Library Selection with a Randomized Repertoire of (βα)8-Barrel Enzymes Results in Unexpected Induction of Gene Expression

Article abstract of DOI:10.1021/acs.biochem.9b00579

The potential of the frequently encountered (βα)₈-barrel fold to acquire new functions was tested by an approach combining random mutagenesis and selection in vivo. For this purpose, the genes encoding 52 different phosphate-binding (βα)₈-barrel proteins were subjected to error-prone PCR and cloned into an expression plasmid. The resulting mixed repertoire was used to transform different auxotrophic Escherichia coli strains, each lacking an enzyme with a phosphate-containing substrate. After plating of the different transformants on minimal medium, growth was observed only for two strains, lacking either the gene for the serine phosphatase SerB or the phosphoserine aminotransferase SerC. The same mutants of the E. coli genes nanE (encoding a putative N-acetylmannosamine-6-phosphate 2-epimerase) and pdxJ (encoding the pyridoxine 5′-phosphate synthase) were responsible for rescuing both ΔserB and ΔserC. Unexpectedly, the complementing NanE and PdxJ variants did not catalyze the SerB or SerC reactions in vitro. Instead, RT-qPCR, RNAseq, and transcriptome analysis showed that they rescue the deletions by enlisting the help of endogenous E. coli enzymes HisB and HisC through exclusive up-regulation of histidine operon transcription. While the promiscuous SerB activity of HisB is well-established, our data indicate that HisC is promiscuous for the SerC reaction, as well. The successful rescue of ΔserB and ΔserC through point mutations and recruitment of additional amino acids in NanE and PdxJ provides another example for the adaptability of the (βα)₈-barrel fold.

Full text of DOI:10.1021/acs.biochem.9b00579

Article
pubs.acs.org/biochemistry
Cite This: Biochemistry XXXX, XXX, XXX−XXX
Library Selection with a Randomized Repertoire of (βα)8‑Barrel
Enzymes Results in Unexpected Induction of Gene Expression
Bettina Rohweder,† Gerhard Lehmann,‡ Norbert Eichner,‡ Tino Polen,§ Chitra Rajendran,†
Fabian Ruperti,† Mona Linde,† Thomas Treiber,‡ Oona Jung,∥ Katja Dettmer,∥ Gunter Meister,‡
Michael Bott,§ Wolfram Gronwald,∥ and Reinhard Sterner*,†
†
̈
Institute of Biophysics and Physical Biochemistry, University of Regensburg, Universitatsstrasse 31, D-93053 Regensburg, Germany
‡
̈
Institute of Biochemistry, Genetics and Microbiology, University of Regensburg, Universitatsstrasse 31, D-93053 Regensburg,
Germany
§IBG-1: Biotechnology, Institute of Bio- and Geosciences, Forschungszentrum Ju
̈
lich GmbH, D-52425 Ju
̈
lich, Germany
∥
̈
Institute of Functional Genomics, University of Regensburg, Universitatsstrasse 31, D-93053 Regensburg, Germany
S
* Supporting Information
ABSTRACT: The potential of the frequently encountered
(βα)8-barrel fold to acquire new functions was tested by an
approach combining random mutagenesis and selection in vivo.
For this purpose, the genes encoding 52 different phosphate-
binding (βα)8-barrel proteins were subjected to error-prone
PCR and cloned into an expression plasmid. The resulting
mixed repertoire was used to transform different auxotrophic
Escherichia coli strains, each lacking an enzyme with a
phosphate-containing substrate. After plating of the different
transformants on minimal medium, growth was observed only
for two strains, lacking either the gene for the serine
phosphatase SerB or the phosphoserine aminotransferase SerC. The same mutants of the E. coli genes nanE (encoding a
putative N-acetylmannosamine-6-phosphate 2-epimerase) and pdxJ (encoding the pyridoxine 5′-phosphate synthase) were
responsible for rescuing both ΔserB and ΔserC. Unexpectedly, the complementing NanE and PdxJ variants did not catalyze the
SerB or SerC reactions in vitro. Instead, RT-qPCR, RNAseq, and transcriptome analysis showed that they rescue the deletions
by enlisting the help of endogenous E. coli enzymes HisB and HisC through exclusive up-regulation of histidine operon
transcription. While the promiscuous SerB activity of HisB is well-established, our data indicate that HisC is promiscuous for
the SerC reaction, as well. The successful rescue of ΔserB and ΔserC through point mutations and recruitment of additional
amino acids in NanE and PdxJ provides another example for the adaptability of the (βα)8-barrel fold.
successfully recapitulated known multicopy suppression events
he mechanisms underlying the acquirement of new
Tenzymatic functions during evolution are still debated. and predicted additional ones, several of which were validated
in vitro.4
One established scenario is the duplication of existing genes.
We wanted to pursue a similar approach, as described.2
This duplication is followed by mutational diversification in
However, instead of using wild-type proteins to restore growth
of deletion strains, we aimed to increase the tested sequence
space through random mutation further. In order to keep
library size manageable, we focused on enzymes belonging to
the (βα)8-barrel fold (Figure 1A). The (βα)8-barrel or TIM
barrel fold is the most common enzyme fold and is found in
five of the six enzyme classes defined by the Enzyme
Commission (oxidoreductases, transferases, lyases, hydrolases,
and isomerases), illustrating its high adaptability and
versatility.5 It is therefore interesting to check the evolutionary
potential of this fold by laboratory experiments. Due to their
one of the daughter genes that leads to the enhancement of an
initially low promiscuous activity.1 Hence, the enzymatic
repertoire of a cell such as Escherichia coli could contain many
promiscuous members. This hypothesis was tested by a
comprehensive experiment in which researchers tried to
complement the growth deficiency of 104 E. coli deletions
strains through systematic overexpression of the whole E. coli
proteome (“multi-copy suppression experiment”).2 For this
purpose, a complete set of 4123 E. coli K-12 genes (the ASKA
library),3 which were cloned into an expression vector, was
used. For as many as 21 deletion strains, growth could be
restored not only by expressing the respective wild-type gene
but also due to the expression of other E. coli genes encoded on
an ASKA plasmid. A method for predicting promiscuous
functions, which is based on an unsupervised BLAST-search,
Received: July 7, 2019
Revised: September 25, 2019
Published: September 26, 2019
© XXXX American Chemical Society
A
DOI: 10.1021/acs.biochem.9b00579
Biochemistry XXXX, XXX, XXX−XXX

Biochemistry
Article
Figure 1. (A) Cartoon representation of the name-giving member of the (βα)8-barrel enzyme superfamily, triose phosphate isomerase TIM (PDB
code 4IOT). The eight central β-sheets forming the barrel structure are colored yellow, while the eight surrounding helices are colored blue. A
sulfate ion that occupies the phosphate-binding pocket formed by main chain NH-groups of active-site loops is shown as a stick model. (B)
Overview of library generation and selection for novel function by complementation of auxotrophic E. coli strains. (C) Reactions catalyzed by
phosphoserine aminotransferase (SerC) and phosphoserine phosphatase (SerB). Abbreviations: 3-p-OH-pyr: 3-phosphohydroxypyruvate; Glu:
glutamate; 2OG: 2-oxoglutarate. (D) The reactions catalyzed by histidinol-phosphate aminotransferase (HisC) and L-histidinol-phosphate
phosphohydrolase (HisB) are chemically similar to the reactions catalyzed by SerC and SerB.
intrinsic stability to mutations, (βα)8-barrels have been proven
ideal targets for directed evolution or protein design
approaches.6,7 We further limited our initial gene pool to
(βα)8-barrel enzymes containing a phosphate-binding site in
their active center. Phosphate is the most common functional
group in the E. coli metabolome, with 35−40% of all
metabolites containing a phosphate group.8 We reasoned
that this existing binding site could function as an anchor that
helps to bind novel substrates and thus facilitates the
enhancement of promiscuous activity levels beyond the
detection limit of the growth assays.
Hence, the ASKA library plasmids corresponding to 52
phosphate-binding (βα)8-barrel proteins from E. coli were
pooled, diversified by error-prone PCR, and used for the
selection of novel functions through complementation of
auxotrophic E. coli deletion strains. Cell growth was observed
for two strains, lacking either the gene for the serine
phosphatase SerB or the phosphoserine aminotransferase
SerC. It turned out that both, the mutated gene pdxJ
(encoding the pyridoxine 5′-phosphate synthase) and the
mutated gene nanE (encoding a putative N-acetylmannos-
amine-6-phosphate 2-epimerase) were responsible for rescuing
both ΔserB and ΔserC. However, instead of catalyzing the SerB
and SerC reactions, the selected PdxJ and NanE variants
unexpectedly induced the expression of the histidine operon,
thus enlisting the help of the endogenous E. coli enzymes
histidinol phosphatase HisB and histidinol phosphate amino-
transferase HisC. While it is established that HisB is
promiscuous for the SerB reaction2,9 our data shows as well
that HisC is promiscuous for the SerC reaction.
Archive)3 was obtained from the National Institute of Genetics
(1111, Yata, Mishima, Shizuoka 411-8540 Japan). Strains were
grown in LB medium or M9-glucose minimal medium (1×
M9-salts, 2 mM MgSO4, 0.1 mM CaCl2, 0.4% Glucose).
Antibiotics and additional substrates were added to the
following concentrations: ampicillin (150 μg/mL), kanamycin
(50 μg/mL), chloramphenicol (30 μg/mL), IPTG (0.5 mM),
pyridoxine (0.1 mg/L), L-histidine, L-histidinol, L-alanine, L-
threonine, L-lysine, sodium acetate, propionic acid, maltotriose,
and 2,6-diaminopimelic acid (12 μg/mL). For ΔserC deletions,
pyridoxine was added to the plates (Table S1). SerC has two
enzymatic functions: Apart from serine biosynthesis, it also
catalyzes a step in the biosynthesis of pyridoxal-5-phosphate.
For ΔhisB and ΔhisC deletions, L-histidinol was added (Table
S1), to allow selection for complementation in the absence of
HisB or HisC, respectively. We used L-histidinol (the product
of the HisB reaction) instead of L-histidine, as L-histidine
functions as a native regulator that downregulates the
expression of the his operon. (With L-histidine added to the
growth plate, wild-type HisB fails to complement ΔserBΔ-
hisB.) Plates were kept at 37 °C up to 4 weeks and checked
regularly for the appearance of colonies. For growth on solid
media, 15 mg/mL agar was added to the growth medium. 14C-
phosphoserine was obtained from Hartmann Analytics. All
other reagents were purchased from Sigma.
Strains. The parent strain E. coli BW25113 and the deletion
strains ΔserB, ΔserC, ΔaroA, ΔaroC, ΔpdxA, ΔpurE, ΔpyrB,
and ΔserA in the BW25113 background (all gene::kan), were
obtained from the National BioResource Project (NIG,
Japan): E. coli (1111, Yata, Mishima, Shizuoka 411−8540
Japan).10 All other deletions strains (ΔserBΔhisB [ΔserB::kan
ΔhisB::cat], ΔserBΔgph [ΔserB::kan Δgph::cat], ΔserBΔytjC
[ΔserB::kan ΔytjC], ΔserBΔserC [ΔserB::kan ΔserC], and
ΔserCΔhisC [ΔserC ΔhisC::kana]) were constructed from the
parent strain BW25113 by the P1 transduction method.11
MATERIALS AND METHODS
■
Oligonucleotides, Growth Media, and Reagents.
Oligonucleotides were obtained from biomers.net. The
ASKA library-gfp (a complete set of E. coli K-12 ORF
B
DOI: 10.1021/acs.biochem.9b00579
Biochemistry XXXX, XXX, XXX−XXX

Biochemistry
Article
Primers used for strain construction and PCR confirmation are
listed in Table S2.
Enzyme Activity Assays. SerB activity was assayed by
monitoring released phosphate with a malachite green assay,
performed with purified protein, as described previously.14
Additionally, we measured phosphate release from 14C-
phosphoserine in a radiometric assay, and 1−4 μM purified
protein was incubated with 4.5 μM 14C-phosphoserine in 50
mM potassium acetate, 20 mM Tris-acetate, 10 mM
magnesium acetate, 100 μg/mL BSA (pH 7.9) for up to 15
h at 37 °C. The products were analyzed and visualized, as
described previously.15
Protein Crystallization. PdxJ(11) was crystallized using
standard hanging drop vapor diffusion methods. The protein
(9.6 mg/mL in 10 mM Tris-HCl (pH 7.5), 30 mM NaCl) was
supplemented with 5 mM phosphoserine, incubated at 4 °C
for 15 min before adding the crystallization buffer (7.5% PEG
6000 and 1.7 M NaCl), and incubation was continued at 18
°C. Crystals of PdxJ(11) grew within 1 week. After transferring
the crystals to 30% ethylene glycol as cryoprotectant, they were
flash-frozen in liquid nitrogen.
Data Collection, Structure Solution, and Refinement.
Data collection was done at Swiss Light Source (SLS),
Switzerland at beamline PXIII and PXI at cryogenic temper-
ature. The data processing was done using XDS,16 and the data
quality assessment was done using phenix.xtriage. Molecular
replacement was performed with Phaser17 within the CCP4i18
suite using 1HO1 and 1M5W as search models. Initial
refinement was performed using REFMAC.19 The model was
further improved in several refinement rounds using automated
restrained refinement with the program PHENIX20 and
interactive modeling with Coot.21 The refinement statistics
are listed in Table S5. The final model was analyzed using the
program MolProbity22 and deposited in the Protein Data Bank
(6RG0).
Library Construction and in Vivo Complementation.
ASKA library clones (vector pCA24N-gfp) were grown
overnight in LB medium with added chloramphenicol. All 52
plasmids (complete list in Table S3) harboring a gene
encoding for a phosphate-binding (βα)8-barrel were prepared
individually (GeneJET Plasmid Miniprep Kit, Thermo Fisher)
and subsequently mixed in equimolar concentrations. This
pool (150 ng/μL) was used as a template for the random-
ization by the error-prone PCR reaction. For the PCR reaction,
50 ng DNA template, 15 U GoTaq, 1× GoTaq buffer, 0.8 mM
MnCl2, 1 mM MgCl2, 0.35 mM dATP, 0.40 mM dCTP, 0.20
mM dGTP, 1.35 mM dTTP, and 0.2 μM of each primer were
added to a total volume of 50 μL. Thirty PCR cycles were
performed (denaturation 95 °C for 45 s, annealing 56 °C for
45 s, elongation 72 °C for 5 min). The resulting mixed
repertoire was cloned into the expression plasmid
pTNA_BsaI12 allowing for low, constitutive expression in E.
coli. The plasmids were used to transform E. coli BW25113
cells, which were then plated on medium containing ampicillin
and incubated at 37 °C. Grown colonies were scratched off the
plates and suspended, followed by the isolation of the (βα)8-
barrel enzyme library. As estimated from the number of grown
colonies and the ligation efficiency tested by colony PCR, the
library contained about 1.4 × 108 independent mutants. To
test for in vivo complementation, electrocompetent cells from
nine different auxotrophic strains were transformed with the
diversified (βα)8-barrel enzyme library, according to standard
protocols. After recovery (1 h shaking in SOC), cells were
washed three times with 1% NaCl and plated on solid M9-
glucose minimal medium supplemented with additives if
necessary and incubated at 37 °C.
Site-Directed Mutagenesis. Single point mutations and
frameshift mutations were introduced by the QuickChange
Site-Directed Mutagenesis System (QCM) protocol developed
by Stratagene (La Jolla, CA) with the primers listed in Table
S2. For the purification of FLAG-tagged proteins PdxJ(X3)
and wild-type PdxJ(+Ext), the corresponding genes were
cloned from pTNA_BsaI into vector pUR21 (Table S4) and
an N-terminal Flag-tag was introduced with the forward
primer.
Protein Expression, Purification, and Characteriza-
tion. Selected mutant genes were recloned for expression into
vector pET21a and used to transform E. coli BL21 cells. The
LB medium (1 L) was inoculated with an overnight culture
and induced with IPTG at an optical density at 600 nm of
about 0.5. Protein expression occurred overnight at 20 °C.
Cells were harvested and disrupted by sonification in 50 mM
Tris-HCl (pH 7.5), 300 mM KCl, and 20 mM imidazole.
Proteins were purified from the soluble fraction of the cell
extracts by Ni2+-affinity (all variants retain the N-terminal
hexahistidine tag of the ASKA library) in 50 mM Tris-HCl
(pH 7.5), 300 mM KCl, using a linear gradient from 20 mM to
500 mM imidazole for protein elution followed by preparative
size exclusion chromatography (25 mM Tris-HCl (pH 7.5),
300 mM KCl). Fractions containing at least 95% pure protein
as judged by SDS-PAGE were pooled, dialyzed against 25 mM
Tris-HCl (pH 7.5), 100 mM KCl, and stored at −80 °C after
flash-freezing in liquid nitrogen. Protein concentrations were
determined by absorption spectroscopy using a molar
extinction coefficient at 280 nm that was calculated from the
amino acid sequence.13
RT-qPCR. Cells were grown in LB medium to mid log
phase. After harvesting, cells were washed three times with 1%
NaCl and used to inoculate minimal M9-glucose medium and
shaken at 30 °C. After the culture reached an optical density at
600 nm of 0.5 cells were mixed with two volumes of
RNAprotect Bacteria Reagent (Qiagen), harvested and stored
at −80 °C. Cells were lysed in RNase-free TE buffer (30 mM
Tris-HCl, 1 mM EDTA, pH 8.0) containing 1 mg/mL
lysozyme for 30 min at room temperature. Total RNA was
prepared using an RNeasy Mini Kit (Qiagen). cDNA was
prepared from 1 μg RNA with the QuantiTect Reverse
Transcription Kit (Qiagen). Genomic DNA was eliminated
prior to the reverse transcriptase reaction according to the Kit’s
protocol. RT-qPCR reactions were carried out in a final
volume of 20 μL containing 1 μL of 1:5 diluted cDNA, 1 μL of
10 μM stock of each primer (primer pairs listed in Table S2),
and 10 μL of SsoFast EvaGreen Supermix (BIO-RAD). The
reactions were measured in a Bio-Rad Cfx96 qPCR system
using 40 cycles of a two-step amplification protocol, with two
technical replicates per sample. To determine relative
expression levels, transcript levels were normalized to the
23S RNA. Samples without reverse transcriptase enzyme were
used to control for specificity.
RNAseq. Cells were grown in LB medium to mid log phase.
After harvesting, cells were washed three times with 1% NaCl
and used to inoculate minimal M9-glucose medium and shaken
at 30 °C. After the culture reached an optical density of 0.5
cells were mixed with two volumes of RNAprotect Bacteria
Reagent (Qiagen), harvested and stored at −80 °C. Cells were
lysed in RNase-free TE buffer (30 mM Tris-HCl, 1 mM
C
DOI: 10.1021/acs.biochem.9b00579
Biochemistry XXXX, XXX, XXX−XXX

Biochemistry
Article
EDTA, pH 8.0) containing 1 mg/mL lysozyme for 30 min at
room temperature. Total RNA was prepared using an RNeasy
Mini Kit (Qiagen). Genomic DNA was removed by RNase-
free DNase (Qiagen). The RNA quality was assayed using a
NanoDrop and a 4200 TapeStation system (Agilent Tech-
nologies Inc.). Ribosomal RNA was depleted using a Ribo-
Zero rRNA Removal Kit Bacteria (Illumina) and successful
rRNA depletion verified by TapeStation. The library was
prepared using a NEBNext Ultra II Directional RNA Library
Prep Kit for Illumina (NEB) with NEBNext Multiplex Oligos
for Illumina (Index Primer Set 1) (NEB). After quality and
quantity assessment on the TapeStation, the libraries were
pooled equimolar. The resulting library pool was quantified
with KAPA library quantification kit (Roche) and finally
sequenced on a MiSeq sequencer (Illumina) performing a 2 ×
80 cycles paired-end run. Of the sequences obtained, the
adapters were trimmed with SeqPrep v1.3.2−1 (github.com/
jstjohn/SeqPrep). The trimmed reads were mapped using the
Bowtie 2 aligner23 with default settings to the E. coli genome
(www.ecogene.org). Samtools v1.3.124 was used to remove
optical duplicates. For counting HTseq-count v0.9.125 was
used against an annotation database (www.ecogene.org_Ecoli_
annotation_110217-065029). Differential analysis was carried
out with R (www.R-project.org) and the Bioconductor26
package DESeq2.27
DNA Microarray Analysis. E. coli BW25113 cells were
grown in minimal M9-glucose medium (with no additional
supplements added) as described above. When the culture
reached an optical density of 0.5, cells were harvested by
centrifugation (3 min, 5000g) with crushed ice to ensure rapid
cooling. The pelleted cells were shock-frozen in liquid nitrogen
and stored at −80 °C. Total RNA was isolated using the
RNeasy system (Qiagen) and quality-checked as described
previously.28 The synthesis of Cy3- or Cy5-labeled cDNA and
two-color hybridization to DNA microarrays was carried out,
as described.28 Custom-made 4 × 44K DNA microarrays were
obtained from Agilent Technologies and were designed using
Agilent’s eArray platform (https://earray.chem.agilent.com/
earray). The array design comprises oligonucleotides for the
annotated genes of E. coli MG1655 and other bacterial
genomes, as well as Agilent’s control spots. After hybridization,
the arrays were washed using Agilent’s wash buffer kit, and the
fluorescence was determined at 532 nm (Cy3-dUTP) and 635
nm (Cy5-dUTP) at 5 μm resolution (GenePix 4000B laser
scanner, GenePix Pro 6.0 software, Molecular Devices). After
quantitative image analysis using the corresponding Agilent’s
gene array list, results containing the non-normalized ratio of
median values were saved as GPR file (GenePix Pro 6.0).
Subsequently, BioConductor R-packages limma and marray
(http://www.bioconductor.org) were used to achieve back-
ground correction, loess-normalization of ratios, and diagnostic
plot generation for array quality control. To check for
differentially expressed genes according to the normalized
ratios the quality criteria (i) Flags ≥ 0 and (ii) signal/noise of
Cy5 (F635Median/B635Median) or Cy3 (F532Median/
B532Median) ≥ 3 were applied. For each comparison
pdxJ(11)/pdxJ(11-Ext) and pdxJ(X3)/wild-type pdxJ(+Ext)
two independent biological samples obtained from separate
cultures were used.
Co-IP, anti-FLAG M2 agarose (Sigma-Aldrich, Deisenhofen)
was used. Prior to use, the matrix was washed twice with cold
PBS and once with the lysis buffer (150 mM KCl, 25 mM Tris
pH 7.5, 2 mM EDTA, 1 mM NaF, 0.5% NP-40, 1 mM DTT, 1
mM AEBSF). The E. coli cultures were washed twice in ice-
cold PBS and lysed by resuspending in lysis buffer. Input
samples of the lysates were taken, mixed with SDS-PAGE
sample buffer, and stored at −20 °C. For immunoprecipitation,
1 mL of the lysate was incubated with 50 μL of the antibody-
coupled beads for 2 h while rotating at 4 °C. After incubation,
the affinity matrix was sedimented by centrifugation for 1 min
at 1000g, and the supernatant was removed. The beads were
incubated four times with the wash buffer (300 mM NaCl, 50
mM Tris (pH 7.5), 0.5% NP-40, 1 mM MgCl2). After the third
washing step, the samples were transferred to new reaction
tubes to minimize contamination by unspecific protein binding
to the tube material. After a final washing step with PBS, the
supernatant was removed quantitatively, and the beads were
eluted by addition of 5 μL of 3× Flag peptide (Sigma-Aldrich,
Deisenhofen) and 35 μL of PBS and rotating for 2 h at 4 °C.
After elution, the affinity matrix was sedimented by
centrifugation for 1 min at 1000g, and the supernatant was
transferred into a new cup. A second elution step with 20 μL of
0.1 M glycine (pH 2.6) for 5 min at room temperature
(stopped by the addition of 1 μL of 1 M NaOH), was applied.
Again, the affinity matrix was sedimented by centrifugation for
1 min at 1000g and the supernatant was transferred into a new
cup. The input (about 15 μg) and the supernatants of the
elution steps were analyzed by SDS-PAGE. Of the supernatant
of the first elution step (approximately 50 μL), 25 μL were
loaded into a gel pocket, and voltage was applied. After near-
complete migration of the applied volume into the stacking gel,
the remaining 25 μL of the first elution step were loaded into
the same gel pocket. Detected protein bands were cut out,
digested with trypsin and analyzed via mass spectrometry.
NMR Metabolomics Analysis. E. coli BW25113 cells were
grown in minimal M9-glucose medium (with no additional
supplements added) as described above. Pelleted cells were
washed three times with 1% NaCl. Cell pellet and liquid
culture supernatant were stored at −20 °C. For metabolite
extraction, cell pellets were incubated overnight in 600 μL of
80% aqueous methanol. To each sample, 10 μL of 80 mM
nicotinate solution were added as extraction standard followed
by centrifugation at 9000g for 5 min. Supernatants were
collected, and remaining pellets were resuspended twice in 200
μL of 80% aqueous methanol followed by centrifugation and
collection of supernatants as described above. For each sample,
supernatants were combined and dried under a gentle stream
of nitrogen to complete dryness. Resulting pellets were
dissolved in 400 μL of deionized water and mixed with 200
μL of 0.1 mol/L phosphate buffer (pH 7.4) and 50 μL of 37.04
mmol/L trimethylsilyl-2,2,3,3-tetradeuteropropionate (TSP)
dissolved in deuterium oxide, the latter serving as an internal
standard for referencing and quantification. For the analysis of
cell culture supernatants, 400 μL of supernatant was mixed
with buffer and deuterium oxide as described above. Here, as
an internal standard for referencing and quantification, 10 μL
of 81.97 mmol/L formic acid was additionally added.
All subsequent NMR analyses were performed on a 600
MHz Avance III spectrometer (Bruker BioSpin, Rheinstetten,
Germany) employing a triple resonance (1H, 13C 15N, 2H lock)
cryogenic probe equipped with z-gradients and an automatic
cooled sample changer. Following established protocols, 1D
Protein Complex Immunoprecipitation (Co-IP). The
FLAG-tagged proteins PdxJ(X3) and wt PdxJ(+Ext) were
overexpressed for 9h in E. coli BW25113 cells in minimal M9-
glucose medium (with no additional supplements added). For
D
DOI: 10.1021/acs.biochem.9b00579
Biochemistry XXXX, XXX, XXX−XXX

Biochemistry
Article
1H NMR spectra were acquired employing either a 1D CPMG
pulse sequence29 for the analysis of cell culture supernatants or
a 1D NOESY pulse sequence30 for samples of cell pellet
extracts. Additionally, for unambiguous metabolite identifica-
To select for novel functions, different auxotrophic E. coli
strains were obtained either from the Keio collection10 or
constructed by using standard P1 transduction methods.11 We
chose auxotrophic strains deleted for enzymes that fulfill a role
in primary metabolism and bind substrates with a phosphate
moiety. The deleted enzymes had to adopt a different fold than
the (βα)8-barrel to exclude wild-type proteins from the library.
Last, in order to decrease the complexity of the selected
reactions, we tried to choose enzymes with as few cofactors as
possible. All used strains (ΔaroA, ΔaroC, ΔpdxA, ΔpurE,
ΔpyrB, ΔserA, ΔserB, ΔserC, ΔdeoC) were transformed with 2
μg (20 × 100 ng) of the diversified (βα)8-barrel enzyme
library, followed by the selection for growth on minimal
medium (Figure 1B). The medium was supplemented with
glucose and, in some cases, additional nutrients according to
the growth requirements of the individual strains (Table S1).
By plating an aliquot of the reaction mixture on rich medium
containing ampicillin, the transformation efficiency was
determined to be higher than 109 cells per μg of plasmid
DNA, ensuring that the entire library was successfully
transformed into each auxotroph.
Colony growth on the minimal medium was observed only
for the strains ΔserB and ΔserC, lacking the serine phosphatase
SerB or the serine phosphate aminotransferase SerC,
respectively (Table S1). SerC and SerB catalyze the final
two, consecutive steps in the biosynthesis of serine: the
conversion of 3-phosphohydroxypyruvate to phosphoserine
and the dephosphorylation of phosphoserine (Figure 1C).
Several hundred colonies appeared within 4−8 days for the
ΔserB complementation and several dozen colonies in 14−20
days for the ΔserC complementation. Prior to plasmid
isolation, about 50 arbitrarily selected colonies of ΔserB and
about 20 arbitrarily selected colonies of ΔserC were restreaked
on fresh minimal medium. Plasmids isolated from these fresh
colonies were used for the retransformation of ΔserB and
ΔserC. These retransformations confirmed the rescue of nine
ΔserB and the rescue of two ΔserC colonies; colonies
consistently grew within two to 5 days. The inserts of the
plasmids from these colonies were sequenced (Table S6).
In order to ensure that only the (βα)8-barrel enzyme coding
sequence was responsible for the rescue of ΔserB and ΔserC,
the genes encoded by the selected plasmids were amplified by
PCR and recloned into the vector pTNA_NN (Table S4). The
genes were again sequenced and once more tested for
complementation of ΔserB and ΔserC. Colonies again
consistently grew within 2−5 days.
1
tion, a 2D H−13C HSQC spectrum was acquired for each
sample. Metabolites were identified by comparison with
spectra of pure reference compounds employing CHENOMX
8.1 (Chenomx Inc., Edmonton, Canada), AMIX 3.9.13
(BrukerBioSpin, Rheinstetten, Germany) and Metabominer.31
1
Metabolites were quantified from 1D H spectra using AMIX
3.9.13 and Chenomx 8.1. To allow for the accurate
determination of peak integrals in the case of partially
overlapping signals, spectral deconvolution employing CHE-
NOMX 8.1 was used. Metabolites of cell culture supernatants
were quantified relative to the formic acid reference signal,
while for cell pellet extracts the TSP reference signal was
employed. Data from cell pellet extracts were normalized
relative to the extraction standard. For the identification of
signals corresponding to an unknown compound not available
1
in the used spectral libraries, additional 2D H−13C HSQC,
1
1
2D H−1H TOCSY, and 2D H−1H NOESY spectra were
acquired. Compound identification was further aided by
hyphenated mass spectrometry employing a Bruker maXis-
Impact quadrupole time-of-flight mass spectrometer (Bruker
Daltonics, Bremen, Germany) coupled to a Dionex Ultimate
3000 high-performance liquid chromatography system (Ther-
mo Fisher Scientific, Idstein, Germany). To this end, samples
were diluted with deionized water (1:4). Based on the exact
mass and fragment patterns together with the corresponding
NMR data the compound was unambiguously identified as N-
succinyl-L,L-2,6-diaminopimelate. Discriminating metabolites
were identified by a two-sided heteroscedastic Student’s t
test while controlling the false discovery rate at the 5% level
according to the method of Benjamini and Hochberg.32
Accession Numbers. The microarray data are accessible in
NCBI’s Gene Expression Omnibus33 through accession no.
GSE126228. The RNAseq data are accessible in NCBI’s Gene
Expression Omnibus through accession no. GSE133289. The
structure of PdxJ(11) was deposited in the Protein Data Bank
under accession code 6RG0.
RESULTS
■
In Vivo Complementation of Auxotrophic Deletions
Strains by a Mixed Library of (βα)8-Barrel Enzymes. To
construct a gene library of diversified (βα)8-barrel enzymes we
selected 52 pCA24N vectors from the ASKA library,3 each
harboring a gene for an E. coli phosphate-binding (βα)8-barrel
enzyme (Table S3). This mixed pool was amplified and
randomized by error-prone PCR, and the resulting mutants
were cloned into the expression vector pTNA_BsaI (Table
S4). This vector contained two distinct BsaI restriction sites
and was designed for the easy generation of large gene
libraries.12 The resulting diversified (βα)8-barrel enzyme
repertoire contained 1.4 × 108 different gene variants.
Sequencing of 20 randomly picked clones revealed that library
members contained 5.5 mutations per 1000 base pairs on
average, corresponding to 4.1 amino acid exchanges. The 20
randomly picked clones belonged to 20 different genes,
indicating that all genes from the initial pool are still present
and equally distributed in the library. Along the same lines,
more than 200 colony PCRs of random clones revealed a
broad range of insert sizes.
Variants of nanE and pdxJ Rescue ΔserB and ΔserC
Strains on Minimal Media. The sequencing results showed
that variants of two different genes, nanE (encoding a putative
N-acetylmannosamine-6-phosphate 2-epimerase) and pdxJ
(encoding the pyridoxine 5′-phosphate synthase), were
responsible for the observed in vivo complementation (Table
S6). The two nanE variants (nanE(7) and nanE(25)), which
were identified through their ability to rescue ΔserB, contained
3 and 6 amino acid exchanges. Variants of pdxJ were identified
through their rescue of ΔserB (7 variants) and or through their
rescue of ΔserC (2 variants). Remarkably, apart from one to
four different amino acid exchanges, all pdxJ variants contained
a ten amino acid extension at the C-terminus (Table S6). This
extension was recruited from the cloning vector pCA24N. Two
deletions in the sequence of the synthesized primers resulted in
a frameshift, leading to the loss of the wild-type stop codon
and the recruitment of an alternative stop codon from the
E
DOI: 10.1021/acs.biochem.9b00579
Biochemistry XXXX, XXX, XXX−XXX

Biochemistry
Article
Figure 2. (A) Sequence of the ten amino acid extension of selected pdxJ variants. The extension is shaded in green. The first 16 bases of the
extension are recruited from the primer sequence; all point mutations or single nucleotide deletions that differentiate selected variants occurred in
the primer sequence. The last 14 bases of the extension along with the stop codon are recruited from the sequence of vector pCA24N. Sequences of
the C-terminal extension of additional pdxJ variants are documented in Figure S1. (B) Growth of deletion strains on M9 minimal medium
transformed with different constructs. The variants pdxJ(11) and pdxJ(X3) rescue the ΔserB strain as well as the double deletion strain
ΔserBΔserC. NanE(7) rescues ΔserB. Constructs lacking the C-terminal extension (pdxJ(11-Ext)), or wild-type pdxJ(+Ext), and wild-type nanE do
not rescue ΔserB. Three different cell concentrations were plated: undiluted (top third of the plates), 10−2 dilution (lower left third of the plates),
and 10−4 dilution (lower right third of the plates). The time until colonies appeared is given in Table 1. Additional complementation experiments
are documented in Figure S2.
Table 1. Overview of in Vivo Complementation of Various ΔserB and ΔserC Single and Double Deletion Strains by Different
a
Genes
gene
deletion strain
pdxJ(11)
pdxJ(X3)
wt pdxJ
wt pdxJ(+Ext)
nanE(7)
wt nanE
wt serB
wt serC
wt hisB
wt hisC
wt hisBhisC
ΔserB
ΔserC
2−3
4−6
4−6
2−3
2−3
−
2−3
4−6
4−6
2−3
2−3
−
−
−
−
NT
NT
NT
NT
−
−
−
NT
NT
NT
NT
3−5
20
−
3−5
3−5
−
−
−
−
NT
NT
NT
NT
1
−
2
−
NT
NT
NT
−
1
−
3
−
NT
NT
−
1
3
3
NT
NT
2
−
−
1
1
−
−
−
−
1
1
1
ΔserBΔserC
ΔserBΔgph
ΔserBΔytjC
ΔserBΔhisB
ΔserCΔhisC
−
−
−
−
23
23
a
The shown deletion strains were transformed with the different variants cloned in vector pTNA_BsaI, and growth times on minimal M9 medium
supplemented with glucose and supplements according to the requirements of the different strains (see Table S1) were recorded. The numbers
correspond to the days until colony growth was observed. “−“: no growth was observed after 30 days. “NT”: not tested.
vector sequence (Figure 2A, Figure S1). During construction
of the library and the selection experiments, we sequenced
more than 70 genes that had been cloned with the same batch
of primers. Yet, this ten amino acid extension was exclusively
observed in the pdxJ variants that were selected through
complementation of ΔserB and ΔserC.
F
DOI: 10.1021/acs.biochem.9b00579
Biochemistry XXXX, XXX, XXX−XXX

Biochemistry
Article
Table 2. Variants of pdxJ Most Frequently Referred to in This Study
gene
amino acid exchanges
C-terminal extension
initially selected by complementation of
pdxJ(11)
pdxJ(X3)
wt pdxJ
wt pdxJ(+Ext)
nanE(7)
E146 K, H198R, M224L, D229V
M224L
−
−
yes
yes
no
yes
no
ΔserB
ΔserC
not applicable
not applicable
ΔserB
S2L, Q6L, Q44R, I72T, L140P, H199R
wt nanE
−
no
not applicable
Subsequent complementation experiments showed that the
nanE and pdxJ variants that were selected via the rescue of
ΔserB could also rescue ΔserC, and vice versa. This was true
for all variants and is paradigmatically shown for pdxJ(11),
pdxJ(X3), and nanE(7) in Table 1.
Growth of the ΔserB strain complemented by pdxJ and nanE
variants was faster (2−3 days and 3−5 days, respectively) than
the growth of the ΔserC strain (4−6 days and 20 days,
respectively). The pdxJ variants most frequently referred to in
this study are listed in Table 2, and their in vivo
complementation of ΔserB is paradigmatically shown in Figure
2B.
Both, the mutations identified in the selected variants of
pdxJ and nanE and the C-terminal extension in the pdxJ
variants are required for the rescue of ΔserB and ΔserC: The
wild-type enzymes nanE and pdxJ, cloned into the same
expression vector as the gene library, do not confer growth.
Furthermore, wild-type pdxJ with the ten amino acid C-
terminal extension, but no additional mutations throughout the
gene (wt pdxJ(+Ext)), does not rescue the serB or serC
deletion either (Table 1). Vice versa, the ten amino acid C-
terminal extension is essential for the ability to complement
ΔserB and ΔserC. Complete removal of the extension totally
abolishes growth for all pdxJ variants tested (Table S6).
We proceeded with mutating the coding sequence of the ten
amino acid C-terminal extension of one fast-growing pdxJ
variant, pdxJ(11). If the extension is shortened by two or more
amino acids, the ability to rescue ΔserB and ΔserC is
completely lost (Table S7). Additionally, we found that
although the extension can be lengthened by one amino acid
without losing the ability to rescue ΔserB and ΔserC,
introducing the bulky amino acid tryptophan at the last
position of the extension results in total loss of the ability to
rescue ΔserB and ΔserC. Next, we constructed a gene library of
the variant pdxJ(11) where we randomized the amino acid at
position 10 of the extension and selected for complementation
of ΔserB. While most amino acids at this position led to rescue
of the deletion strain, a few, mostly bulky, amino acids
(phenylalanine, glutamate, glutamine, lysine, valine, and
cysteine), were never found at this position.
selected for their ability to complement ΔserB, can only
complement ΔserC with slower growth times compared to the
pdxJ variants and were not able to complement the double
deletion strain ΔserBΔserC, for unknown reasons.
Characterization of Purified NanE and PdxJ Variants.
Both isolated nanE variants and three isolated pdxJ variants
(with or without the C-terminal extension), along with the
wild-type genes serB, serC, nanE, and pdxJ (with and without
the extension), were cloned into the expression vector pET21a.
The corresponding recombinant proteins containing an N-
terminal hexahistidine tag were expressed in E. coli and could
readily be purified to homogeneity from the soluble cell extract
using metal chelate affinity chromatography. Circular dichro-
ism spectroscopy and analytical gel filtration chromatography
showed that all purified proteins were well folded and
confirmed that the amino acid exchanges and the C-terminal
extension did not change the oligomerization state compared
to the wild-type proteins. Along the same lines, the structure of
the variant PdxJ(11) as determined by X-ray crystallography
(PDB code 6RG0) turned out to be similar to the structure of
wild-type PdxJ (PDB code 1M5W), with an overall rmsd of 2.0
Å for 239 Cα-atoms (Figure S3). Unfortunately, the 10 amino
C-terminal extension is not visible in the structure, indicating
high structural flexibility.
The results of the in vivo complementation described above
suggested that the selected NanE and PdxJ variants possess
SerB activity. In order to test this, we performed two different
assays with the purified proteins. However, neither in end-
point assays using radio-labeled phosphoserine as the substrate,
nor by detecting free phosphate cleaved from phosphoserine
with malachite green,14 could we detect any enzymatic activity.
Addition of various divalent metal ions or of crude cell extract
did not result in activity either. Wild-type SerB from E. coli was
used as positive control and showed activity levels comparable
to previous studies.34,35 We concluded that the selected nanE
and pdxJ variants do not rescue ΔserB by providing the missing
enzymatic function.
The NanE and PdxJ Variants Rescue ΔserB and ΔserC
by Recruiting HisB and HisC. Previous studies have shown
that novel sequences can rescue deletion strains not by
providing the missing enzymatic function, but by recruiting an
endogenous, promiscuous E. coli enzyme.36,37 In E. coli, three
phosphatases, HisB, Gph, and YtjC, can rescue ΔserB when
overexpressed.2 To test whether any of these phosphatases play
a role in the rescue of ΔserB by the nanE and pdxJ variants, we
constructed the double deletion strains ΔserBΔhisB,
ΔserBΔgph, and ΔserBΔytjC. Whereas, the additional deletion
of the phosphoglycolate phosphatase Gph or the putative
phosphoglycerate mutase YtjC had no effect on the rescue of
ΔserB, the double deletion strain ΔserBΔhisB could no longer
be complemented by our selected nanE or pdxJ variants (Table
1). This finding suggests that the nanE and pdxJ variants
recruit the promiscuous histidinol phosphatase HisB, which
To confirm that the variant protein PdxJ, and not the
mRNA, is responsible for the observed phenotype, we mutated
the first bases of pdxJ(11) to introduce either an early stop
codon or an early frame shift. In both cases, the sequence of
the translated protein is severely affected, whereas the mRNA
remains practically identical. Growth assays showed that both
mutants no longer rescue ΔserB or ΔserC.
Furthermore, we performed complementation experiments
with a newly constructed double deletion strain ΔserBΔserC.
As expected, the pdxJ(11) and pdxJ(X3) variants could rescue
ΔserBΔserC, illustrating that these variants are simultaneously
capable of complementing the lack of both serine biosynthesis
enzymes (Figure 2B, Table 1). The nanE variants, which were
G
DOI: 10.1021/acs.biochem.9b00579
Biochemistry XXXX, XXX, XXX−XXX

Biochemistry
Article
Figure 3. Expression of pdxJ(11) increases histidine operon mRNA levels up to 7−16-fold. Bars show the abundance of transcripts in cells
expressing pdxJ(11) relative to the same cell line expressing pdxJ(11-Ext) as a negative control. Genes of the histidine operon are colored orange.
(A) qRT-PCR identifies changes in relative mRNA levels of the four amino acid biosynthesis genes hisB, hisC, trpCF, and proA upon expression of
pdxJ(11). Addition of L-histidine (12 μg/mL) to the growth medium prevents induction of his operon expression through overexpressed pdxJ(11).
Expression levels were normalized with the housekeeping 23S rRNA gene. (B) DNA microarray analysis and RNAseq identifies fold changes in
mRNA levels. Shown are differentially expressed genes (more than 2-fold change in DNA microarray analysis and p-adjust value < 0.01 and more
than a 2-fold change in RNAseq) in cells expressing pdxJ(11) compared to the negative controls lacking the extension (pdxJ(11-Ext)). DNA
microarray: Values given correspond to average fold changes and the standard deviation calculated from two arrays of two independent biological
replicates. RNAseq: Values are based on two to three independent experiments, each with biologically independent cultures. All DNA microarray
data are shown in Table S8, and all RNAseq data are shown in Table S9.
dephosphorylates histidinol-phosphate in the biosynthesis of
histidine (Figure 1D) and is known to catalyze the SerB
reaction.9,36
Analogous to the RT-qPCR measurements described above,
total RNA was isolated from the pseudo wild-type strain
BW25113 expressing either pdxJ(11), pdxJ(X3), nanE(7), or
the negative controls pdxJ lacking the C-terminal extension
(pdxJ(11-Ext)), wild-type pdxJ with the C-terminal extension
(wt pdxJ(+Ext)), and wild-type serB. Although the tran-
scription of a large number of genes was found to be altered
through expression of pdxJ(11), pdxJ(X3), and nanE(7), only
eight genes belonging to the histidine operon (his operon)
were found to be significantly and exclusively overexpressed
under all conditions tested (Table S8, Table S9, Figure 3B).
Thus, expression of pdxJ and nanE variants leads to selective
induction of his operon expression and, through recruitment of
promiscuous HisB and HisC, to the rescue of the ΔserB and
ΔserC strains.
Attempts To Clarify How the NanE and PdJ Variants
Induce the Expression of the Histidine Operon. In E. coli,
transcription of the his operon is stimulated by guanosine 5′-
diphosphate 3′-diphosphate (ppGpp), the effector of the
stringent response,38 under conditions of amino acid starvation
and in cells grown in minimal medium. Other amino acids
biosynthetic operons are simultaneously up-regulated through
ppGpp, while the translation apparatus is subject to a large-
scale downregulation.39 In cells expressing pdxJ(11), pdxJ(X3),
and nanE(7), we did not observe the typical gene expression
pattern of the stringent response (Table S8, Table S9, Figure
3B). Thus, we conclude that ppGpp is not involved in the
rescue mechanism.
However, this did not explain the rescue of ΔserC through
the expression of the nanE and pdxJ variants. Still, the
similarity of the enzymatic reactions of SerC and the histidinol
phosphate aminotransferase HisC (Figure 1D) let us construct
the double deletion strain ΔserCΔhisC. And, as was the case
with ΔserBΔhisB, the selected nanE and pdxJ variants could
not confer growth to this strain (Table 1). Furthermore, we
could confirm that overexpression of hisC can indeed rescue
the ΔserC strain and that the combined overexpression of both
hisB and hisC rescues the double deletion strain ΔserBΔserC
(Table 1). These results strongly indicate that HisC is
promiscuous for the SerC reaction, just as HisB has
promiscuous SerB activity.2,34 This promiscuity of HisC was
probably missed in the previous large-scale complementation
analysis2 due to a deletion in the coding sequence of hisC in
the ASKA vector pCA24N-hisC, which we detected by
sequencing.
The NanE and PdxJ Variants Induce the Expression of
the Histidine Operon. We were interested in uncovering the
mechanism by which the selected variants induce the
overproduction of the endogenous E. coli HisB and HisC
proteins. In a first step to solve this enigma, we performed RT-
qPCR to quantify possible expression changes of the hisB and
hisC genes in the presence of PdxJ variants. Due to the very
slow growth of the deletion strains in liquid minimal medium,
total RNA was isolated from the pseudo-wild-type strain
BW25113 (the parent strain of the deletion strains used in this
study), expressing either pdxJ(11) or pdxJ(X3). Expression of
either pdxJ variant led to 7 to 16-fold increased hisB and hisC
mRNA levels, whereas other amino acid biosynthetic genes
were not affected (Figure 3A).
Furthermore, transcription of the his operon is regulated by
attenuation.40 Attenuation is based on the finding that an
excess of the amino acid L-histidine leads to the production of
the leader peptide HisI, which results in an mRNA secondary
structure that functions as a terminator signal for the RNA
polymerase. If L-histidine is limited, the ribosome stalls at the
histidine codons of HisI, leading to the formation of an
alternative secondary structure (the “antiterminator”) and
subsequent transcription of the whole operon. We tested
To analyze alterations of the transcriptome in the presence
of PdxJ and NanE variants in an unbiased manner, we
performed both DNA microarray and RNAseq analysis.
H
DOI: 10.1021/acs.biochem.9b00579
Biochemistry XXXX, XXX, XXX−XXX

Biochemistry
Article
whether the presence of an excess concentration of L-histidine
affects the transcription of the his operon when pdxJ and nanE
variants are overexpressed. We found that pdxJ(11), pdxJ(X3),
and nanE(7) could no longer rescue the ΔserB and ΔserC
strains when L-histidine is added to the growth medium. RT-
qPCR verified these results: In the presence of L-histidine, the
genes hisB and hisC of the his operon are down-regulated even
under overexpression of pdxJ(11) (Figure 3A) or pdxJ(X3)
(Figure S4). Thus, the endogenous attenuation mechanism is
probably unaffected by the pdxJ and nanE variants. However,
we cannot rule out that pdxJ(11) and pdxJ(X3) relieve
background attenuation with low efficiency, and that such a
weak effect can be overridden by the attenuation that occurs in
the presence of L-histidine.
In order to identify potential interaction partners of
PdxJ(X3) that might mediate up-regulation of the his operon,
protein complex immunoprecipitation (Co-IP) was performed;
wild-type pdxJ(+Ext), which lacks the M224L exchange, was
used as a negative control. An N-terminal Flag-tag was
attached to both proteins, which were then expressed in
BW25113 cells. Following cell lysis, Co-IP was performed with
the help of anti-Flag-tag agarose beads. Whole protein extracts
(input), as well as the supernatants of two subsequent elution
steps, were analyzed by SDS-PAGE. Identified bands were cut
out, digested with trypsin and analyzed via mass spectrometry.
The input and both elution steps contained main bands
corresponding to PdxJ(X3) and wild-type pdxJ(+Ext), which
shows that both proteins were overexpressed and successfully
bound to and eluted from the anti-Flag-tag agarose beads. Two
additional minor bands corresponded to the heavy and light
chains of the antibody coupled to the beads. However, no
interaction partner of PdxJ(X3) could be detected.
(DAP). We expected that, if one of the identified metabolites
had a regulatory function, the rescue of ΔserB would be
impaired by its addition (analogous to the addition of L-
histidine to the growth medium). Unfortunately, none of the
compounds had such an effect. ΔserB cells expressing pdxJ(11)
grow normally on minimal medium even when all seven
compounds are combined in high concentrations (12 μg/mL
or 120 μg/mL).
DISCUSSION
■
By generating a large library of diversified (βα)8-barrel
enzymes, we sought to select for novel enzymatic functions
through auxotrophy complementation. We were able to
identify variants in our library that allowed the growth of the
deletion strains ΔserB and ΔserC. However, in contrast to our
initial expectations, these variants do not substitute for the
missing SerB or SerC by catalyzing the same enzymatic
reaction. Obviously, although the (βα)8-barrel fold is thought
to be very versatile, a pool of only 52 enzymes containing a few
mutations each is still a very limited set and possibly too small
for the evolution of a novel catalytic activity or substrate
specificity. Another reason for our lack of success might be a
too-small number of introduced mutations. On average, each
member of our mixed library contained about four amino acid
exchanges. Assuming that the most promising substitutions are
located in active site loops, which comprise no more than
about 25% of all residues of the (βα)8-barrel scaffold, a single
exchange would have had to suffice for establishing a novel
catalytic activity. Instead of identifying such a new enzymatic
activity, we found that the selected PdxJ and NanE variants
lead to the recruitment of HisB and HisC.
It has been shown previously that overexpression of hisB is
sufficient to rescue the serB deletion2 and that HisB has
promiscuous SerB activity.9 SerB and HisB are both members
of the HAD-like hydrolase superfamily, SCOP classification
c.67.1,41 that share the same protein fold and catalyze similar
reactions (EC 3.1.3), the dephosphorylation of phosphoserine
(Figure 1C) and the dephosphorylation of histidinol
phosphate (Figure 1D), respectively. However, in this work,
we could show that the relationship between the serine and
histidine biosynthetic pathways extends to SerC and HisC, as
well. Analogous to SerB and HisB, SerC and HisC are
structurally and functionally related: Both enzymes belong to
the same SCOP41 superfamily c.67.1 and both function as
PLP-dependent aminotransferases (EC 2.6.1) that catalyze the
transfer of an amino group from donor L-glutamate to their
respective substrates (Figures 1C,D). Overexpression of hisC is
sufficient to rescue the serC deletion, illustrating that HisC
retains a promiscuous activity for the SerC reaction. Our
results confirm the common evolutionary origin of histidine
and serine biosynthetic genes as postulated early on.42
PdxJ and nanE variants described in this study rescue the
deletion of SerB or SerC through exclusive up-regulation of
histidine operon expression. However, the mechanism behind
this up-regulation remains unclear. While we have ruled out
any involvement of the stringent response effector ppGpp, it is
principally possible that PdxJ and NanE variants bind directly
to the regulatory region of the his operon DNA or mRNA.
However, this seems unlikely as the acquired mutations (both
in the gene sequence and the C-terminal extension) do not add
positive charges that might be expected for a novel nucleic acid
binding site. Furthermore, our variants are not the only
examples of novel proteins that can rescue ΔserB through up-
We hypothesized that small molecules from the E. coli
metabolome could be recruited by the selected variants and
up-regulate the transcription of the his operon. In order to
identify such small-molecule regulators, we compared the total
metabolome of BW25113 cells expressing either pdxJ(11) or
the negative control pdxJ(11-Ext), lacking the C-terminal
extension, by a global NMR based metabolic screen. We
analyzed both cellular metabolites and the culture supernatant
of cells grown in minimal medium in the exponential growth
phase. In each case, three biological replicates were
investigated per group. Following metabolite identification
and quantification, discriminating metabolites were identified
by a two-sided heteroscedastic Student’s t test. While we did
not identify metabolites that are more abundant in cells
expressing pdxJ(11), we found 5 small molecules (alanine,
threonine, lysine, maltotriose, and the intermediate of lysine
biosynthesis N-succinyl-L,L-2,6-diaminopimelate (SDAP)) that
are significantly over-represented in whole cells expressing the
control pdxJ(11-Ext), and two metabolites (propionate and
acetate) that are significantly over-represented in the culture
supernatant of the same cells (Table S10). We speculated that
one (or more) of the compounds that accumulated in cells
expressing pdxJ(11-Ext) might function as negative regulators
for his operon transcription. Expression of pdxJ(11) would lead
to a drop in their concentration and hence indirectly up-
regulate the his operon. To test this hypothesis, we added the
identified compounds to the growth medium of ΔserB cells
expressing pdxJ(11) (to a concentration of 12 μg/mL or 120
μg/mL). In the case of SDAP, we had to substitute the
originally identified substance with the structurally related and
commercially available desuccinylated 2,6-diaminopimelic acid
I
DOI: 10.1021/acs.biochem.9b00579
Biochemistry XXXX, XXX, XXX−XXX

Biochemistry
Article
regulation of his operon mRNA: The 4-helix bundle SynSerB3,
that shares no sequence or structural identity with our variants
nor with the C-terminal extension, and adopts a completely
different protein fold, analogously stimulates the expression of
the his operon.36 Therefore, it seems that an unknown and
specific activator or repressor of the his operon exists in the E.
coli proteome and that this potential activator or repressor
might be the target of both PdxJ and NanE variants and
SynSerB3. However, our Co-IP experiments performed with
PdxJ(X3) failed to detect such a hypothetical interaction
partner.
Lab-made, small proteins have been shown to provide a life-
sustaining function by changing gene expression patterns in E.
coli,36,37 analogous to the results presented in this study.
Whereas these designed proteins are 4-helix bundles with no
sequence similarity to naturally occurring proteins, the variants
presented in this study belong to the (βα)8-barrel fold. Natural
(βα)8-barrels are enzymes belonging to five of the six enzyme
classes, and the evolvability of this fold and its ability to acquire
new enzymatic functions has been illustrated repeatedly.43 To
our knowledge, the variants described in this study are the first
examples of (βα)8-barrel enzymes that function not as enzymes
but have a regulatory function and lead to altered gene
expression patterns in E. coli. As such, they provide a new
example of the adaptability of the (βα)8-barrel fold.
ACKNOWLEDGMENTS
■
̈
The authors thank Joseph Sperl and Elisabeth Worle for the E.
coli deletion strains ΔdeoC and ΔserBΔgph and Marc Creus for
suggesting using DAP as a substitute for commercially
unavailable SDAP. We are grateful to Ingrid Araya and
Kevin Heizler for their advice on the Co-IP experiments.
REFERENCES
■
(1) Khersonsky, O., and Tawfik, D. S. (2010) Enzyme promiscuity: a
mechanistic and evolutionary perspective. Annu. Rev. Biochem. 79,
471−505.
(2) Patrick, W. M., Quandt, E. M., Swartzlander, D. B., and
Matsumura, I. (2007) Multicopy suppression underpins metabolic
evolvability. Mol. Biol. Evol. 24, 2716−2722.
(3) Kitagawa, M., Ara, T., Arifuzzaman, M., Ioka-Nakamichi, T.,
Inamoto, E., Toyonaga, H., and Mori, H. (2006) Complete set of
ORF clones of Escherichia coli ASKA library (a complete set of E. coli
K-12 ORF archive): unique resources for biological research. DNA
Res. 12, 291−299.
(4) Oberhardt, M. A., Zarecki, R., Reshef, L., Xia, F., Duran-Frigola,
M., Schreiber, R., Henry, C. S., Ben-Tal, N., Dwyer, D. J., Gophna, U.,
and Ruppin, E. (2016) Systems-Wide Prediction of Enzyme
Promiscuity Reveals a New Underground Alternative Route for
Pyridoxal 5′-Phosphate Production in E. coli. PLoS Comput. Biol. 12,
e1004705.
̈
(5) Sterner, R., and Hocker, B. (2005) Catalytic versatility, stability,
and evolution of the (βα)8-barrel enzyme fold. Chem. Rev. 105, 4038−
4055.
ASSOCIATED CONTENT
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.bio-
chem.9b00579.
■
S
(6) List, F., Sterner, R., and Wilmanns, M. (2011) Related (βα)8-
barrel proteins in histidine and tryptophan biosynthesis: a paradigm to
study enzyme evolution. ChemBioChem 12, 1487−1494.
(7) Obexer, R., Godina, A., Garrabou, X., Mittl, P. R., Baker, D.,
Griffiths, A. D., and Hilvert, D. (2017) Emergence of a catalytic tetrad
during evolution of a highly active artificial aldolase. Nat. Chem. 9,
50−56.
(8) Nobeli, I., Ponstingl, H., Krissinel, E. B., and Thornton, J. M.
(2003) A structure-based anatomy of the E. coli metabolome. J. Mol.
Biol. 334, 697−719.
(9) Yip, S. H., and Matsumura, I. (2013) Substrate ambiguous
enzymes within the Escherichia coli proteome offer different
evolutionary solutions to the same problem. Mol. Biol. Evol. 30,
2001−2012.
(10) Baba, T., Ara, T., Hasegawa, M., Takai, Y., Okumura, Y., Baba,
M., Datsenko, K. A., Tomita, M., Wanner, B. L., and Mori, H. (2006)
Construction of Escherichia coli K-12 in-frame, single-gene knockout
mutants: the Keio collection. Mol. Syst. Biol. 2, 2006.0008.
(11) Datsenko, K. A., and Wanner, B. L. (2000) One-step
inactivation of chromosomal genes in Escherichia coli K-12 using
PCR products. Proc. Natl. Acad. Sci. U. S. A. 97, 6640−6645.
(12) Rohweder, B., Semmelmann, F., Endres, C., and Sterner, R.
(2018) Standardized cloning vectors for protein production and
generation of large gene libraries in Escherichia coli. BioTechniques 64,
24−26.
(13) Pace, C. N., Vajdos, F., Fee, L., Grimsley, G., and Gray, T.
(1995) How to measure and predict the molar absorption coefficient
of a protein. Protein Sci. 4, 2411−2423.
(14) Geladopoulos, T. P., Sotiroudis, T. G., and Evangelopoulos, A.
E. (1991) A malachite green colorimetric assay for protein
phosphatase activity. Anal. Biochem. 192, 112−116.
E. coli deletions strains and growth media used for
selection; list of oligonucleotides used in this study; 52
phosphate-binding (βα)8-barrel genes pooled for the
generation of the diversified library; list of expression
vectors used in this study; crystal structure of PdxJ(11),
data collection and refinement statistics; variants of nanE
and pdxJ selected by in vivo complementation; effect of
the modification of the C-terminal extension of pdxJ
variants for in vivo complementation of ΔserB; DNA
microarray analysis to identify genes exhibiting more
than a 2-fold change of mRNA levels; RNAseq to
identify genes exhibiting more than a 2-fold change of
mRNA levels; NMR analysis of metabolites in whole
cells and culture supernatants; sequence of the ten
amino acid extension of pdxJ variants; growth of deletion
strains on M9 minimal medium transformed with
different constructs; overlay of variant PdxJ(11) with
wild-type PdxJ; qRT-PCR identifies changes in relative
mRNA levels of the four amino acid biosynthesis genes
hisB, hisC, trpCF, and proA upon expression of pdxJ(X3)
(PDF)
AUTHOR INFORMATION
Corresponding Author
*E-mail: reinhard.sterner@ur.de. Phone: +49-941-943-3015.
ORCID
Michael Bott: 0000-0002-4701-8254
Reinhard Sterner: 0000-0001-8177-8460
Notes
■
(15) Linde, M., Heyn, K., Merkl, R., Sterner, R., and Babinger, P.
(2018) Hexamerization of geranylgeranylglyceryl phosphate synthase
ensures structural integrity and catalytic activity at high temperatures.
Biochemistry 57, 2335−2348.
(16) Kabsch, W. (1993) Automatic processing of rotation diffraction
data from crystals of initially unknown symmetry and cell constants. J.
Appl. Crystallogr. 26, 795−800.
The authors declare no competing financial interest.
J
DOI: 10.1021/acs.biochem.9b00579
Biochemistry XXXX, XXX, XXX−XXX

Biochemistry
Article
(17) McCoy, A. J., Grosse-Kunstleve, R. W., Adams, P. D., Winn, M.
D., Storoni, L. C., and Read, R. J. (2007) Phaser crystallographic
software. J. Appl. Crystallogr. 40, 658−674.
(35) Wang, W., Cho, H. S., Kim, R., Jancarik, J., Yokota, H., Nguyen,
H. H., Grigoriev, I. V., Wemmer, D. E., and Kim, S. H. (2002)
Structural characterization of the reaction pathway in phosphoserine
phosphatase: crystallographic “snapshots” of intermediate states. J.
Mol. Biol. 319, 421−431.
(18) Potterton, E., McNicholas, S., Krissinel, E., Cowtan, K., and
Noble, M. (2002) The CCP4 molecular-graphics project. Acta
Crystallogr., Sect. D: Biol. Crystallogr. 58, 1955−1957.
(36) Digianantonio, K. M., and Hecht, M. H. (2016) A protein
constructed de novo enables cell growth by altering gene regulation.
Proc. Natl. Acad. Sci. U. S. A. 113, 2400−2405.
(19) Murshudov, G. N., Vagin, A. A., and Dodson, E. J. (1997)
Refinement of macromolecular structures by the maximum-likelihood
method. Acta Crystallogr., Sect. D: Biol. Crystallogr. 53, 240−255.
(20) Adams, P. D., Grosse-Kunstleve, R. W., Hung, L. W., Ioerger, T.
R., McCoy, A. J., Moriarty, N. W., Read, R. J., Sacchettini, J. C.,
Sauter, N. K., and Terwilliger, T. C. (2002) PHENIX: building new
software for automated crystallographic structure determination. Acta
Crystallogr., Sect. D: Biol. Crystallogr. 58, 1948−1954.
(37) Digianantonio, K. M., Korolev, M., and Hecht, M. H. (2017) A
non-natural protein rescues cells deleted for a key enzyme in central
metabolism. ACS Synth. Biol. 6, 694−700.
(38) Artz, S. W., and Holzschu, D. (1983) Histidine Biosynthesis
and Its Regulation. In Amino Acids: Biosynthesis and Genetic Regulation
(Somerville, K. M. H. a. R. L., Ed.), pp 379−404, Addison-Wesley
Publishing Co, Reading, MA.
(39) Cashel, M., and Rudd, K. E. (1987) The Stringent Response. In
Escherichia Coli and Salmonella Typhimurium: Cellular and Molecular
Biology (Neidhardt, F. C., Ingraham, J. L., Low, K. B., Magasanik, B.,
Schaechter, M., and Umbarger, H. E., Eds.), pp 1410−1438, American
Society for Microbiology, Washington, D.C.
(40) Kolter, R., and Yanofsky, C. (1982) Attenuation in amino acid
biosynthetic operons. Annu. Rev. Genet. 16, 113−134.
(41) Lo Conte, L., Brenner, S. E., Hubbard, T. J., Chothia, C., and
Murzin, A. G. (2002) SCOP database in 2002: refinements
accommodate structural genomics. Nucleic Acids Res. 30, 264−267.
(42) Jensen, R. A. (1976) Enzyme recruitment in evolution of new
function. Annu. Rev. Microbiol. 30, 409−425.
(43) Gerlt, J. A., and Raushel, F. M. (2003) Evolution of function in
(βα)8-barrel enzymes. Curr. Opin. Chem. Biol. 7, 252−264.
(21) Emsley, P., and Cowtan, K. (2004) Coot: model-building tools
for molecular graphics. Acta Crystallogr., Sect. D: Biol. Crystallogr. 60,
2126−2132.
(22) Davis, I. W., Leaver-Fay, A., Chen, V. B., Block, J. N., Kapral, G.
J., Wang, X., Murray, L. W., Arendall, W. B., 3rd, Snoeyink, J.,
Richardson, J. S., and Richardson, D. C. (2007) MolProbity: all-atom
contacts and structure validation for proteins and nucleic acids.
Nucleic Acids Res. 35, W375−383.
(23) Langmead, B., and Salzberg, S. L. (2012) Fast gapped-read
alignment with Bowtie 2. Nat. Methods 9, 357−359.
(24) Li, H., Handsaker, B., Wysoker, A., Fennell, T., Ruan, J.,
Homer, N., Marth, G., Abecasis, G., and Durbin, R. (2009) The
Sequence Alignment/Map format and SAMtools. Bioinformatics 25,
2078−2079.
(25) Anders, S., Pyl, P. T., and Huber, W. (2015) HTSeq–a Python
framework to work with high-throughput sequencing data. Bio-
informatics 31, 166−169.
(26) Huber, W., Carey, V. J., Gentleman, R., Anders, S., Carlson, M.,
Carvalho, B. S., Bravo, H. C., Davis, S., Gatto, L., Girke, T., Gottardo,
R., Hahne, F., Hansen, K. D., Irizarry, R. A., Lawrence, M., Love, M. I.,
MacDonald, J., Obenchain, V., Oles, A. K., Pages, H., Reyes, A.,
Shannon, P., Smyth, G. K., Tenenbaum, D., Waldron, L., and Morgan,
M. (2015) Orchestrating high-throughput genomic analysis with
Bioconductor. Nat. Methods 12, 115−121.
(27) Love, M. I., Huber, W., and Anders, S. (2014) Moderated
estimation of fold change and dispersion for RNA-seq data with
DESeq2. Genome Biol. 15, 550.
(28) Kranz, A., Steinmann, A., Degner, U., Mengus-Kaya, A.,
Matamouros, S., Bott, M., and Polen, T. (2018) Global mRNA decay
and 23S rRNA fragmentation in Gluconobacter oxydans 621H. BMC
Genomics 19, 753.
(29) Wallmeier, J., Samol, C., Ellmann, L., Zacharias, H. U., Vogl, F.
C., Garcia, M., Dettmer, K., Oefner, P. J., and Gronwald, W. (2017)
Quantification of Metabolites by NMR Spectroscopy in the Presence
of Protein. J. Proteome Res. 16, 1784−1796.
(30) Gronwald, W., Klein, M. S., Kaspar, H., Fagerer, S. R.,
Nurnberger, N., Dettmer, K., Bertsch, T., and Oefner, P. J. (2008)
Urinary metabolite quantification employing 2D NMR spectroscopy.
Anal. Chem. 80, 9288−9297.
(31) Xia, J., Bjorndahl, T. C., Tang, P., and Wishart, D. S. (2008)
MetaboMiner–semi-automated identification of metabolites from 2D
NMR spectra of complex biofluids. BMC Bioinf. 9, 507.
(32) Benjamini, Y., and Hochberg, Y. (1995) Controlling the False
Discovery Rate-a Practical and Powerful Approach to Multiple
Testing. J. R. Stat. Soc. B-Stat. Meth. 57, 289−300.
(33) Edgar, R., Domrachev, M., and Lash, A. E. (2002) Gene
Expression Omnibus: NCBI gene expression and hybridization array
data repository. Nucleic Acids Res. 30, 207−210.
(34) Kuznetsova, E., Proudfoot, M., Gonzalez, C. F., Brown, G.,
Omelchenko, M. V., Borozan, I., Carmel, L., Wolf, Y. I., Mori, H.,
Savchenko, A. V., Arrowsmith, C. H., Koonin, E. V., Edwards, A. M.,
and Yakunin, A. F. (2006) Genome-wide analysis of substrate
specificities of the Escherichia coli haloacid dehalogenase-like
phosphatase family. J. Biol. Chem. 281, 36149−36161.
K
DOI: 10.1021/acs.biochem.9b00579
Biochemistry XXXX, XXX, XXX−XXX