Genome sequence and analysis of Escherichia coli production strain LS5218.

Escherichia coli strain LS5218 is a useful host for the production of fatty acid derived products, but the genetics underlying this utility have not been fully investigated. Here, we report the genome sequence of LS5218 and a list of large mutations and single nucleotide permutations (SNPs) relative to E. coli K-12 strain MG1655. We discuss how genetic differences may affect the physiological differences between LS5218 and MG1655. We find that LS5218 is more closely related to E. coli strain NCM3722 and suspect that small genetic differences between K-12 derived strains may have a significant impact on metabolic engineering efforts.

Introduction

Escherichia coli strain LS5218 is frequently studied for the production of polyhydroxyalkanoates (PHAs) from mixtures of sugars and organics acids (Agnew et al., 2012, Nduko et al., 2012, Salamanca-cardona et al., 2014). LS5218 is selected because of two commonly cited differences from other E. coli strains – mutations in fadR (fadR601) and atoC (atoC(c)). The fadR601 disrupts expression of FadR thereby deregulating the fad genes that encode enzymes responsible for β-oxidation (Fujita et al., 2007). AtoC is an activator of the atoDAEB operon, encoding enzymes required for catabolism of acetoacetate and other short-chain organic acids (Lioliou et al., 2005, Theodorou et al., 2011). The atoC(c) mutation alters the regulator and causes constitutive expression and upregulation of the atoDAEB operon (Jenkins and Nunn, 1987, Matta et al., 2007). The mutations in E. coli LS5218 allow for increased uptake and utilization of a wider array of fatty acid chain-lengths and make it well-adapted for the engineering of short chain length-co-medium chain length (SCL-co-MCL) copolymers and poly(3-hydroxybutyrate-co-3-hydroxyvalerate) [P(3HB-co-3HV)] (Rhie and Dennis, 1995, Tappel et al., 2012). Despite its widespread use in PHA production studies, the genome sequence of E. coli LS5218 has not been made publicly available. This is in part due to the common assumption that it is a close relative of the sequenced E. coli K-12 strain MG1655.

While a variety of E. coli strains are widely used by researchers, the history of their isolation is not as widely known. The original E. coli K-12 was isolated in 1922 and deposited in the Stanford University strain collection in 1925 (Neidhardt et al., 1996). The two main wild-type K-12 strains, WG1 from J. Lederberg and EMG2 from Clowes and Hayes, are subcultures of the Stanford K-12 strain. The published derivation of E. coli LS5218 involved a two-step screening of spontaneous mutants on selective media (Fig. 1A) (Spratt et al., 1981). Strain RS3010 was a spontaneous mutant of the Lederberg E. coli K-12 strain selected for growth on decanoate, to isolate a mutant with upregulated β-oxidation gene expression (Simons et al., 1980). Strain LS5218 was generated as a spontaneous mutant of RS3010 selected for on butyrate in order to isolate a strain with the ability to metabolize SCL fatty acids (Spratt et al., 1981). E. coli MG1655 was derived from an original K-12 isolate from the Lederberg lab through a two-step process designed to cure out the bacteriophage lambda (UV radiation and blood agar selection) and the F plasmid (acridine orange) (Blattner et al., 1997a). E. coli MG1655 and E. coli LS5218 appear to be derived from the same E. coli K-12 isolate (the Lederberg K-12 strain), but differences in their derivation histories convinced us to sequence E. coli LS5218 to know the exact genetic background of this production strain. Here, we report the genome sequence of E. coli LS5218 and an analysis of its content relative to E. coli MG1655 and a closer relative E. coli NCM3722.

Fig. 1

Comparison of LS5218 and MG1655. A. Diagram of the published derivation paths for LS5218 and MG1655. B. Pan-genome phylogenetic tree for E. coli K-12 strains. Strains listed with published name, or accession number if a published name was not listed. Spon. Mut., spontaneous mutation.

Results and discussion

E. coli LS5218 genomic DNA was sequenced using paired end reads on a HiSeq. 2500 System, then assembled into 121 contigs using SPAdes (Bankevich et al., 2012). The draft genome was deposited in GenBank (GCA_002007165.1) and the NCBI Prokaryotic Genome Annotation Pipeline (PGAP) automatically assigned annotations. Using the annotated protein features for LS5218, we generated a phylogenetic tree comparing LS5218 with 21 completely sequenced E. coli K-12 derivatives using the Bacterial Pan Genome Analysis pipeline (BPGA) (Chaudhari et al., 2016). The pan genome analysis compiled a set of core genes common in all strains, accessory genes present in at least two or more strains, and unique genes only found in a single strain. The algorithm used this information to perform phylogenetic clustering of the E. coli K-12 derivative strains based on their variable gene content (Fig. 1B). The phylogenetic clustering shows that LS5218 is more closely related to the newly sequenced strain E. coli NCM3722 than it is to MG1655.

Next, we compared the newly assembled LS5218 genome with the E. coli MG1655 reference genome to evaluate the genetic relationship between the two strains. We used the Mauve genome alignment software (Darling et al., 2004, Darling et al., 2010) to align the genome contigs for LS5218 against MG1655 and found 17 large differences in the form of gene insertions, deletions and genome rearrangements between the two strains (Table 1). We also performed single nucleotide polymorphism (SNP) and indel analysis with FreeBayes (Garrison and Marth, 2012), and found 74 small differences listed in Table 2. Through this analysis, we confirmed the presence of mutations in fadR and atoC.

Table 1

Table of Large insertions and deletions between MG1655 and LS5218.

Location	MG1655	LS5218	Comments
257905‐258680	IS1I	No insert	Intact crl gene in LS5218
279599‐291070b	No insert	Deletion – recombination at insA elements	Deletion of 11 genes of cryptic prophage CP4–6
574587‐575785b	insH1	No insert	Intact nmpC gene in LS5218
687850‐689049b	insH1	No insert	IS5 upstream of gltIJKL operon in MG1655
807329b	No insert	λ phage	Wild type λ phage in LS5218
916878a	No insert	Insertion in ybjD	Premature stop codon
1299498–1300697	IS5U	No insert	upstream of oppA
1878573b	No insert	IS5	Disrupted yeaP gene in LS5218
1978505–1979294b	IS1	Tn1000	Insertions upstream of flhDC
2101742–2102945b	IS5	No insert	Intact wbbL gene in LS5218
2110297–2128593b	No insert	IS1 and 18 kb deletion	Deleted: rfbA, rfbD, rfbB, galF, wcaM, wcaL, wcaK, wzxC, wcaJ, cpsG, cpsB, wcaI, fcl, gmd, wcaF
2170165–2171620b	IS3	No insert	Intact gatR in LS5218
3130145a	IS5	3.5 kb insert	Inserted: fatty acyl-AMP ligase, short chain dehydrogenase, ACP binding site family protein
3365549–3366752b	IS5	No insert	Intact yhcE gene in LS5218
4480807b	No insert	IS1	Disrupted yjgN gene in LS5218
4498173–4499513b	IS2	No insert	Insertion in MG1655 between pseudogenes in KpLE2
F Plasmidb	No	Yes

Similar position but different from reported mutation in NCM3722 (Lyons et al., 2011).

, Mutation also reported for NCM3722 (Lyons et al., 2011).

Table 2

Table of SNPs and indels between MG1655 and LS5218.

Location	Gene	Type	CDNA change	AA change
280113	insX	CDS	AAGCTG→GGCTA	Lys82fs
1101543	csgG	CDS	A→T	Lys48b
1330578	yciN	CDS	∆G	Ile31fs
2173360	gatC	CDS	∆GG	Val306fs
2210942	yehQ	CDS	T→G	b615Glu
2278174	yejG	CDS	Insert CTGCTGGT	Phe22fs
2665747	csiE	CDS	C→T	Gln105b
2867455	arpoS	CDS	C→T	Gln33b
3130140	yghO	CDS	A→T	Lys2b
3473612	arpsG	CDS	T→A	Leu157b
3560455	glpR	CDS	Insert C	His51fs
3662700	amdtF	CDS	C→T	Gln763b
3815879	arph	CDS	Insert G	Glu224fs
3951535	ailvG	CDS	Insert AT	Gln327fs
290103	argF	CDS	T→A	Phe68Tyr
290174	argF	CDS	TACAGAAGCTTACC→AAGCCAAACTCACT	ValGln40GluAla
290192	argF	CDS	ATGGCAAG→GCGGTAAA	Asn36Ser
290221	argF	CDS	AC→GA	Gln28Lys
378700	afrmA	CDS	T→G	Val291Gly
579285	ybcV	CDS	A→G	Ile104Val
616676	entF	CDS	C→A	Asp840Glu
903248	aartP	CDS	C→A	Leu163Met
1169836	ldtC	CDS	T→C	Leu180Pro
1235101	fadR	CDS	T→A	Leu55Gln
1301992	aoppA	CDS	A→T	Asn271Tyr
1301999	aoppA	CDS	G→A	Ser273Asn
1302190	aoppA	CDS	A→G	Asn337Asp
1305442	aoppD	CDS	T→G	Val230Gly
1306736	aoppF	CDS	T→G	Ser325Ala
1337394	aacnA	CDS	A→G	Ser522Gly
1358859	apuuP	CDS	A→G	Tyr110Cys
1643679	aydfU	CDS	T→A	Leu209Gln
1652331	aintQ	CDS	T→C	Phe261Leu
1894839	apabB	CDS	T→C	Leu12Pro
2003346	afliC	CDS	C→A	Asn87Lys
2040433	ayedY	CDS	C→A	Ala319Asp
2322251	atoC	CDS	T→G	Ile129Ser
3035546	prfB	CDS	A→G	Thr246Ala
3214757	arpoD	CDS	T→C	Tyr571His
3300572	yhbS	CDS	G→A	Asp63Asn
3388041	aaaeB	CDS	A→C	Thr50Pro
3554135	amalT	CDS	T→A	Trp351Arg
3725176	aglyQ	CDS	A→C	Glu48Ala
4243857	amalF	CDS	G→T	Gly407Cys
4300405	mdtP	CDS	A→T	Gln209Leu
4342047	amelA	CDS	T→A	Leu46Gln
289241	yagI	Upstream	C→A (−79)
289281	yagI	Upstream	TTGG→CTGA (−119)
579146	nmpC	Upstream	T→C (−2321)
579651	nmpC	Upstream	G→A (−2826)
579671	nmpC	Upstream	A→G (−2846)
579717	nmpC	Upstream	T→G (−2892)
579811	nmpC	Upstream	G→A (−2986)
687852	hscC	Upstream	C→A (−4459)
696470	ybeX	Upstream	G→A (−4686)
1299464	insZ	Upstream	A→C (−4142)
1665170	clcB	Upstream	A→C (−145)
1979271	cheA	Upstream	ATG→TTT (−3947)
2118488	wcaN	Upstream	G→A (−4161)
2118495	wcaN	Upstream	C→A (−4168)
2118501	wcaN	Upstream	TGTGCTCGGGTCTT→AGGTCC (−4175)
2118526	wcaN	Upstream	T→A (−4199)
2118560	wcaN	Upstream	Insert T (−4233)
2118599	wcaN	Upstream	TGTGCTCGGGACC→GCGTACAGATT (−4272)
2118649	wcaN	Upstream	C→T (−4322)
2725818	akgtP	Upstream	T→C (−72)
3707947	dppD	Upstream	G→T (−4099)
4035734	fadB	Upstream	A→C (−4763)
4166470	trmA	Upstream	G→A (−3200)
4223638	arpA	Upstream	A→G (−1151)
4296380	nrfD	Downstream	Insert CG (4948)
4510238	yjhD	Upstream	A→C (−3382)
4542681	nanM	Upstream	∆A (−3917)

Fs, frameshift.

Mutation also reported for NCM3722 (Lyons et al., 2011).

, stop codon.

The L55Q mutation in fadR replaces a hydrophobic leucine with a hydrophilic glutamine within the DNA binding domain. This change likely affects the interaction of fadR with the DNA backbone (van Aalten et al., 2000, Xu et al., 2001). The atoC mutation, I129S, is responsible for conferring constitutive expression of the ato operon, however the mechanism of this action remains unknown. Beyond the expected mutations, the major insertions and deletions were concentrated around insertion elements whereas the small SNPs were distributed evenly throughout the genome. Coverage of the LS5218 sequence compared to MG1655 (Fig. 2) highlights the position of known insertion elements in MG1655 for comparison of the large and small differences along with the assembled contigs.

Fig. 2

Circular plot of LS5218 features mapped to MG1655. The outer purple histogram displays the coverage of LS5218 sequencing reads as mapped to MG1655. Average coverage was 200× with breaks displayed at genomic regions that differ between the strains. The contigs generated from the LS5218 de novo assembly are blue. Many of these breaks correspond to locations of native MG1655 insertion sequences (green bars). The large insertions and deletions of LS5218 are displayed in red and labeled. SNPs are spread throughout with those in coding regions shown in orange and those upstream of genes shown in light blue. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

We found that LS5218 has numerous insertions, deletions, genomic arrangements, and SNPs as well as the presence of the F plasmid. The highlighted 17 large insertions and deletions as well as the 74 SNPs could affect gene expression beyond the anticipated changes in fatty acid degradation pathways. One of the primary differences between MG1655 and LS5218 is the rph mutation. It is known that MG1655 has a frameshift mutation in rph that also causes pyrimidine starvation due to polar effects on the downstream pyrE gene (Blattner et al., 1997b). MG1655 also has a known frameshift in ilvG that affects expression of a branched-chain amino acid biosynthesis operon (Lawther et al., 1982), which is not present in LS5218. The fact that LS5218 does not have these mutations may partially explain why we see increased growth rates in LS5218 when compared with MG1655 on MOPS minimal media (Fig. 3).

Fig. 3

E. coli MG1655 and LS5218 growth rate in MOPS minimal media with glucose or glucose supplemented with 20 μg/mL uracil.

During our evaluation of large genomic changes we found a 3.5 kb insert in LS5218 containing three genes putatively annotated for fatty acid and secondary metabolite biosynthesis. These genes have homology towards an acyl-carrier protein (B1R43_RS14595), an aldehyde/flavonoid reductase with an NAD(P) binding site (B1R43_RS14600) and a fatty acyl-AMP ligase (B1R43_14605). None of these enzymes have been studied, but their putative annotations suggest that they could augment fatty acid metabolism with new or enhanced enzymes. Among the small changes compared to MG1655, mutations in rpoS (sigma-28) and rpoD (sigma-70) could have large pleiotropic effects on the cell. LS5218 also has a mutation in prfB (release factor 2), similar to that of E. coli BL21. These mutations could explain the differences in gene expression (identified by microarray) between MG1655 and NCM3722, a close LS5218 relative (Soupene et al., 2003). This study showed significantly higher mRNA expression of flagella and chemotaxis and lower expression of galactitol and maltose operon and regulons (Soupene et al., 2003).

Conclusions

The genome sequence of E. coli LS5218 disproves a commonly held assumption about its relationship to the reference K12 strain. LS5218 is a close relative of NCM3722 and not MG1655, with many of the genomic differences reported here also seen in a comparison of the E. coli strains MG1655 and NCM3722 (Table 1, Table 2). Unfortunately the strain history for NCM3722 was lost (Lyons et al., 2011, Soupene et al., 2003) so we do not know if they are directly related. We theorize, based off phylogeny and common genetic variations, that NCM3722 and LS5218 share a similar derivation path and are a better representation of the original E. coli K-12 isolate than MG1655 (Soupene et al., 2003). The sequence of E. coli LS5218 allows us to have a better understanding of the genetic background for this widely used production strain and raises the question whether other mutations, in addition to fadR601 and atoC(c), could be contributing to the improved production rates compared to other E. coli derivatives (Salamanca-cardona et al., 2014, Tappel et al., 2012, Ushimaru et al., 2015). The additional overlooked differences between LS5218 and MG1655 highlight the fact that genetic background is an important feature when selecting a host for metabolic engineering. The choice may have profound effects on successful engineering and strain performance.

Materials and methods

DNA was isolated from LS5218 using the Wizard® Genomic DNA Purification Kit (Promega) and sequenced by the University of Wisconsin Biotechnology Center. A paired end library was run on an Illuminia Hi-Seq. 2500. Sequencing generated 5431,968 reads (2 × 250). A de novo assembly was created using SPAdes (Bankevich et al., 2012). The draft genome contained 121 contigs (200 bp or greater) with an N50 of 159,470. The genome length was 4699,198 with an average coverage of 279X. The assembly included the complete F plasmid (67,502 bp) and bacteriophage phiX174 (5513 bp). The draft genome was annotated through the NCBI Prokaryotic Genome Annotation Pipeline (PGAP). The genome sequence has been deposited in GenBank under bioproject PRJNA374891 and accession number MVJG00000000. Reads have been deposited to the Sequence Read Archive with accession number SRR5572609.

Sequencing reads (as FASTQ files) of E. coli LS5218 were mapped to completed reference genomes E. coli K12 MG1655 (GCA_000005845.2) and E. coli NCM3722 (GCF_001043215.1) using Bowtie2 using the “fast-local” setting (Langmead and Salzberg, 2012). The output sequence alignment map (SAM) file was converted to a binary alignment map (BAM) file and sorted using SAMtools (Li et al., 2009). Variants were then called using FreeBayes (Garrison and Marth, 2012) and Naïve Variant Caller (Galaxy open source bioinformatics tool) (Goto et al., 2011). Variant calls were then annotated using SnpEff (Cingolani et al., 2012b) and variant calls with a quality of less than 30 were sorted out using SnpSift (Cingolani et al., 2012a). Large gaps and insertions were isolated using progressive Mauve alignment with default settings (Darling et al., 2004, Darling et al., 2010) and the pan-genome for the E. coli K-12 strains was generated with BPGA (default settings) (Chaudhari et al., 2016).

Specific growth rates calculated from growth curves generated in MOPS minimal media (Neidhardt et al., 1974) supplemented with 0.2 wt% glucose and 20 μg/mL uracil, when indicated. OD600 measurements were taken at 30 min intervals by a Tecan m200.