Scientific paper
Gene Expression Profiling of Recombinant Protein Producing E. coli at Suboptimal Growth Temperature
I	V	-	^ V	i	^
Mitja Mahni~,1* [pela Baebler,3 Andrej Blejec,3 [pela Jalen,4 Kristina Gruden,3
Viktor Menart and Simona Jev{evar2
1 Development center Slovenia, Lek Pharmaceuticals d.d., a Sandoz company, Verovskova 57, 1526 Ljubljana, Slovenia 2 Sandoz Biopharmaceuticals Menges, Lek Pharmaceuticals d.d., a Sandoz company, Kolodvorska 27, 1234 Menges, Slovenia 3 Department of Biotechnology and Systems Biology, National Institute of Biology, Vecna pot 111, 1000 Ljubljana, Slovenia 4 Laboratory for Biosynthesis and Biotransformation, National Institute of Chemistry, Hajdrihova 19, 1000 Ljubljana, Slovenia
* Corresponding author: E-mail: mitja.mahnic@sandoz.com;
Phone: + 386 1 5803275
Received: 07-03-2011 Dedicated to inspiring mentor dr. Viktor Menart.
Abstract
Recent studies have revealed that at lower cultivation temperatures (25 °C) much higher percentage of correctly folded recombinant hG-CSF protein can be extracted from inclusion bodies. Hence, the goal of our research was to investigate mechanisms determining characteristics of non-classical inclusion bodies production using gene expression profiling, focusing on proteases and chaperones gene expression. Statistical analysis of microarray data showed prominent changes in energy metabolism, in metabolism of amino acids and nucleotides, as well as in biosynthesis of cofactors and secondary metabolites if the culture was grown below its optimal temperature. Moreover, 24 differentially expressed up to now known genes classified among proteases, chaperones and other heat or stress related genes. Among chaperones UspE and among proteases YaeL and YeaZ might play an important role in accumulation of correctly folded recombinant proteins. Membrane localized protease yaeL gene was found to have higher activity at 25 °C and is thus potentially functionally related to the more efficient recombinant protein production at lower temperatures. The results of this study represent advance in the understanding of recombinant protein production in E. coli. Genes potentially influencing production of recombinant protein at lower growth temperature represent basis for further research towards improvement of E. coli production strains as well as fermentation process.
Keywords: Recombinant protein production, non-classical inclusion bodies, E. coli, expression microarrays, YaeL protease, GroEL chaperone
1. Introduction
The „art nouveau" of the modern biopharmaceuti-cal industry is getting towards understanding of the mechanisms underlying recombinant protein production in different organisms. The obtained knowledge should further contribute to improvements of fermentation processes and thus gain economical advantages for the industry.
Escherichia coli is the most widely used recombinant protein producing organism due to its ease of cultivation and fast production rate. Protein misfolding is a com-
mon event during bacterial over-expression of recombinant genes.1 Incorrectly folded or misfolded proteins can appear as a result of cell exposure to the environmental stress, such as elevated temperatures and over-expression of recombinant genes. The resulting misfolded proteins may be degraded by proteases, fold by chaperones, or aggregated and sequestered as inclusion bodies (IBs).2 Hence, a common limitation of recombinant protein production in bacteria is the formation of insoluble protein aggregates known as IBs.3 It has been believed for a long time that IB proteins are biologically inactive and therefore
undesired in bioprocesses.4 The potential of chaperones in assisting folding of misfolded proteins has been investigated from several aspects.5 On the other hand, it has already been reported that functional proteins could be easily extracted from IBs using non denaturing mild detergents and polar solvents provided that cultures were grown at lower temperatures.3,4,6-9 Such IBs, termed „non-classical" inclusion bodies (ncIBs) by Jevševar et al. (2005),7 are defined by containing large amount of correctly folded protein precursor produced in E. coli at lower temperature (around 25 °C). Compared to classical IBs they are characterized by higher fragility and solubility, irreversible contraction at acidic pH and most importantly, by a high amount of correctly folded target protein or its precursor.6
One of the most important recombinant proteins in the field of modern oncology is human granulocyte colony stimulating factor (hG-CSF) protein. Due to its regulatory role in the growth, differentiation, survival, and activation of neutrophils and their precursors, hG-CSF is central to neutrophil-based immune defenses1. Four types of hG-CSF are clinically available: a glycosylated form (lenograstim) produced in CHO cells, an N-terminal replaced nonglycosylated form of granulocyte colony-stimulating factor (nartograstim),10 and nonglycosylated form (filgrastim), both produced by using the expression in E. coli.1 In addition to aforementioned forms long acting form of filgrastim - PEGfilgrastim, a modified PEG-ylated filgrastim enabling less frequent administration has been available since 2002.
As shown before7 cultivation temperature was the most important variable affecting properties of hG-CSF IBs and thus its efficient production. Therefore the goal of our research was to investigate mechanisms determining characteristics of ncIBs production by comparing physiology of recombinant E. coli [BL21 (DE3)] at three different temperatures (T = 25 °C (suboptimal), 37 °C (optimal) and 42 °C (heat shock), respectively) using gene expression profiling approach. As formation of various proteases and chaperones under different temperature conditions was previously reported6-8,11 we have inspected behavior of these genes in more detail.
2. Experimental
2. 1. Cultures and Plasmids
In this study the recombinant E. coli strain BL21 (DE3) (Novagen), carrying expression plasmid pET3a without hG-CSF insert (control strain) or with hG-CSF insert [Fopt5] (production strain) was used. hG-CSF insert ([Fopt5]) was prepared as described in7.
2. 2. Culture Conditions
Bacterial inoculum of the production and control strain was prepared in a shake flask culture and grown
overnight at 25 °C and at 160 rpm in the LBPG/amp100 medium7. After reaching optical density of OD600nm ~ 4 the inoculum was transferred to the GYSP medium and immediately induced with IPTG. The cultures were then incubated in shake flasks at 160 rpm and at three different temperatures (i.e. T = 25 °C, 37 °C and 42 °C, respectively) until the appropriate culture's optical density (OD), indicating the transition from the exponent to the stationary phase was reached (OD600nm ~ 10 for the culture grown at 25 °C and OD600 nm ~ 4 for the cultures grown at 37 °C or 42 °C).7 At that point cultures were stabilized with RNA protect Bacteria Reagent (Qiagen), aliquoted, centrifuged and the bacterial pellet was stored at - 80 °C for further RNA and protein expression analysis. The cultivation experiment was repeated 3 times, thus yielding 18 samples altogether.
2. 3. Isolation of Total RNA and DNase Treatment
RNA isolation and DNase treatment was performed as described by Petek et al. (2010),12 except for substituting lysostaphin with lysocyme (500 mg/ml) in the cell lysis step. RNA quality, quantity and integrity were checked by NanoDrop (NanoDrop Technologies, USA), gel elec-trophoresis and Bioanalyzer (Agilent Technologies).
2. 4. Microarray Hybridization
Purified RNA (approximately 30 pg) was used for the cDNA synthesis and direct labeling (Superscript II, In-vitrogen). Luciferase control mRNA (1 ng/pg; Promega) and 3 pg of random primers were added to each RNA sample. This was followed by 10 minutes of incubation at 70 °C and immediate chilling on ice. cDNA synthesis was carried out using SuperScript II reverse transcriptase (In-vitrogen) according to manufacturer's instructions. Synthesized cDNA was purified using MinElute PCR purification Kit (Qiagen). The concentration of cDNA, efficiency of dye (Cy5) integration and integrity of labeled cDNA were checked by NanoDrop and gel electrophore-sis. Pre-designed oligonucleotide microarrays (Custom-Array™ 12K Microarray, Combimatrix Corporation) containing 12,000 features arrays of complete E. coli genome (4,200 genes, positive and negative control sequences) were used. Labeled cDNA was hybridized to the arrays according to the protocol recommended by CombiMatrix except for using 2X formamide based hybridization buffer (Genisphere) containing Salmon testis DNA (1 pg / pl, Sigma) and shorter hybridization time (1h).
2. 5. Microarray Imaging and Data Analysis
After hybridization semiconductor microarray surfaces were covered by imaging solution and were scanned using a fluorescence LS200 scanner (TECAN).13 Combi-
matrix Microarray Images Software was used for image analysis and quality control. Further data analysis was pe-formed in R software environment for statistical computing and graphics (http://www.r-project.org/). Bioconduc-tor's packages affy, limma and KEGGsoap were used for quality control, preprocessing, statistical significance testing14 and annotation. The data was normalized using the quantile normalization. Intensities of the factory-built in control probes were compared to information on Combi-matrix FAQ internet site to confirm the validity of the chosen preprocessing approach. Differentially expressed transcripts were functionally analysed according to Gene Ontology (GO) using GSEA15. Normalized data was further analyzed for significance using the linear models with different contrast settings and empirical Bayes (p<0.05). To minimize the possibility of false positive results, all log2 values of gene expression ratios between -0.5 and 0.5 were considered not relevant and were excluded from the data interpretation16. The final results were expressed as log2 values (logFC) of the ratios between the mean expression in sample groups. Moreover, the dataset was defined by GeneID identification tags for easier access to different knowledge databases. Differentially expressed genes (DE) were visualized in the EcoCyc database (http://www.ecocyc.org/expression.html) that allows representation of the obtained results on the metabolic and signaling pathways.
The microarray data have been deposited in NCBI's Gene expression Omnibus and are accessible through GEO Series accession number GSE25561 (http://www. ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE25561).
2. 6. Real-Time PCR
Six DE genes yeaZ, ydcP, groEL, ecpD, torD and uspD (yiiT) were selected for real-time PCR analysis based on TaqMan® MGB™ technology17. 16S rRNA was used as the reference gene. Gene specific sequences were chosen for assay design using NCBI BLAST (http://blast. ncbi.nlm.nih.gov/Blast.cgi). The assays were designed by Applied Biosystems, primer and probe sequences are listed in Table 1. Total RNA (approximately 3 ^g) was reverse transcribed using High Capacity cDNA Reverse Transcription Kit (Applied Biosystems) according to manufactures instructions.
Real-time PCR reaction were set up as described in Petek et al. (2010)12 in ABI PRISM 7900 HT Fast Sequence Detection System (Applied Biosystems) using 5 |jl reactions and standard cycling parameters. Data quality control and analysis was performed as described in Petek et al. (2010).12 For the purpose of comparison of qPCR and microarray data, the results were expressed as the log2 of the ratio between relative gene expressions at two different temperatures for the control and production strain and as the ratio between relative gene expressions in control and production strain grown at the same temperature. Statistical significance of differences in gene expression was calculated using the same model as in analysis of microar-ray data.13
2. 7. Western Blotting
SDS-PAGE was performed using a 4-12% Nu-PAGER Novex Bis-Tris gel (Invitrogen) according to manufacturer's protocols. Prior to electrophoresis all samples were resuspended in 10 mM TRIS/HCl (pH = 8.0) buffer and diluted according to their final OD to obtain similar sample loads. Samples were further treated by addition of NuPAGER LDS sample buffer, denatured for 10 minutes at T = 70 °C and applied to gel. Electrophoresis was performed at 200 V, 125 mA for 40 minutes at room temperature. Proteins that were separated with SDS-PAGE were afterwards transferred onto the Nitrocellulose membrane by using iBlotTM Dry blotting System (Invitrogen). Immunodetection was made by primary antibody (GroEL antibody mouse Monoclonal, IgG1, Antibodies-online GmbH) followed by secondary antibodies (Anti-mouse IgG - HRP, Sigma). Colorime-tric detection was achieved with addition of detection solution (mixture of solution A (15 mg of 4-chloro-1-naftol dissolved in 5 ml of Methanol) and solution B (15 ml of H2O2 added to 25 ml of TBS (pH = 7.5). At the end the membrane was imaged and obtained images were further analyzed by ImageJ program (Image Processing and Analysis in Java, http://rsbweb.nih.gov/ij/) used for optical density measurements. We semi-quantified all proteins from electrophoresis gel images and GroEL from western blotting membranes images and determine the relative GroEL content in different samples.
Table 1: Primer and probe sequences used for real-time PCR
Gene	Forward primer	Reverse primer	Probe
yeaZ	CGCTGATATTCGGCCCAGTAAA	GTGCTGGCAGCCATTGAC	CTTCGCCCATTCGCG
ydcP	GATATTGGCGCGTTCGATTCG	GAGATGATCTTTCGCCACTTTCAAT	CAGGCCGATAAATTT
groEL	GCAACTCTGGTTGTTAACACCAT	TGCAGCATAGCTTTACGACGAT	AACCGCAGCGACTTT
ecpD	GAACACGCTCTCTCTGTCTTTAGG	CAAACGTGGGCAAACAATCAAATT	ACAGCCAGCACCTCAC
torD	ACAGGACGAGCAAGAGATTAAACG	CGTTGAAATTGCCGCTGGTTT	CCCTGCCTCAACTAAC
uspD	CGGATAACAAGCTGTATAAACTGACGAA	GCATTTCTCCGCGTTCAATACG	TCGGCCATTGAATATT
16S rRNA	GGAGTACGGCCGCAAGGT	CATGCTCCACCGCTTGTG	AAAACTCAAATGAATTGACG
3. Results
It is known that cultivation at T = 25 °C yields high amounts of correctly folded hG-CSF protein within the IBs7. 98% of the expressed hG-CSF was in the form of IBs and only 2% was produced in the cytoplasm.8 In these conditions, recombinant hG-CSF protein accumulated to yield 35% to 40% of the total proteins of E. coli, almost 50% of hG-CSF extracted from the IBs showed biological activity.7,8 In order to investigate the underlying mechanisms of this phenomenon the experimental design was set to profile E. coli gene expression at different cultivation temperatures. Therefore, control (recombinant strain carrying empty expression plasmid) and production strain (recombinant strain carrying expression plasmid containing hG-CSF) were incubated at three different temperatures, 25 °C, 37 °C and 42 °C, respectively. The cultures were sampled just before transition into stationary state, when the cells were still fully viable and already under stress of recombinant protein production. At this point maximal plasmid copy number was determined as well as maximal productivity was achieved - accumulation level of hG-CSF reached plateau.7,18 After initial data analysis special attention was given to the temperature dependent changes in expression levels of proteases, chaperones and other heat or stress related genes.
3. 1. Overview of Identified Gene Expression Differences
Linear models were set to identify DE genes in comparisons of the following group pairs: control strain cultures grown at 37 °C compared to 25 °C (C37_25), control strain cultures grown at 42 °C compared to 25 °C (C42_25), production strain cultures grown at 37 °C compared to 25 °C (P37_25), production strain cultures grown at 42 °C compared to 25 °C (P42_25) as well as comparison of control and production strain cultures grown at 25 °C (P_C25), 37 °C (P_C37), 42 °C (P_C42). 282 DE genes showed statistically significant differences in expression levels if comparing control strain cultures grown at T = 37 °C and T = 25 °C and 226 DE genes if comparing control strain cultures grown at T = 42 °C and T = 25 °C. In production strain 28 and 34 DE genes showed changes in expression levels at T = 37 °C and T = 42 °C compared to T = 25 °C, respectively. Comparison in gene expression profiles of control and production strain grown at the same cultivation temperature identified on average 113 DE genes; i.e. from 74 to 161 DE genes at all studied temperatures. Additionally, two-way analysis of variance was performed to test for the effects of recombinant protein production (P_C) and growth temperature as well as interaction of both factors (Supporting information 1). The percentage of DE genes identified was similarly up to about 7% (i.e. up to about 305 DE genes) as in pair-wise comparisons (Fig.1). When comparing control
S3
-a -a

¿»I
"T3 <D
is
o
I
s -a
o
o <
w
M
O
t
U <N M Tt
. a ,
^ £
la	a
a	R
a	V
13	U a
i!
H a
S
M
V -
= g —
S
M
V -
Ö
T3
v
S
M V
13 £
O —
Ö
C
-D O T3
'[S .a o
T3 —< <u
•SSS
-D3 0
ZSrS rt b^
O 00
co in
(N CO P- P-CO CO in OOO OOO OOO
u
m <<
no	o
o	o
o	o
o	o
M
S	K
-a	^
» .5	"ft
o C	o
ft <	S3
9	K
o	o
^	ft
CO	«
s
S	S3
o	ö
M C ; T3 '
s '
-a
• S	"O
z	a
NO	t-
t-	00
NO	00
CO	CO
o	o
o	o
o	o
^ jo « <«
ä Ž ^ pi
O	CO
00	^
O	00
CO	o
O	rt
o	o
o	o
c o 'm
S	0
ö	*
oo	no a\	trt	in in	in
o	o
o	o
o	o
ä « O Sä 0-2T3
00 c • •a c •
c
M o
rt fH M rt
m
rs <u O &
no	00
o\	ay
O	rt
NO	00
o o o o o o
o ft
-a £
2 ^
S JŽ
F U J (N in co rt CO O "^t^rt NO NO O OOO OOO OOO
.8 o o
O rt
& U
ON tiN
o\ o o o
&	3
•■ö	-a
o	o
<3	f^
. g	3
°	s
ft	o
CO	CO
co	in
iN	-^t
in	in
o	o
o	o
S <<
I * £ £
«
"3 X ^ <u .a ~ ro T3
M
•a ft ^^
di &
CO	ON
■^t	iN
co	in ^ o
rt	co
o	o
o	o
O 00 NO co co
o NO oo OOO
C
^ M M Č ¡2
c
z
i-J & in rt oo iN rt o\ O O rt
o o o o
3
M
V -
13
I o
TT -
3
M -
Ö
3
M
V -
13 £ O
O
TT
o
ft Ö
li ¡9 is
U
Ö ö
a> co
o a
eg c
T3 S3
ftfto HHrS OOoi
• >3
a
2 ^
13 £ S3 o Q ¡3
(N	CO
t-	r-
CO	CO
o	o
o	o
o	o
-■tino
rninm
ooo ooo ooo
<<
o no
o
NOOO
ooo ooo ooo
Ü 3 •■ö T3
o o
<3 f^
. S $ ft O
CO CO
co in (N -it in in
CD CD O O
-D
S3
§ s
i3 <1
a S3
o
) u c '<3
¡fig O
c G u
So <u
£ £
CO ON O ^(NCO
oono nmj
ooo ooo

« c o		C3	u o .a &	13 C3 O		
ft		C	o	« C3		
M ro		• ii	.a ift			
u						
la ro c		13 « c	J	C3 C J3	oo 13	
o	13	o			T3	(S « M C3 T3
^	C3 i-H	^	«	o		
o (J	CD Is	o o	c o	C3 O	¡3	
					'<3	
o ^	13 m « u		u 3	'j3	ID u	ro
H		H			ft	ft
NO		o	s	IN	t-	CO
in		NO			in	CO
		1—1	J3	00	-t	IN
CD		o		-t	NO	00
O		o		o	o	o
O		o		o	o	o
O		o		o	o	o
p<	ft
CO	CO
o	o 00
O	<N
O	-t
o	o
o	o
is U
-t in -t in -t o o
eg c
T3 C
C
'<3
2 a
T3
S3 £ IN 00 o
a _ o
13
-D
C3
eg Ü S3 ta
£ ® ®
-D S3
0
g ^	^
g	13
§
s u	s
C^ MD CN CN ^
^o oo o in in ooo ooo ooo
Ö tt o Sä
0-2T3
£	®
ft	Pi
t-	CD
in	in
-t	o\
NO	NO
o	o
o	o
o	o
c
o • a
Pi >
00 <N
o o oo . N in
83 '
o o
T3 C
c '<3
u ft T3 ro T3
£ S3 £ CM 00 o
in in
o o o o
3
M
V -
13
I o
3
M -
ft Ö
3
M
V -
13 £ O
O
a Ö
c o > "-Ö u
C3 <U
u
-D
§
NO
t-	<N
in	NO
in	in
o	o
o	o
o	o
ft
"o -o
<3 1:
u ^ g
O ft U
in
c in co o -t o\
NO NO NO OOO
ooo ooo
u
g n
c
o
o
a §
b c
• ^
CD «
T3
1 .a u
-a
c
o
o «
S3
o
tZ ^ o
H rt O
eg t3
c
ao e
in oo
NO o
oo	o
a\
IN
CO	CO
o	o
o	o
o	o
in
co	o
o	o
o	o
o	o
C^	N!D
00	o
(N	CO
^	o\
o	o
o	o
o	o
ro -a
£ „ c ¡3 ■st r^ in co -t o in
■■t in
o o o o
in in
o o o o
strain cultures grown at T = 37 °C and T = 42 °C to cultures grown at T = 25 °C the ratio between up-regulated and down-regulated genes was on average 50 : 50 (35 : 31 DE genes, respectively) as it was true also for the gene expression profiles of the production strain (data not shown). When comparing control and production strain cultures grown at the same temperature the ratio between up-regulated and down-regulated genes was on average 65 : 35 in favor of down-regulated at T = 25 °C (7 : 3 DE genes, respectively) and T = 42 °C (11 : 9 DE genes, res-
P C	T37 25
91 / 48 \ 305 / 14 \/ 76 \
223
T42_25	3223
Figure 1: Venn diagram of differentially expressed genes identified by analysis of variance. P_C: strain dependent changes in gene expression, T37_25, T42_25 - temperature dependent changes in gene expression. p-value cut-off 0.05.
pectively), respectively and 55 : 45 (7 : 6 DE genes, respectively) in favor of up-regulated at T = 37 °C (Supporting information 2).
For easier biological interpretation functional analysis of gene expression changes was performed using GSEA (Table 2). To gain a better overview of the studied processes DE genes were visualized in EcoCyc Omics viewer which is based on its own functional ontology system (Supporting information 3). Prominent changes in energy metabolism, in metabolism of amino acids and nuc-leotides, as well as in biosynthesis of cofactors and secondary metabolites if the culture was grown below its optimal temperature were observed. Some DE genes are annotated as transporters. Also some signaling pathways related genes were identified as DE if comparing strains grown at different temperatures. If comparing production to control strain grown at suboptimal temperature genes identified as DE are related to carbohydrates biosynthesis, amino acids biosynthesis and in biosynthesis of cofactors, while at higher growth temperatures (T = 37 °C and T = 42 °C) more general changes in energy metabolism, metabolism of amino acids and secondary products were detected (Supporting information 2).
3. 2. Gene Expression Analysis of
Chaperones, Proteases and Other Stress Related Genes
According to EcoCyc and KEGG databases and literature survey,1,19,20 131 of E. coli's genes are coding for molecular chaperones, proteases, sigma factors and other heat or stress related proteins.
Table 3: Temperature dependent changes in gene expression of chaperones, proteases and other heat shock and stress induced proteins in recombinant E. coli BL21 (DE3) strain: Genes identified as DE and having negative logFC (log2) ratio in particular comparison are indicated as 'down', while DE genes with positive logFC ratio are indicated as 'up' regulated. Other classification is aligned with EcoCyc classification of genes. Comparisons of gene expression of production strain grown at different temperatures (P42_25, P37_25) and control strain grown at different temperatures (C42_25, C37_25) are presented. Functional classification of proteases was performed according to EcoCyc (http://www.ecocyc.org/expres-sion.html).
EcoCyc ID	Gene (ecoid)	Protein function	Comparison
P37_25 P42_25 C37_25 C42_25
Chaperones and other stress induced proteins
EG10599	groEL	chaperone	up		up
EG10241	dnaK	chaperone			up
EG10240	dnaJ	chaperone	up		
EG11534	ibpA	chaperone			up
G6463	clpS	chaperone		down	
EG12055	ccmE	chaperone			down
EG11973	ecpD	chaperone		down	
G378	fliJ	predicted chaperone		down	
EG11976	fliQ	flagellar biosythesis		down	down
G7039	fliY	cystine binding protein			down
EG12195	torD	chaperone		down	
EG11877	uspD	stress protein			up
EG11246	uspE	stress protein		up	
G7743	hslR	heat shock protein			down
Cytoplasm proteases G6991 yeaZ	Hypothetical protease		down	
Cytoplasmic membrane proteases EG10397 glpG EG10956 sohB EG12436 yaeL	intramembrane serine protease GlpG Possible Ser protease SohB RseP zinc protease down	down	up down	down
Periplasm proteases
EG11893	dacD
EG10463	degP
G6746	ydcP
PBP-6B, D-alanyl-D-alanine carboxypeptidase Serine protese Do Putative protease YdcP
down up
up
Table 4: Comparison of gene expression of proteases, chaperones and stress induced proteins in production and control BL21 (DE3) strain at different temperatures: Genes identified as DE and having negative logFC ratio in particular comparison are indicated as 'down', while DE genes with positive logFC ratio are indicated as 'up' regulated. Other classification is aligned with EcoCyc classification of genes. Comparisons of production versus control strain at 25 °C (P_C25), 37 °C (P_C37) and 42 °C (P_C42) are presented. Gene names and their classification is aligned with EcoCyc (http://www.ecocyc.org/expression.html).
EcoCyc ID Gene (ecoid)	Protein function		Comparison	
		P_C25	P_C37	P_C42
Sigma factors and regulators				
EG10897 rpoH (a32)	sigma factor controlling the heat shock response			down
EG12121 rssB	sigma regulator			down
Chaperones and stress induced proteins				
G7039 fliY	cystine binding protein			up
G6357 hscC	Chaperone			up
EG11877 uspD	stress protein			down
EG11246 uspE	stress protein	up		up
Cytoplasm proteases				
G6991 yeaZ	Hypothetical protease	down		
Cytoplasmic membrane proteases				
EG10956 sohB	Possible Ser protease SohB	down		
EG12436 yaeL	RseP zinc protease	up		
Periplasm proteases				
EG11893 dacD	PBP-6B, D-alanyl-D-alanine carboxypeptidase			up
Expression of these genes in our growth conditions was inspected in more detail. All together 24 genes showed significant changes in their expression levels in any of the comparisons (Table 3, 4). Most distinctive are gene expression profiles of cultures grown at 42 °C as most of DE genes were identified in comparisons C42_25, P42_25 and P_C42. As the heat shock response is well studied in E.coli we have compared our results to already published data, but our focus was set more to the differences in gene expression related to suboptimal growth temperature (P37_25, C37_25 and P_C25). Few genes coding for chaperones (clpS, ecpD, fliJ, torD, uspE) and a glpG gene coding for cytoplasmic membrane protease were regulated specifically due to change of temperature from 25 °C to 37 °C (and not to 42 °C). Interestingly, uspE gene was found to be the only chaperone gene and yaeL the only protease gene with higher expression in production strain compared to control strain when grown at 25 °C (Table 4). Similarly, yeaZ gene coding for putative cyto-plasmic protease and sohB gene coding for putative peri-
plasmic protease were found to be down-regulated in production strain when grown at 25 °C (Table 4).
3. 3. Verification of Microarray Results by Real-time PCR
Quantitative real-time PCR (qPCR) was used to verify microarray results. Among genes identified as DE in at least one of the comparisons six were chosen for further analysis: four genes coding for chaperones (groEL, ecpD, torD, uspD (yiit)) and two genes coding for proteases (yeaZ, ydcP). Gene expression changes obtained by mi-croarrays were in all comparisons confirmed by qPCR, except for two (Table 5).
3. 4. Temperature Dependent Accumulation of GroEL Chaperone
Information on gene expression was for GroEL chape-rone complemented also on the level of protein accumulation.
Table 5: Comparison of gene expression measurements by mi-croarrays and quantitative real-time PCR: Difference in expression is given as logFC. Comparisons of production versus control strain at 25 °C (P_C25) and 42 °C (P_C42) are presented as well as comparisons of production strain grown at different temperatures (P42_25) and control strain grown at different temperatures (C42_25, C37_25). M - microarray analysis results, qPCR - quantitative real-time PCR results. NS - nonsignificant difference in expression as detected in microarray analysis.
Gene	Method	P42_25	C37_25	C42_25	P_C25	P_C42
groEL	M	0.7	NS	1.7	NS	NS
	qPCR	1.6	-0.7	1.1		
ecpD	M	NS	-1.2	NS	NS	NS
	qPCR		-0.7			
torD	M	NS	-1.2	NS	NS	NS
	qPCR		-1.0			
uspD	M	1.1	NS	2.5	NS	-1.4
	qPCR	1.6		1.0		
yeaZ	M	NS	-2.2	NS	-1.6	NS
	qPCR		-0.9		-1.4	
ydcP	M	1.6	NS	NS	NS	NS
	qPCR	0.9				
The results showed (Figure 2) that GroEL protein content is increasing with increasing temperature in both cultures which is in line with the results of microarray hybridizations and qPCR. However, expression levels of GroEL are higher at T = 37 °C than at T = 42 °C in production strain. Statistically significant differences were obtained for comparison of protein content at 42 °C and 25 °C in control strain (p < 0.05) and for comparison of control and production strain at 42 °C (p < 0.05).
4. Discussion
Up to now many studies of E. coli transcriptome analysis changes during heat shock and/or recombinant
Figure 2: Recombinant E. coli GroEL content at different growth temperatures. Western blot analysis was performed for GroEL content in control strain (grey columns; (C)) and production strain (white columns; (P)). Bars represent standard error of biological replicate measurements. Statistically significant differences are marked with * (p < 0.05). Original gel and membrane images can be found in Supporting information 4.
protein production have been reported.21-23 However, not many studies focused on transcriptome changes in E. coli that occur during recombinant protein production at lower temperatures (i.e. at T = 25 °C) in particular with regard to formation of non classical IBs. Although in conditions investigated, we have observed limited changes in gene expression due to recombinant protein production in E. coli, our study suggests that some chaperones and proteases play an important role in the efficient production of the recombinant protein hG-CSF at suboptimal growth temperature. Interestingly, very limited changes in gene expression due to recombinant protein production in E. coli were observed (in contrast to control strain) in metabolic and regulatory pathways irrespective of the temperature growth condition (data not shown). Such small differences might indicate that the overexpression of recombinant protein could be more important stress factor for the recombinant protein producing cells which is able to „diminish" the temperature associated changes on gene expression.
4. 1. Role of Chaperones in Production of Active Recombinant Protein hG-CSF at Lower Temperature
The role of chaperones is to assist in proper folding of both native and recombinant cell proteins24. For example, it was shown that overproduction of GroES and GroEL or DnaK and DnaJ should prevent protein aggregation.25,26 We have observed increased expression levels of groEL (Table 3) at T = 42 °C in comparison to T = 25 °C in both, control and production strain and only in one strain type for dnaK and dnaJ which is in agreement with the previously published results27,28 and should assist in folding of proteins in heat stress conditions. Under stress E. coli needs chaperones first for its own survival and only afterwards for production of recombinant proteins. Thus, at elevated temperatures or overproduction of recombinant proteins chaperones might become limiting.24 In our study, the levels of GroEL were found to be lower in production strain compared to control (Fig. 2), indicating even further lack of this protein in producing strain. The increased expression level in comparison between production and control strain grown at T = 25 °C was observed for uspE gene (Table 4). The observed increased levels of uspE genes might therefore only confirm that they are induced due to a variety of stresses (e.g. heat shock or recombinant protein production)29 while the exact mechanism of their function remains unknown.
Flagellum-specific chaperones (e.g. FliJ, FliQ and FliY), the type III cytoplasmic chaperone family members, assist in folding and export of flagellar proteins.30 fli-Y, coding for FliY, a periplasmic31 cystine-binding pro-tein32 participates in regulation of class III transcription through regulation oF (rpoF, a28) activity.33 rpoF is a minor sigma factor responsible for initiation of transcription
of a number of genes involved in motility.34 Higher activity of fliY gene (a significant decrease in expression level at T = 42 °C in comparison to T = 25 °C in the control strain, Table 3) at suboptimal growth temperature, activates rpoF whose expression levels remain unchanged (see Supporting information). Similarly, fliQ and fliJ coding for integral membrane components of the flagellar export apparatus30,34 showed increased expression levels at T = 25 °C in the control strain (Table 3). Direct implication of the increased activity of flagellum specific chaperones on efficiency of hG-CSF protein expression and folding on the basis of currently known functions of Fli proteins cannot be taken. These results suggest that E. coli flagellum-specific chaperones might have some so far unidentified functions that could contribute to improved folding of recombinant proteins.
4. 2. Hypothetical Protease YeaZ is Repressed in Recombinant E.coli Producing hG-CSF When Grown at 25 °C
At elevated temperatures cytoplasm proteases and cytoplasmic membrane proteases are in majority down-regulated and periplasm proteases up-regulated (Table 3). The only cytoplasm protease that has higher gene activity at lower growth temperature (in the control strain) and could thus contribute to more efficient recombinant protein production is YeaZ. Potential role of periplasm located proteases in structure of IBs is even more obscure. Interestingly however, the expression level of hypothethical protease yeaZ gene was decreased at 25 °C in production strain if compared to control strain (Table 4). Although not much data is available on function of this protein, it is known that together with two essential proteins YjeE and YgjD it forms an interaction network whose cellular role remains unknown.35 Nevertheless, YeaZ could be an important factor in protein interactions that lead to different accumulation levels of recombinant protein and presents interesting area for future research.
4. 3. Membrane Localized Protease YaeL is Induced in Recombinant E.coli at Suboptimal Temperature
It is known that the function of YaeL is to cleave RseA in the cytoplasmic or intramembrane region36 and to exhibit proteolytic activity toward RpoE and RpoH.37 Continual degradation of RseA by YaeL is needed to provide the cell with sufficient free RpoE to support cell viability.38 According to our results yaeL is upregulated at suboptimal temperature in production strain if compared to both temperatures (Table 3). This suggests that cell viability is increased at suboptimal temperature, hence YaeL could be an important factor in protein interactions that lead to different accumulation levels of re-
combinant proteins and presents interesting area for future research.
5. Conclusions
Escherichia coli adaptation to internal and external signals is an extremely important mechanism and must be tightly regulated in order to assure survival of the organism. Recombinant protein production is representing a stress for bacterial cells as is also change in the growth temperature. An interesting and so far not understood phenomenon is the increased content of correctly folded recombinant protein in IBs of E. coli grown at 25 °C.
Our hypothesis was that chaperones and/or proteases might have an important role in this process. The lowered activity of a certain group of proteases should render the newly formed molecules of hG-CSF less vulnerable to proteolytic hydrolysis. On the other hand, some cha-perones could be identified that contribute to proper folding of protein in IBs at suboptimal growth temperature. We have found several chaperones (Table 3) that were regulated in our experiment, mostly however had increased levels detected in cultures grown at 42 °C and not at 25 °C. The only chaperone with higher gene activity in production strain grown at 25 °C was UspE (Table 4). No other experimental data is currently available that would link these proteins to folding of recombinant proteins. Among proteases, cytoplasmic proteases are of main interest since IBs are mostly found in the bacterial cytoplasm.24 Gene expression of cytoplasm protease yeaZ was found to be decreased at 25 °C in production strain whereas gene expression of cytoplasmic membrane protease yaeL was found to be increased at 25 °C in production strain (Table 4). Thus yaeL and yeaZ genes are candidates for improved recombinant protein production at lower growth temperature. Taken altogether, gene expression profiling provided additional information on mechanisms of recombinant protein production in E. coli at suboptimal growth temperatures. Further research is needed to directly confirm the potential role of candidate genes in this process.
6. Acknowledgements
The authors would like to acknowledge Neža Turnšek and Katja Barle for help with sample preparations and qPCR analysis, Karmen Čerkic and Maruša Hafner for help with cultivation of E. coli and Mateja Skok for Western blotting analysis. The work was partially conducted at the Center for Functional Genomics and Bio-Chips, Ljubljana, Slovenia and was also supported by Sandoz, Biopharmaceuticals, Lek Pharmaceuticals d.d. and Slovenian research agency programs P1-0340 and P4-0165. This work is dedicated to inspiring mentor
dr. Viktor Menart who has unexpectedly passed away during the study.
7. References
1.	L. S. A. Vanz, G. Renard, S. M. Palma, J. M. Chies, S. L. Dalmora, L. A. Basso, D. S. Santos, Microb. Cell Fact. 2008, 7, 13.
2.	M. Miot, J-M. Betton, Microb. Cell Fact. 2004, 3, 4.
3.	S. Ventura, A. Villaverde, Trends Biotechnol. 2006, 24, 17985.
4.	E. García-Fruitós, N. González-Montalbán, M. Morell, A. Vera, R. M. Ferraz, A. Arís, S. Ventura, A. Villaverde, Microb. Cell Fact. 2005, 4, 27.
5.	J. Tydemers, A. Mogk, B. Bukau, Nat. Rev. Mol. Cell Biol. 2010, 11, 777-788.
6.	S. Peternel, M. Bele, V. Gaberc-Porekar, V. Menart, Microb. Cell Fact. 2006, 5, P23.
7.	S. Jevševar, V. Gaberc-Porekar, I. Fonda, B. Podobnik, J. Gr-dadolnik, V. Menart, Biotechnol. Prog. 2005, 21, 632-639.
8.	S. Peternel, J. Grdadolnik, V. Gaberc-Porekar, R. Komel, Microb. Cell Fact. 2008, 7, 34.
9.	M. Martínez-Alonso, N. González-Montalbán, E. García-Fruitós, A. Villaverde, Microb. Cell Fact. 2009, 8, 4.
10.	K. Maruyama, K. Tsuji, R. Tanaka, K. Yamada, Y. Kodera, T. Nakahata, Bone Marrow Transplant. 1998, 22, 313-320.
11.	S. Jevševar, J. Palčič, S. Jalen, A. Pavko, Acta Chim. Slov. 2007, 54, 360-365.
12.	M. Petek, S. Baebler, D. Kuzman, A. Rotter, Z. Podlesek, K. Gruden, M. Ravnikar, U. Urleb, BMC Microbiol. 2010, 10, 159.
13.	S. Baebler, H. Krečič-Stres, A. Rotter, P. Kogovšek, K. Cankar, E.J. Kok, K. Gruden, M. Kovač, J. Žel, M. Pompe-No-vak, M. Ravnikar, Mol. Plant Pathol. 2009, 10, 263-75.
14.	G. K. Smyth, Stat. Appl. Genet. Mol. Biol. 2004, 3, Article 3.
15.	A. Subramanian, P. Tamayo, V. K. Mootha, S. Mukherjee, B. L. Ebert, M. A. Gillette, A. Paulovich, S.L. Pomeroy, T. R. Golub, E.S. Lander, J. P. Mesirov, Proc. Natl. Acad. Sci. U.S.A. 2005, 102(43), 15278-9.
16.	M. Pompe-Novak, K. Gruden, S. Baebler, H. Krečič-Stres, M. Kovač, M. Jongsma, M. Ravnikar, Physiol. Mol. Plant Pathol. 2006, 67, 237-247.
17.	T. Nolan, R. E. Hands, S. A. Bustin, Nat. Protoc. 2006, 1, 1559-1582.
18.	M. Skulj., V. Okrslar, S. Jalen, S. Jevsevar, P. Slanc, B. Stru-kelj, V. Menart, Microb. Cell Fact. 2008, 7, 6.
19.	M. Arifuzzaman, M. Maeda, A. Itoh, K. Nishikata, C. Takita, R. Saito, T. Ara, K. Nakahigashi, H-C. Huang, A. Hirai, K. Tsuzuki, S. Nakamura, M. Altaf-Ul-Amin, T. Oshima, T. Baba, N. Yamamoto, T. Kawamura, T. Ioka-Nakamichi, M. Ki-tagawa, M. Tomita, S. Kanaya, C. Wada, H. Mori, Genome Res. 2006, 16, 686-691.
20.	S. Gottesman, Annu. Rev. Genet. 1996, 30, 465-506.
21.	S. W. Harcum, F. T. Haddadin, J. Ind. Microbiol. Biotechnol. 2006, 33, 801-814.
22.	M. D. Carruthers, C. Minion, FEMS Microbiol. Lett. 2009, 295, 96-102.
23.	M. Gadgil, V. Kapur, W-S. Hu, Biotechnol. Prog. 2005, 21, 689-699.
24.	A. Villaverde, M. M. Carrio, Biotechnol. Lett. 2003, 25, 1385-1395.
25.	A. Gragerov, E. Nudler, N. Komissarova, G. A. Gaitanaris, M. E. Gottesman, V. Nikiforov, Proc. Natl. Acad. Sci. USA
1992,	89, 10341-10344.
26.	F. U. Hartl, Nature 1996, 381, 571-579.
27.	M. Y. Sherman, A. L. Goldberg, Proc. Natl. Acad. Sci. USA
1993,	90, 8648-8652.
28.	J. Chatellier, F. Hill, P. A. Lund, A. R. Fersht, Proc. Natl. Acad. Sci. USA 1998, 95, 9861-9866.
29.	N. Gustavsson, A. Diez, T. Nyström, Mol. Microbiol. 2002, 43, 107-17.
30.	T. Minamino, K. Namba, J. Mol. Microbiol. Biotechnol. 2004, 7, 5-17.
31.	D. S. Mytelka, M. J. Chamberlin, J. Bacteriol. 1996, 178, 24-34.
32.	J. D. Butler, S. W. Levin, A. Fachiano, L. Miele, A. B. Muk-heriee, Life Sci. 1993; 52, 1209-15.
33.	T. Minamino, R. M. Macnab, J. Bacteriol. 1999, 181, 1388-94.
34.	X. Mao, Y. Zhang, Y. Xu, PLoS One, 2011, 6-7, e22556.
35.	J. I. Handford, B. Ize, G. Buchanan, G. P. Butland, J. Greenblat, A. Emili, T. Palmer, J. Bacteriol. 2009, 191, 4701-4.
36.	K. Kanehara, K. Ito, Y. Akiyama, Genes Dev. 2002, 16, 2147-2155.
37.	C. Dartigalongue, H. Loferer, S. Raina, EMBO J., 2001, 20, 5908-5918.
38.	S. E. Ades, Curr. Opin. Microbiol., 2004, 7, 157-162.
Povzetek
V nedavnih raziskavah je bilo dokazano, da se v inkluzijskih telesih pri nižjih temperaturah gojenja (25 °C) izloča večji delež pravilno zvitega rekombinantnega proteina hG-CSF. Naša raziskava je bila zato usmerjena v preučevanje in razlago procesov, ki sodelujejo pri tvorbi »neklasičnih« inkluzijskih teles predvsem s pomočjo preučevanja genskega izražanja proteaz in šaperonov. V primeru gojenja kultur pri suboptimalni temperaturi smo s statistično analizo oligonu-kleotidnih mikromrež dobili vidne rezultate sprememb v izražanju genov vpletenih v energetski metabolizem, v meta-bolizem aminokislin in nukleotidov ter kofaktorjev in sekundarnih metabolitov. Izmed do zdaj znanih genov, ki nosijo zapis za šaperone, proteaze in ostale s stresom povezane proteine jih je bilo kar 24 statistično značilno diferencialno izraženih. Tako so UspE med šaperoni in YaeL ter YeaZ med proteazami najbolj pomembni kandidati s potencialno vlogo pri akumulaciji aktivnih rekombinantnih proteinov. Ugotovili smo tudi, da ima pri nižjih temperaturah gen yaeL, ki nosi zapis za izražanje v membrani lokalizirane proteaze YaeL večjo aktivnost kar ga funkcijsko potencialno povezuje z bolj učinkovitim izražanjem rekombinantnih proteinov. Rezultati raziskave zato predstavljajo napredek v razumevanju produkcije rekombinantih proteinov bakterije E. coli. Prav ti geni, ki pri nižji temperaturi gojenja potencialno vplivajo na produkcijo rekombinantnih proteinov predstavljajo osnovo za nadaljnje raziskave pri izboljšavi produkcijskih sevov bakterije E. coli kot tudi pri izboljšavi fermentacijskih procesov.
Supporting information
Supporting information 1
Results of statistical testing for differentially expressed genes. Two-way analysis of variance was performed with the cut-off p-value 0.05. DE genes due to recombinant protein production (P_C), growth temperature (37_25, 42_25) and interaction of both factors (interaction) are listed. Each worksheet within the Excel file (ANOVA analysis) consists of different columns descri-
bing numerical identification of a gene (geneid), EcoCyc identification number (ecocyc), gene name (ecoid), definition and class of the protein, statistical significance (P. value),.... and normalized data using quantile normalization at different temperatures (e.g. N-25-BL21-10-1 stands for normalized data of strain of E. coli control strain grown at T = 25 °C and harvested at OD = 10).
Supporting information 2
Table 1: Temperature dependent changes in energy metabolism, in metabolism of amino acids, and of nucleotides, in biosynthesis of cofactors and secondary metabolites of recombinant BL21 (DE3) strain: Comparisons of gene expression of control strain grown at different temperatures (C42_25, C37_25)- Genes are identified as DE and having negative logFC ratio in particular comparison are indicated as 'down', while DE genes with positive logFC ratio are indicated as 'up' regulated. Gene names and their classification is aligned with EcoCyc (http://www.ecocyc.org/ex-pression.html).
Gene Protein function	Pathways	Comparison	
(ecoid) Cellular function		C37_25	C42_25
C1 compounds utilization and assimilation			
frmA glutathione-dependent formaldehyde	formaldehyde oxidation II	down	down
dehydrogenase	(gluthathione dependent)		
alcohols degradation			
glpB glycerol-3-phosphate dehydrogenase (anaerobic)	glycerol degradation I	up	up
glpD glycerol 3-phosphate dehydrogenase (aerobic)	glycerol degradation I	up	up
carbohydrates biosynthesis			
eno enolase	gluconeogenesis I	up	
ybhA pyridoxal phosphatase/fructose 1,6-bisphosphatase gluconeogenesis I		down	
galM galactose-1-epimerase	galactose degradation I (Leloir pathway)	up	
rbsD ribose pyranase	ribose degradation	down	
mdh malate dehydrogenase	gluconeogenesis I		up
pgi phosphoglucose isomerase	gluconeogenesis I		up
mngB a-mannosidase	2-O-a-mannosyl-D-glycerate degradation	down	down
araB L-ribulokinase	L-arabinose degradation I	down	down
fatty acid and lipids biosynthesis			
lpxP palmitoleoyl acyltransferase	superpathway of KDO2-lipid A biosythesis		down
waaA KDO transferase	lipid-A-precursor biosynthesis	up	
cfa cyclopropane fatty acyl phospholipid synthase	cyclopropane fatty acid (CFA) biosynthesis	up	
waaA KDO transferase	superpathway of KDO2-lipid A biosythesis	up	
Fermentation			
acnB 2-methylisocitrate dehydratase	2-methylcitrate cycle I	down	down
acnB bifunctional aconitate hydratase 2	mixed acid fermentation	down	down
mdh malate dehydrogenase	mixed acid fermentation		up
pykF pyruvate kinase I	mixed acid fermentation	up	
ldhA D-lactate dehydrogenase	mixed acid fermentation	down	
frdB fumarate reductase iron-sulfur protein	mixed acid fermentation	down	
Glycolysis			
pgi phosphoglucose isomerase	glyoxylate bypass and TCA		up
mdh malate dehydrogenase	glyoxylate bypass and TCA		down
sucC succinyl-CoA synthetase (P subunit)	glyoxylate bypass and TCA		down
acnB bifunctional aconitate hydratase 2	glyoxylate bypass and TCA		down
ybhA pyridoxal phosphatase/fructose 1,6-bisphosphatase glycolysis I		down	
Gene Protein function	Pathways	Comparison	
(ecoid) Cellular function		C37_25	C42_25
Glycolysis			
pfkB 6-phosphofructokinase II	glycolysis I -Etner-Doudoroff	up	
eno enolase	glycolysis I Etner-Doudoroff	up	
pykF pyruvate kinase I	glycolysis I Etner-Doudoroff	up	
ybhA pyridoxal phosphatase/fructose 1,6-bisphosphatase glycolysis I-TCA		down	
aceB	glycolysis I-TCA	down	
acnB bifunctional aconitate hydratase 2	glycolysis I-TCA	down	
pfkA malate synthase A	glycolysis I-TCA	up	
eno enolase	glycolysis I-TCA	up	
pykF pyruvate kinase I	glycolysis I-TCA	up	
inorganic nutrients metabolism			
ssuD FMNH2-dependent alkanesulfonate	two-component alkanesulfonate	down	down
monooxygenase cysH 3'-phospho-adenylylsulfate reductase	monooxygenase sulfate reduction I (assimilatory)		up
cysJ sulfite reductase (flavoprotein subunit complex)	sulfate reduction I (assimilatory)		up
respiration			
acnB bifunctional aconitate hydratase 2	anaerobic respiration	down	down
mdh malate dehydrogenase	anaerobic respiration		up
pykF pyruvate kinase I	anaerobic respiration	up	
eno enolase	anaerobic respiration	up	
frdB fumarate reductase iron-sulfur protein	anaerobic respiration	down	
pentose phosphate pathways			
tktA transketolase I	pentose phosphate pathway	down	
rpe ribulose phosphate 3-epimerase	pentose phosphate pathway	up	
superpathway of glycolysis and Entner Doudoroff			
pgi phosphoglucose isomerase	glycolysis I - Enter-Doudoroff		up
cell structures biosynthesis			
pgi phosphoglucose isomerase	colanic acid building blocks biosythesis		up
rffG dTDP-glucose 4,6-dehydratase 2	enterobacterial common antigen biosythesis	down	down
glf UDP-galactopyranose mutase	dTDP-L-rhamnose biosythesis I	up	
rfbC dTDP-4-dehydrorhamnose 3,5-epimerase	dTDP-L-rhamnose biosythesis I	down	
amine and polyamines biosynthesis			
argI ornithine carbamoyltransferase	aminopropylcadaverine biosynthesis		down
nagB glucosamine-6-phosphate deaminase	N-acetylglucosamine degradation		up
amino acids biosynthesis			
ydiB quinate/shikimate dehydrogenase	tryptophan biosythesis	down	
trpC phosphoribosylanthranilate isomerase /	tryptophan biosythesis	down	
indole-3-glycerol phosphate synthase			
trpB tryptophan synthase, P subunit dimer	tryptophan biosythesis	up	
ilvI catalytic subunit of acetolactate synthase III	leucine biosynthesis	down	
ilvI catalytic subunit of acetolactate synthase III	isoleucine biosynthesis (from threonine)	down	
ansB L-asparagine aminohydrolase II	glutamate degradation II	down	
ilvI catalytic subunit of acetolactate synthase III	isoleucine biosynthesis I (from threonine)	down	
speE spermidine synthase	aminopropylcadaverine biosynthesis	down	
gcvT aminomethyltransferase	glycine cleavage complex	down	
dapE N-succinyl-L-diaminopimelate desuccinylase	lysine biosythesis I	down	down
proA glutamate-5-semialdehyde dehydrogenase	proline biosythesis I		down
iscS cysteine desulfurase	alanine biosythesis I		up
cysK cysteine synthase A	cysteine biosythesis I		up
nucleosides and nucleotides biosynthesis			
pyrE orotate phosphoribosyltransferase	pyrimidine ribonucleotides interconversion (de novo biosythesis)	up	up
deoD purine nucleoside phosphorylase deoD-type	salvage pathways of adenine, hypoxanthine and their nucleosides		up
deoD purine nucleoside phosphorylase deoD-type	salvage pathways of guanine, xanthine and their nucleosides		up
dcd dCTP deaminase	pyrimidine deoxyribonucleotides de novo biosythesis	down	down
			
Gene (ecoid)	Protein function Cellular function	Pathways	Comparison C37_25 C42_25
nucleosides and nucleotides biosynthesis
trpC indole-3-glycerol phosphate synthase / phosphoribosylanthranilate isomerase cdd cytidine deaminase udp uridine phosphorylase
tryptophan biosythesis
salvage pathways of pyrimidine ribonucleotides salvage pathways of pyrimidine ribonucleotides
up up
down
nucleosides and nucleotides degradation and recycling
deoD purine nucleoside phosphorylase deoD-type deoD purine nucleoside phosphorylase deoD-type cdd cytidine deaminase cdd cytidine deaminase udp uridine phosphorylase
purine deoxyribonucleosides degradation degradation of purine ribonucleosides pyrimidine deoxyribonucleosides degradation up degradation of pyrimidine ribonucleoside	up
degradation of pyrimidine ribonucleoside	up
up up
secondary metabolites degradation
idnD	L-idonate 5-dehydrogenase
idnO	5-keto-D-gluconate 5-reductase
garL	a-dehydro-P-deoxy-D-glucarate aldolase
pfkB	6-phosphofructokinase II
gatZ	Subunit of tagatose-1,6-bisphosphate aldolase 2
gatY	Subunit of tagatose-1,6-bisphosphate aldolase 2
L-idonate degradation
L-idonate degradation	up
D-glucarate and D-galactarate degradation	up
galactitol degradation	up
galactitol degradation	up
galactitol degradation	up
down up
cofactors, prosthetic groups, electron carriers biosynthesis
cysG uroporphyrin III C-methyltransferase cobC predicted a-ribazole-5'-P phosphatase thiC thiamin biosynthesis protein ThiC thiG thiazole synthase gshB glutathione synthetase ribE 6,7-dimethyl-8-ribityllumazine synthase ubiH 2-octaprenyl-6-methoxyphenol hydroxylase btuR cobinamide adenosyltransferase / cobalamin
adenosyltransferase cobU cobinamide-P guanylyltransferase / cobinamide kinase
thiM hydroxyethylthiazole kinase
nadK NAD kinase
pdxH pyridoxine 5'-phosphate oxidase /
pyridoxamine 5'-phosphate oxidase folE GTP cyclohydrolase I folB dihydroneopterin aldolase gcvT Aminomethyltransferase
siroheme biosythesis
adenosylcobalamin salvage from cobinamide I thiamin biosynthesis I thiamin biosynthesis I glutathione biosythesis flavin biosythesis I (bacteria) ubiquinone-8 biosythesis (prokaryotic) adenosylcobalamin salvage from cobinamide I
down down down
up
adenosylcobalamin salvage from cobinamide I	up
thiamin biosynthesis I	up
NAD phosphorylation and dephosphorylation	down
pyridoxal 5'-phosphate biosythesis and salvage	up
tetrahydrofolate biosythesis	up
tetrahydrofolate biosythesis	up
formylTHF biosythesis I	down
up down down down up up down
Table 2: Functional analysis of differences in gene expression observed between production and control BL21 (DE3) strain grown at different temperatures: Genes identified as DE and having negative logFC ratio in particular comparison are indicated as 'down', while DE genes with positive logFC ratio are indicated as 'up' regulated. Other classification is aligned with EcoCyc classification of genes. Comparisons of production versus control strain at 25 °C (P_C25), 37 °C (P_C37) and 42 °C (P_C42) are presented. Gene names and their classification is aligned with EcoCyc (http://www.ecocyc.org/expression.html).
Gene Protein function	Pathways		Comparison	
(ecoid)				
Energy metabolism		P_C25	P_C37	P_C42
inorganic nutrients metabolism				
ssuD FMNH2-dependent alkanesulfonate	two-component alkanesulfonate monooxygenase	down	down	
monooxygenase				
carbohydrates biosynthesis				
mngB a-mannosidase	2-O-a-mannosyl-D-glycerate degradation	down		
araB L-ribulokinase	L-arabinose degradation I	down		
galM galactose-1-epimerase	galactose degradation I (Leloir pathway)	up		
mdh malate dehydrogenase	gluconeogenesis I			down
pgi phosphoglucose isomerase	gluconeogenesis I			down
Gene Protein function (ecoid)	Pathways		Comparison	
Energy metabolism		P_C25	P_C37	P_C42
Fermentation acnB 2-methylisocitrate dehydratase mdh malate dehydrogenase	2-methylcitrate cycle I mixed acid fermentation		uP	down
pentose phosphate pathways rpe ribulose phosphate 3-epimerase tktA transketolase I talA transaldolase A gnd 6-phosphogluconate dehydrogenase (decarboxylating)	pentose phosphate pathway pentose phosphate pathway pentose phosphate pathway (oxidative branch) pentose phosphate pathway (oxidative branch)		down up	down up
fatty acid and lipids biosynthesis cfa cyclopropane fatty acyl phospholipid synthase	cyclopropane fatty acid (CFA) biosynthesis		down	
Respiration mdh malate dehydrogenase acnB 2-methylisocitrate dehydratase	anaerobic respiration anaerobic respiration		up	down
Glycolysis mdh malate dehydrogenase pgi phosphoglucose isomerase acnB 2-methylisocitrate dehydratase	glyoxylate bypass and TCA glyoxylate bypass and TCA glyoxylate bypass and TCA		up	down down
superpathway of glycolysis and Entner Doudoroff pgi phosphoglucose isomerase glycolysis I - Enter-Doudoroff				down
carbohydrates degradation rpib allose-6-phosphate isomerase / ribose-5-phosphate isomerase B	D-allose degradation			up
amino acids degradation cysK cysteine synthase A	L-cysteine degradation II			down
Metabolism of amino acids				
amino acids biosynthesis ydiB quinate/shikimate dehydrogenase trpC phosphoribosylanthranilate isomerase / indole-3-glycerol phosphate synthase dapE N-succinyl-L-diaminopimelate desuccinylase gcvT Aminomethyltransferase cysK cysteine synthase A proA glutamate-5-semialdehyde dehydrogenase iscS cysteine desulfurase dapA dihydrodipicolinate synthase	tryptophan biosythesis tryptophan biosythesis lysine biosythesis I glycine cleavage complex cysteine biosythesis I proline biosythesis I alanine biosythesis I aspartate biosynthesis	down down	up	down down down up down down
amino acids degradation cysK cysteine synthase A	L-cysteine degradation II			down
cell structures biosynthesis pgi phosphoglucose isomerase glf UDP-galactopyranose mutase	colanic acid building blocks biosythesis 0-antigen building blocks biosynthesis (E. coli)			down up
Biosynthesis of cofactors and secondary metabolites				
secondary metabolites degradation gatY Subunit of tagatose-1,6-bisphosphate aldolase 2 gatZ Subunit of tagatose-1,6-bisphosphate aldolase 2 idnD L-idonate 5-dehydrogenase yjjN predicted L-galactonate oxidoreductase uxaB altronate oxidoreductase	galactitol degradation galactitol degradation L-idonate degradation L-galactonate degradation D-galacturonate degradation I	up	down down up	up up
Gene	Protein function	Pathways		Comparison
(ecoid)				
	Energy metabolism		P_C25	P_C37 P_C42
cofactors, prosthetic groups, electron carriers biosynthesis				
cobU	cobinamide-P guanylyltransferase / cobinamide kinase	adenosylcobalamin salvage from cobinamide I	up	
thiC	thiamin biosynthesis protein ThiC	thiamin biosynthesis I	down	
cobC	predicted a-ribazole-5'-P phosphatase	adenosylcobalamin salvage from cobinamide I		uP
thiG	thiazole synthase	thiamin biosynthesis I	down	
gcvT	aminomethyltransferase	formylTHF biosythesis I		down
frmA	glutathione-dependent formaldehyde	formaldehyde oxidation II (gluthathione dependent)		uP
	dehydrogenase	(gluthathione dependent)		
pdxH	pyridoxine 5'-phosphate oxidase / pyridoxamine 5'-phosphate oxidase	pyridoxal 5'-phosphate biosythesis and salvage		down
cysG	uroporphyrin III C-methyltransferase	siroheme biosythesis		down
btuR	cobinamide adenosyltransferase / cobalamin adenosyltransferase	adenosylcobalamin salvage from cobinamide I		uP
cobs	cobalamin 5'-phosphate synthase / cobalamin synthase	adenosylcobalamin salvage from cobinamide I		uP
thiM	hydroxyethylthiazole kinase	thiamin biosynthesis I		uP
gshB	glutathione synthetase	glutathione biosythesis		down
ribE	6,7-dimethyl-8-ribityllumazine synthase	flavin biosythesis I (bacteria)		down
ribF	bifunctional riboflavin kinase / FMN adenylyltransferase			uP
Supporting information 3
Figure 1: Overview of changes in gene expression of recombinant E. coli grown at different temperatures DE gene sets from comparison A) between 37 °C and 25 °C, and B) between 42 °C and 25 °C in control strain are presented using EcoCyc Omics Viewer [http://biocyc.org/ECOLI/over-view-expression-map]. Lines in the diagram correspond to reactions; nodes correspond to cell components (□ - carbohydrates, A - amino acids, O - proteins, V - cofactors, T - tRNA, O - other). Biosynthetic pathways are positioned in the left of the cytoplasm, degradative pathways in the right, signaling pathways in the bottom-left. Reactions not assigned to any pathway are in the far right of the cytoplasm. In the inner and outer membranes are transporters (with arrows) and other membrane proteins. In the periplasm are periplasmic enzymes and other periplasmic proteins. Log-FC ratios of DE genes are presented coloured according to the scale on the image. Each figure (A and B) is marked with red letters to show cellular function. A = cofactors, prosthetic groups, electron carriers biosynthesis; B = fatty acids and lipids biosynthesis; C = signal transduction pathways; D = glycolysis; E = fermentation; F = respiration; G = pentose phosphate pathways; H = amines and polyamines biosynthesis; I = superpath-way of glycolysis and Entner-Doudoroff; J = inorganic nutrients metabolism; K = secondary metabolites degradation; L= alcohols degradation; M= transporters
B)
Expre ,. .1 l.. i. t2.4
Supporting information 4
Original images supporting figure 2 of the manuscript
A) SDS - PAGE of E. coli proteins and B) Western blott analysis for GroEL: 1) Standard, 2) P42-1, 3) P42-2, 4) P42-3, 5) P37-1, 6) P37-2, 7) P37-3, 8) P25-1, 9) P25-2, 10) P25-3 (A) 11) Standard, 12) C42-1, 13) C42-2, 14) C42-3, 15) C37-1, 16) C37-2, 17) C37-3, 18) C25-1, 19) C25-2