
For all enzymes a distinct dependence of the total
tions of cultivation conditions or E. coli expression turnover to Fe(II) and ascorbate concentrations was
strains as well as denaturation techniques were not found. Optimal Fe(II) concentrations were around
successful. An overnight expression with the cold 0.1 mM for the cis-selective enzymes whereas trans-
shock vectors pCold I and pCold II at 168C gave a P4H was slightly more iron tolerant and had an opti-
reasonable fraction of soluble and active protein mum at ca. 0.5 mM. In contrast to prolyl hydroxylas-
(
data not shown). Best results were obtained by coex- es, ascorbate is not essential for catalysis with proline
[
4,14]
pression of the genes in the expression vector pET28b hydroxylases.
Nevertheless, the presence of l-as-
with the chaperone system GroEL/GroES (Figure 1, corbate in an appropriate amount increases the pro-
a). S. coelicolor GroEL/GroES expressed in vector duction of hydroxyprolines significantly. This effect
[
13]
pETcoco-2 was used in this study,
comparable yields were also obtained with the com- zymes is generally explained by the ability of ascor-
mercially available chaperone vector pG-KJE8 encod- bate to reduce iron(III) to the catalytically relevant
even though which is observed for many a-KG-dependent en-
AHCTUNGTRENNUNG
[
15]
ing GroEL/GroES from E. coli among other proteins. iron(II) species.
At ascorbate concentrations
Yields in a range of 5–10 mg protein/L medium were >3 mM, however, an inhibitory effect is dominating.
obtained from shake flask cultures after Ni-NTA pu- For instance, the conversion with trans-P4H at 6 mM
2+
rification and desalting for all three proteins ascorbate was lower than without (2 mM Fe in both
(
Figure 1, b).
cases). The inhibitory effect was confirmed by meas-
uring the total turnover of conversions with trans-
2+
P4H and 0.5 mM Fe in dependence on the ascorbate
concentration (see Supporting Information).
At adequate conditions a moderate, but virtually
constant activity over two hours was observed for all
Figure 1. a) Expression of trans-P4H in E. coli BL21 Codon Plus RP, soluble fractions; lane 1: trans-P4H/pET28b(+); lane
: trans-P4H/pET28b(+) + chaperones/pETcoco2; lane 3: chaperones only; lane 4: negative control. b) Purified trans-P4H,
cis-4PH and cis-3PH. Proline hydroxylase bands are marked with arrows.
2
Adv. Synth. Catal. 2011, 353, 1375 – 1383
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1377

Christian Klein and Wolfgang Hꢀttel
FULL PAPERS
Figure 2. Separation of hydroxyprolines in HPLC-chromatogram (Hyp=hydroxyproline; wavelength: excitation=254 nm,
emission=316 nm).
Figure 4. Hydroxyproline production with trans-P4H, cis-
P4H and cis-P3H. Optimal temperatures according to the
[4,11]
literature were applied for each enzyme.
three proline hydroxylases, although some protein
precipitated already minutes after start of the reaction
(
Figure 4). After 24 h no further activity could be de-
tected. Attempts to stabilize the enzymes with addi-
tives were not successful (data not shown). Neverthe-
less, a product concentration of 8 mM hydroxyproline
ꢀ
1
was achieved in the presence of approx. 0.5 mg·mL
trans-P4H. With the cis-selective enzymes activities
and total turnover were slightly lower under compara-
ble conditions.
In vivo Production of Hydroxyprolines
Hydroxyprolines were generated by feeding l-proline
Figure 3. Hydroxyproline production at different concentra-
tions of FeSO4 alone and in the presence of l-ascorbate
(
6.25 mM) to an exponentionally growing culture of
the corresponding E. coli strain and incubation at
(
molar ratio FeSO /l-ascorbate=1:3). Since concentrations
4
range over three orders of magnitude, a logarithmic scale 288C for three days. HPLC analysis of the product
was used. mixture revealed quantitative substrate conversion
1378
asc.wiley-vch.de
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2011, 353, 1375 – 1383

A Simple Procedure for Selective Hydroxylation of l-Proline and l-Pipecolic Acid
Table 1. Hydroxylation of l-proline in E. coli cultures ex-
pressing trans-4-, cis-3- and cis-4-proline hydroxylases.
Table 2. Conversion of l-pipecolic acid with trans-4-, cis-3-
and cis-4-proline hydroxylases.
Enzyme
Product
Conversion I[bs]olated
[a]
to product
Substrate:
Hydroxylase Product
Conversion to
product
cis-P3H
69%
61% (305 mg)
[a]
cis-P3H
cis-P4H
17%
cis-P4H
41%
63%
35% (175 mg)
51% (255 mg)
3
3%
trans-P4H
(
1)+33% (2)
[
[
a]
b]
5
00 mg l-proline were converted in 800 mL culture for 3
days at 288C. The conversion was determined by HPLC.
Since the product still contained inorganic salts after ion
exchange chromatography, yields were determined by
HPLC.
trans-P4H
68%
[a]
8
mg (1 mM) pipecolic acid were incubated with fresh
enzyme preparation for 14 h at 378C (cis-P3H, trans-
P4H) and 258C (cis-P4H), respectively. No other prod-
ucts were detected in this assay (HPLC, NMR).
and product concentrations >3.75 mM (60%) in the
case of trans-P4H and cis-P3H and 2.56 mM (41%)
for cis-P4H (Table 1). The loss of substrate is due to
metabolization by E. coli. It can be substantially re- In vitro Hydroxylation of l-Pipecolic Acid
duced by deletion of the proline dehydrogenase gene
[11a]
putA in E. coli.
To circumvent a laborious product In the scope of this project it should be demonstrated
purification including a separation from other amino that our system also applies to non-natural substrates.
acids, an amino acid-free minimal medium based on In the case of trans-P4H and cis-P3H a relatively re-
glycerol, acetate and ammonium sulfate was used. laxed substrate specificity has been found for proline
[
12]
Glucose was also avoided, since it suppresses the am- derivatives.
Especially the six-membered homo-
plification of the pETcoco-based chaperone vector. logue of l-proline, l-pipecolic acid, appeared to be a
The concentration of l-proline was set low enough to good substrate for these enzymes, and the hydroxylat-
ensure a quantitative conversion within 72 h incuba- ed products are of synthetic interest too. Thus l-pipe-
tion time. After that the hydroxyprolines were isolat- colic acid was chosen as model substrate. The selectiv-
ed from the culture broth by two steps of ion-ex- ity of the reactions was confirmed by NMR spectros-
change chromatography. In this way 80–90% of the copy of the isolated products. In coincidence with the
corresponding product was obtained. In the medium literature, trans-P4H gave trans-5-hydroxypipecolic
and also in the isolated product only the expected acid in a good yield (Table 1) whereas the conversion
[
12]
regio- and stereoisomer for the corresponding enzyme with cis-P3H was rather poor.
Yet the product
was found. Some inorganic salts, especially sodium could be unambiguously identified as the cis-3-hy-
1
1
chloride, were co-eluted with the product, but did not droxylated pipecolic acid by H- H-COSY-NMR spec-
affect follow up chemistry. If the salt-free compound troscopy. The coupling constant between H-2 and H-3
is required, this can be readily separated by further was 1.6 Hz, which is in good agreement with the cis-
chromatography or recrystallization. Occasionally product (7.1 Hz for the corresponding trans-product;
[
16]
proteinogenic amino acids, especially valine, were en- see also Supporting Information).
With cis-P4H,
riched in the medium and coeluted with hydroxypro- the conversion of l-pipecolic acid was about 55%
line. So far the reason for this is not fully understood, compared to that observed for l-proline under the
but it is evident that the amount of other amino acids identical conditions. Two products were obtained in
can be reduced to traces at optimal fermentation con- an approximate 1:1 ratio (Table 2 and HPLC chroma-
1
ditions (cf. H NMR spectra in Supporting Informa- togram in Supporting Information). To the best of our
tion).
knowledge, all conversions with proline hydroxylases
reported so far gave a single product exclusively.
Through H- H-COSY-NMR spectroscopy the com-
pounds were identified as the 3- and 5-hydroxylated
1
1
Adv. Synth. Catal. 2011, 353, 1375 – 1383
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1379

Christian Klein and Wolfgang Hꢀttel
FULL PAPERS
Table 3. Hydroxylation of l-pipecolic acid with recombinant E. coli producing trans-P4H.
Reaction Conversion to product Isolated yield
[a]
7
8%
61% (=35 mg)
[a]
Eluted in a fraction together with 13 mg trans-4-hydroxyproline. Product concentrations were
determined by HPLC.
congeners. In the HPLC assay one of the products Discussion
had the same retention time as cis-3-hydroxypipecolic
1
acid and the H NMR coupling constants were identi- Whereas heterologous expression of proline hydroxy-
cal (see Supporting Information). The second com- lases in E. coli with standard vectors yielded almost
pound was eluted approx. 2 min later than trans-5-hy- exclusively insoluble protein, coexpression with the
droxy-l-pipecolic acid indicating that the product is chaperone system GroES/GroEL gave a soluble pro-
1
ꢀ1
cis-5-hydroxylated. In the H NMR spectrum the cou- tein fraction of up to 10 mg·L medium. In analytical
pling constants were in good agreement with those of conversions the best total turnover was observed at
cis-5-hydroxy-l-pipecolic acid reported in the litera- iron(II) concentrations of 0.1–0.5 mM, lower than ap-
[17]
[4,11]
ture.
So for both products a cis-configuration can plied for these enzymes in the literature.
It was
be assigned, coinciding with the cis-selectivity of the found that iron(II) concentrations already in a low
enzyme.
millimolar range have a strong inhibitory effect. Be-
sides l-proline, the non-physiological substrate l-pipe-
colic acid was tested. From all conversions enough
product for a comprehensive chemical analysis includ-
ing 2D-NMR analysis was obtained, even though the
enzymatic activity was rather poor with cis-P3H.
In vivo Hydroxylation of l-Pipecolic Acid with trans-
P4H
Although some milligrams of hydroxypipecolic acids However, isolated proline hydroxylases have a rela-
were obtained through catalysis with isolated en- tively low specific activity and tend to denaturation
zymes, this method cannot be considered as an effi- under reaction conditions, so that in vitro conversions
cient synthetic approach and there is little potential in production scale are not profitable. In contrast, hy-
for up-scale. In the longer term, an in vivo method as droxylation of l-proline with recombinant E. coli
used for hydroxyprolines is desirable. However, l-pi- strains proved to be very efficient in an optimized in-
[
11]
pecolic acid is not a proteinogenic amino acid and the dustrial process. Thus, we established an analogue
uptake of the compound by living cells might be a method for our strains applicable in a laboratory
limiting factor. To test the general feasibility of this scale. l-Proline in a concentration of 6.25 mM was
approach, a 4 mM solution of l-pipecolic acid was quantitatively converted in shake flask cultures. Hy-
converted with an E. coli strain expressing trans-P4H droxyprolines were obtained as single regio- and ste-
under the same conditions as applied for proline hy- reoisomers in 41–69% yield in the medium depending
droxylation.
on the enzyme. Since an amino acid-free minimal
A surprisingly high product concentration was medium was used, the complex down-stream process
found in the fermentation broth (3.12 mM) and after for product purification applied in industry could be
ion exchange chromatography the product was ob- replaced by two straightforward ion exchange chro-
tained in 61% yield (Table 3). Neither substrate nor matography steps, from which 80–90% of the product
other regio- or stereoisomers of the product were was recovered. The product fractions still contained
found in the amino acid isolate. Since endogenously inorganic salts, especially sodium chloride, which did
produced l-proline was also hydroxylated by E. coli, not interfere in follow-up chemistry. With regard to
trans-4-hydroxyproline was obtained in the same frac- organic compounds hydroxyprolines could be ob-
tion as the product after chromatographic work-up tained in purities >95%. In a few cultures valine was
1
(
see HPLC chromatogram and H NMR spectrum in found in considerable amounts as a by-product, how-
Supporting Information).
ever, this can be avoided by careful control of the fer-
mentation conditions. The transformations described
here were not optimized for maximal yields and we
expect that product concentrations far beyond
ꢀ
1
0.6 g·L are feasible after some optimization.
1380
asc.wiley-vch.de
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2011, 353, 1375 – 1383

A Simple Procedure for Selective Hydroxylation of l-Proline and l-Pipecolic Acid
Surprisingly, conversions of l-pipecolic acid with Bacterial Strains, Plasmids and Proline Hydroxylase
cis-P4H yielded cis-3-hydroxy-l-pipecolic acid besides Genes
the expected cis-5-isomer. This demonstrates that the
Escherichia coli BL21 Codon Plus RP with chaperone ex-
catalytic properties of this enzyme are very similar to
pression vector pL1SL2 was kindly provided by Peter F.
that of cis-P3H, although the sequence identity is
[13]
Leadly (University of Cambrige, U.K.).
encoding plasmid pG-KJE8 (Takara Bio Inc., Japan) was
not related to any of the cis-selective enzymes (11– also used for coexpression. Expression plasmid pET-28b (+)
4% sequence identity), so that a different evolution- (Novagen, Darmstadt, Germany) was applied for gene clon-
The chaperone
[
18]
rather moderate (34%).
In contrast, trans-P4H is
1
ary origin can be postulated for this enzyme.
ing and expression. The genes of trans-P4H (GenBank No.
BAA20094) and cis-P3H type (Protein Data base:
E5RA) were synthesized by ATG:biosynthetics GmbH
Merzhausen, Germany) under optimized codon usage for
E. coli. The gene of cis-P4H (GenBank No. CAC47686) was
synthesized by the same company without C-terminal stop
codon for expression of the C-terminal HisTag in pET-28b
I
In the longer term, we are interested in the selec-
tive functionalization of non-physiological substrates
on a preparative scale. Therefore, we tested the feasi-
bility of our in vivo approach by converting a 4 mM
solution of l-pipecolic acid with trans-P4H producing
E. coli. 78% of the substrate was hydroxylated selec-
tively to trans-5-hydroxy pipecolic acid. This shows
1
(
(
+). All three genes were initially inserted into the SmaI
site of pBSK II. For cold-shock expression the vectors
that in principle amino acids other than l-proline can pCold I and pCold II (Takara Bio Inc., Japan) were used
be oxidized on a preparative scale too. Hydroxypro- and gene expression was performed according to the manu-
line produced by E. coli was found as a side product facturerꢃs instructions.
in this culture which was isolated with the product
after ion exchange chromatography. Currently we are Construction of the Expression Plasmids
investigating whether an advanced separation proce-
dure is required for product purification or if the for-
mation of hydroxyproline can be suppressed by opti-
mization of the fermentation conditions.
The proline hydroxylase genes were amplified from the corꢄ-
responding pBSK II-based plasmid by PCR with Phusion
High-Fidelity DNA Polymerase (New England Biolabs,
NEB, Ipswich, Massachusetts) using T7 and T3 standard pri-
mers. The PCR products were digested with SalI and NotI
(
NEB) in case of trans-P4H and cis-P3H and with NotI and
Conclusions
HindIII in case of cis-P4H. The DNA fragments were ligat-
ed into the corresponding sites of pET-28b (+) and verified
by sequencing (Eurofins MWG Operon, Ebersberg, Germa-
ny). The plasmids were transformed into E. coli BL21
Codon Plus RP containing pL1SL2, an expression vector for
the Streptomyces coelicolor chaperones GroEL 1, GroEL 2,
In summary, a comprehensive approach to three dif-
ferent proline hydroxylases including protein expres-
sion, biotransfomation, product purification, analytics
and their application in biocatalysis is presented. It
allows the synthesis of a diversity of hydroxyprolines
and hydroxylated proline derivatives with standard
laboratory equipment. It is highly flexible and new
isoenzymes or substrates are easily integrated. Where-
[13]
and GroES.
Preparation and Purification of Recombinant
Enzymes
as the scale of in vitro production is restricted by the Recombinant E. coli cells were grown aerobically at 378C
and 140 rpm in 800 mL Terrific Broth (TB) with ampicillin
intrinsically limited stability of the enzymes, hydroxy-
ꢀ
1
ꢀ1
(
(
100 mgmL ), chloramphenicol (34 mgmL ), kanamycin
1
prolines can be obtained efficiently on a preparative
scale through in vivo transformation of proline in
simple shake flask cultures. First experiments with l-
pipecolic acid suggest that non-natural substrates can
also be hydroxylated on a preparative scale.
ꢀ
50 mgmL ) and l-arabinose (0.01% w/v). At an OD
=
00
6
1
.2, isopropyl-ß-d-thiogalactopyranoside (IPTG) was added
to a final concentration of 0.2 mM and the cultures were in-
cubated for 16 h at 288C and 140 rpm. The cells were har-
vested by centrifugation (3300ꢅg, 20 min, 48C) and resus-
pended in 15 mL equilibration buffer [imidazole (20 mM),
glycerol (20% v/v), Mes (50 mM, pH 6.5), Tween 20 (0.1%
v/v)] and sonicated on ice for 7ꢅ10 s in 30 s intervals (Bran-
son Sonifier 250, Danbury, USA). After centrifugation
Experimental Section
(
9290ꢅg, 60 min, 48C), the enzymes were isolated from the
Chemicals and Culture Media
cell-free extract using Ni-NTA superflow chromatography
according to the manufacturerꢃs instructions (Qiagen
GmbH, Hilden, Germany). Imidazole was removed via PD-
Chemicals, antibiotics and culture media were purchased
from Sigma Aldrich Chemie GmbH (Munich, Germany), A.
Hartenstein GmbH (Wꢀrzburg, Germany) and Carl Roth
GmbH & Co. KG (Karlsruhe, Germany).
1
0 desalting columns (GE Healthcare, Piscataway, USA) so
that the enzymes were eluted in Mes buffer (50 mM,
pH 6.5). The concentrations of purified protein were mea-
sured with a Qubitꢆfluorometer (Invitrogen, Paisley, UK).
For conversions described here fresh enzyme preparations
were used. However, enzyme solutions or lyophyllisates can
Adv. Synth. Catal. 2011, 353, 1375 – 1383
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1381

Christian Klein and Wolfgang Hꢀttel
FULL PAPERS
ab ce ti sv ti ot yr .e d for months at ꢀ808C without significant loss off In vivo Production of Hydroxyprolines
E. coli production strains were grown aerobically in a 2-L
flask with 800 mL medium 5: glycerol (1% v/v), (NH ) SO
4
2
4
Derivatization and HPLC Assay
(
10 g/L), NaCl (2 g/L), K HPO (1 g/L), MgSO4 (1 g/L),
2 4
Hydroxyprolines and l-proline were analyzed by HPLC
NH Cl (5 g/L), sodium acetate (5 g/L), supplemented with
4
after precolumn derivatization (FMOC/ADAM) as fol-
ampicillin (100 mg/mL), chloramphenicol (34 mg/mL), kana-
mycin (50 mg/mL) and l-arabinose (0.01% w/v) at 378C and
140 rpm. At OD600 =1.1–1.4 IPTG was added to a final con-
centration of 0.2 mM. Subsequently, 500 mg l-proline
[19]
lows. To 40 mL of sample solution, 40 mL of sodium borate
buffer (0.5M, pH 7.7) were added. Then 80 mL of 9-fluoren-
yl-methoxycarbonyl chloride (FMOC-Cl, 1.5 mM in ace-
tone) were added and the mixture was vortexed for 60 s.
After that 100 mL of 1-adamantanamine (ADAM, 40 mM in
acetone and sodium borate buffer (pH 7.7), 1:1) were added
and the mixture was vortexed for 45 s. For analysis, the mix-
ture was diluted with 140 mL HPLC buffer A (20% acetoni-
(6.25 mM), 111 mg FeSO
7H O (0.5 mM) and 1.45 g a-KG
4
2
disodium salt·2H O (8 mM) were added. The cells were in-
2
cubated for further 72 h at 288C and 140 rpm. After 24 h
and 48 h 4 mL glycerol (6.6 mL/L) were supplemented.
tril/80% 50 mM sodium acetate pH 5). HPLC analysis was In vivo Production of trans-5-Hydroxypipecolic Acid
performed with an Agilent 1100 series HPLC system (Agi-
lent Technologies, Santa Clara, USA) equipped with an Agi-
lent fluorescence detector (1100 series) and a LiChrosorb
RP 18–5 m column (250ꢅ4.0 mm, CS GmbH, Langerwehe,
The transformation was performed as described for the pro-
duction of hydroxyprolines, but on a 100 mL scale. l-Pipe-
colic acid (52 mg, 4 mM) was added to the culture instead of
l-proline.
Germany) maintained at 458C. A gradient of buffer A and
buffer B (80% acetonitrile/20% 50 mM sodium acetate
pH 5) was applied (gradient program 1, buffer A: 1. 0–2 min
Isolation of Hydroxylated Products
1
2
3
00%; 2–17 min: 100%!91%; 17–25 min: 91%!70%; 25–
7 min: 70%!0%; 27–32 min: 0%; 32–34 min: 0%!100%;
4–40 min: 100%). The injection volume was 10 mL and the
The cultures were centrifuged (3300ꢅg, 45 min, 48C) and
the supernatant was adjusted with HCl to pH 1.3–1.6 and
passed through a column with 110 mL of cation-exchange
resin [Dowex 50 Wꢅ8, 50–100 mesh (H form)]. The frac-
tions containing hydroxylation products were passed
through a second column with 140 mL of anion-exchange
ꢀ
1
flow rate 0.75 mLmin . Derivatized hydroxylprolines were
detected fluorometrically at 254 nm excitation wavelength
and at 316 nm emission wavelength.
ꢄ
+
ꢄ
ꢀ
resin [Dowex 1ꢅ8, 20–50 mesh (OH form)]. The product
fractions were combined and concentrated under reduced
pressure and air-dried to obtain the products as a brownish
solid containing inorganic salts. Isolated yields determined
Assay for l-Proline
The in vitro activity assay of proline hydroxylases were car-
ried out in a reaction mixture containing Mes buffer
2
0
by HPLC: cis-3-Hyp: 305 mg (61%), [a] : ꢀ61.3 (c 0.3,
(
50 mM, pH 6.5), l-proline (8 mM), a-KG (14 mM), ferrous
D
2
0
D
H O); cis-4-Hyp: 175 mg (35%), [a] : ꢀ43.9 (c 0.3, H O);
2
2
sulfate (0.5 mM), l-ascorbate (1.5 mM) and enzyme prepa-
ration in a final volume of 1 mL. Reactions with trans-P4H
and cis-P3H were carried out at 378C and reactions with cis-
P4H at 258C. The hydroxyproline concentration was mea-
sured by HPLC as described above.
trans-4-Hyp: 255 mg, (51%).
Acknowledgements
Assay for l-Pipecolic Acid
We thank Prof. Michael Mꢀller for his generous support. The
skillful technical assistance by Ms. Olga Fuchs and Ms. Alex-
andra Walter is gratefully acknowledged. We also thank Mr.
Volker Brecht for measuring the NMR-spectra. This work
was financially supported by the Deutsche Bundesstiftung
Umwelt (DBU, ChemBioTech: AZ13234-32).
Conversions of l-pipecolic acid were performed in Mes
buffer (50 mM, pH 6.5), l-pipecolic acid (1 mM), a-KG
(
10 mM), ferrous sulfate (0.5 mM), l-ascorbate (1.5 mM)
and enzyme preparation in a final volume of 60 mL (4ꢅ
5 mL). Reactions with trans-P4H and cis-P3H were carried
out at 378C for 14 h and reactions with cis-P4H at 258C for
4 h. Reaction products were isolated by ion-exchange chro-
1
1
matography as described below. The hydroxylated l-pipecol-
ic acids were analyzed by HPLC as Fmoc derivatives (gradi-
ent program 2, buffer A: 0–2 min: 100%; 2–19 min: 100%
References
[
1] a) B. M. Trost, C. S. Brindle, Chem. Soc. Rev. 2010, 39,
600–1632; b) S. G. Zlotin, A. S. Kucherenko, I. P. Be-
!
75%; 19–25 min: 75%!45%; 25–27 min: 45%!0%; 27–
1
3
2 min: 0%; 32–34 min: 0%!100%; 34–40 min: 100%), by
letskaya, Russian Chem. Rev. 2009, 78, 737–784.
2] a) P. Remuzon, Tetrahedron 1996, 52, 13803–13835;
b) R. M. Johnston, L. N. Chu, M. Liu, S. L. Goldberg,
A. Goswami, R. N. Patel, Enzyme Microb. Technol.
mass spectroscopy (photospray ionization, Agilent Technol-
ogies, Santa Clara, USA) and NMR spectroscopy (Bruker
Avance DRX 400, Buker Biospin GmbH, Rheinstetten,
Germany).
[
2
009, 45, 484–490.
3] R. P. Hausinger, Crit. Rev. Biochem. Mol. Biol. 2004,
9, 21–68.
[
[
3
4] a) H. Mori, T. Shibasaki, K. Yano, A. Ozaki, J. Bacter-
iol. 1997, 179, 5677–5683; b) R. Hara, K. Kino, Bio-
1382
asc.wiley-vch.de
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2011, 353, 1375 – 1383

A Simple Procedure for Selective Hydroxylation of l-Proline and l-Pipecolic Acid
chem. Biophys. Res. Commun. 2009, 379, 882–886;
c) T. Shibasaki, H. Mori, S. Chiba, A. Ozaki, Appl. En-
viron. Microbiol. 1999, 65, 4028–4031.
T. Shibasaki, K. Ando, S. Chiba, U.S. Patent 5,854,040,
1998; c) A. Ozaki, H. Mori, T. Shibasaki, K. Ando, K.
Ochiai, S. Chiba, Y. Uosaki, U.S. Patent 6,413,748 B1,
2002.
[
[
[
5] I. J. Clifton, L. C. Hsueh, J. E. Baldwin, K. Harlos, C. J.
Schoefield, Eur. J. Biochem. 2001, 268, 6625–6636.
6] L. Petersen, R. Olewinski, P. Salmon, N. Connors,
Appl. Microbiol. Biotechnol. 2003, 62, 263–267.
7] a) C. C. Lawrence, W. J. Soby, R. A. Field, J. E. Bald-
win, C. J. Schoefield, Biochem. J. 1996, 313, 185–192;
b) H. Mori, T. Shibasaki, Y. Uozaki, K. Ochiai, A.
Ozaki, Appl. Environ. Microbiol. 1996, 62, 1903–1907.
8] Y. Izumi, I. Chibata, T. Itah, Angew. Chem. 1978, 90,
[12] T. Shibasaki, W. Sakuri, A. Hasegawa, Y. Uosaki, H.
Mori, M. Yoshida, A. Ozaki, Tetrahedron Lett. 1999,
4
0, 5227–5230.
[
[
[
13] L. Betancor, M. J. Fernandez, K. J. Weissman, P. F.
Leadly, ChemBioChem 2008, 9, 2962–2966.
14] R. Myllylꢁ, E.-R. Kuutti-Savolainen, K. I. Kivirikko,
Biochem. Biophys. Res. Commun. 1978, 83, 441–448.
15] a) R. Kuttan, A. N. Radhakrishan, Adv. Enzymol. 1973,
[
1
87–194; Angew. Chem. Int. Ed. Engl. 1978, 17, 176–
3
7, 273–347; b) L. Tuderman, R. Myllylꢁ, K. I. Kivirik-
1
83.
ko, Eur. J. Biochem. 1977, 80, 341–348.
[
9] a) G. Bellon, R. Berg, F. Chastang, A. Malgras, J. P.
Borel, Anal. Biochem. 1984, 137, 151–155; b) A.
Madau, G. Porzi, S. Sandri, Tetrahedron: Asymmetry
[
16] a) G. Kusano, H. Ogawa, A. Takahashi, S. Nozoe, K.
Yokoyama, Chem. Pharm. Bull. 1987, 35, 3482–3486;
b) C. Greck, F. Ferreira, J. P. Gen eˇ t, Tetrahedron Lett.
1
996, 7, 825–830; c) W. R. Ewing, B. D. Harris, K. L.
1
996, 37, 2031–2034.
17] C. Herdeis, E. Heller, Tetrahedron: Asymmetry 1993,
, 2085–2094.
Bhat, M. M. Joullie, Tetrahedron 1986, 42, 2421–2428;
d) J. Hausler, U. Schmidt, Liebigs Ann. Chem. 1979, 11,
[
[
4
1
881–1889.
18] Amino acid identities were determined by sequence
alignment using ExPASy Proteomics tools (data not
shown).
19] J. Kunte, E. A. Galinski, H. G. Trꢀper, J. Microbiol.
Methods 1993, 17, 129–136.
[
10] a) B. Lubec, M. Sternberg, R. Mallinger, W. Radner,
W. Vycudilik, J. Hꢁusler, G. Lubec, Amino Acids 1993,
4
, 249–254; b) L. Metzner, J. Kalbitz, M. Brandsch, J.
[
Pharmacol. Exp. Ther. 2004, 309, 28–35.
[
11] a) T. Shibasaki, H. Mori, A. Ozaki, J. Biosci. Biotech-
nol. Biochem. 2000, 64, 746–750; b) A. Ozaki, H. Mori,
Adv. Synth. Catal. 2011, 353, 1375 – 1383
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1383