high-efficiency scarless genetic modification in ...€¦ · 14/4/2014 · 21 genetic...
TRANSCRIPT
1
1
High-efficiency Scarless Genetic Modification in 2
Escherichia coli Using Lambda-Red Recombination and I-SceI 3
Cleavage 4
5
Method for Bacterial Genetic Modification 6
7
Junjie Yang1,2, Bingbing Sun1,2, He Huang1,2, Yu Jiang1,2, Liuyang Diao1,2, Biao Chen1,2, 8
Chongmao Xu1,2, Xin Wang1,3, Jinle Liu1,3, Weihong Jiang1, Sheng Yang1,2# 9
10
1 Key Laboratory of Synthetic Biology, Institute of Plant Physiology and Ecology, Shanghai 11
Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai 200032, China 12
2 Shanghai Research Center of Industrial Biotechnology, Shanghai 201201 , China 13
3 University of the Chinese Academy of Sciences, Beijing 100080, China 14
15
16
# Correspondence to: Sheng Yang, 300 Fenglin Road, Shanghai 200032, China; Tel: 17
+86-21-54924173; Fax: +86-21-54924015; Email: [email protected]; 18
19
AEM Accepts, published online ahead of print on 18 April 2014Appl. Environ. Microbiol. doi:10.1128/AEM.00313-14Copyright © 2014, American Society for Microbiology. All Rights Reserved.
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
2
ABSTRACT 20
Genetic modifications of bacterial chromosomes are important for both fundamental and 21
applied research. Here, we developed an efficient, easy-to-use system for genetic modification 22
of the Escherichia coli chromosome. The system uses two plasmids, lambda (λ)-Red 23
recombination and I-SceI cleavage. An intermediate strain is generated by integration of a 24
resistance marker gene(s) and I-SceI recognition sites in or near the target gene locus using 25
λ-Red PCR targeting. The intermediate strain is transformed with a donor plasmid carrying the 26
target gene fragment with the desired modification flanked by I-SceI recognition sites, together 27
with a bifunctional helper plasmid for λ-Red recombination and I-SceI endonuclease. I-SceI 28
cleavage of the chromosome and donor plasmid allows λ-Red recombination between 29
chromosomal breaks and linear double strand DNA from the donor plasmid. Genetic 30
modifications are introduced into the chromosome and the placement of the I-SceI sites 31
determines the nature of the recombination and the modification. This method was 32
successfully used for cadA knock-out, gdhA knock-in, pepD seamless deletion, site-directed 33
mutagenesis of the essential metK gene, and replacement of metK with Rickettsia 34
S-adenosylmethionine transporter gene. This effective method can be used with both essential 35
and nonessential genes modification, and will benefit basic and applied genetic research. 36
Key words: I-SceI endonuclease; lambda Red recombination; genetic modification; 37
markerless; Escherichia coli 38
39
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
3
40
INTRODUCTION 41
Lambda (λ)-Red recombination, which triggers recombination between a bacterial 42
chromosome and foreign, linear double-stranded DNA (dsDNA), is a common technique in 43
Escherichia coli for genetic modification (1-4). The λ-Red recombination method relies on 44
three proteins: 5′→3′ exonuclease exo, the single-stranded DNA (ssDNA)-binding protein beta, 45
and the degradation inhibiting protein gam (4). The λ-Red recombination replaces a specific 46
chromosomal sequence with a selectable antibiotic resistance gene flanked by homologous 47
arms. Homologous recombination is efficient for linear DNA with homologous arms of 48
approximate 40 or more bp (2, 3). For markerless genetic modification, the antibiotic 49
resistance cassette integrated into the chromosome must be removed. Common methods for 50
markerless genetic modification are based on the Cre-loxP or Flp-FRT systems (3, 5). These 51
methods leave a single loxP or FRT scar sequence at the chromosomal locus, which can have 52
undesirable consequences such as native-locus gene replacement or mutagenesis. If a 53
second modification site is close to a former site, recombination can occur between the loxP or 54
FRT sequences. 55
Counterselection is effective for generating genetic modifications without leaving 56
selectable markers or scars. Published protocols are usually based on two successive rounds 57
of recombination: (1) integration of a positively selectable marker (e.g., an antibiotic resistance 58
gene), and (2) selection for marker loss (counterselection). Counterselectable markers include 59
Bacillus subtilis SacB (6), E. coli galactokinase (galK) (7) and thymidylate synthase A (thyA) (8, 60
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
4
9), streptomycin resistance (10) and the fusaric acid sensitivity system (11). These methods 61
have several disadvantages including: (1) some counterselectable markers, such as SacB, 62
must be combined with a positively selectable marker, which increases cassette length; and (2) 63
some counterselectable markers as galK or thyA can be used only in host strains with a 64
particular auxotrophy. 65
The yeast Saccharomyces cerevisiae endonuclease I-SceI is a novel tool for unmarked 66
chromosomal manipulation. I-SceI is an intron-encoded endonuclease with an unusually long 67
recognition sequence of 18 bp (12, 13). Bacterial genomes usually do not have natural I-SceI 68
recognition sites. If an I-SceI recognition site is introduced into a bacterial genome, the 69
expression of I-SceI endonuclease induces lethal double-strand breaks (DSBs) within the 70
genome, inhibiting DNA propagation and cell growth. Several methods for site-directed 71
mutagenesis of BACs or bacterial genomes that use I-SceI expression are reported (14-17). 72
These methods are based on two successive rounds of recombination, similar to other 73
counterselection systems. The approaches can be categorized into two types (18). The first 74
type requires integration of a duplicated sequence with the I-SceI recognition site sequences 75
in a first round of recombination. The duplicated sequence is as a substrate for recombination 76
in the second round (16, 17, 19). This type of approach is efficient, but requires complex 77
design and construction. The second type of approach includes two rounds of separated 78
recombination. Blank, et al. reported a method for scarless mutagenesis of the Salmonella 79
enterica chromosome (20). Resistance marker gene and the I-SceI recognition site(s) are 80
integrated into genome, and then dsDNA fragment with modification replaces the resistance 81
marker gene. Kuhlman and Cox reported another method that integrates large fragments into 82
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
5
the E coli genome (14, 15). Resistance gene flanked by I-SceI recognition sites and 25bp 83
exogenous sequences for “landing pads” are inserted into the chromosome. Then a fragment 84
carrying insertion is excised by I-SceI from a donor plasmid and incorporated into the genome 85
via recombination at the landing pads (14, 15). 86
Another technique, “Gene Gorging”, also uses two plasmids, λ-Red recombination and 87
I-SceI cleavage. I-SceI cleavage of the donor plasmid creates a linear dsDNA substrate, which 88
would replace the wild type chromosomal allele. "Gene Gorging" frequencies for precise 89
mutations in the E. coli genome are not sufficiently high, only 1-15%, limiting its use (21). 90
In this study, a new, two-plasmid method for unmarked genetic modification of E. coli was 91
established. The method uses λ-Red recombination and I-SceI cleavage. E. coli endogenous 92
sequences were used as substrates for recombination and no scar was introduced into 93
genome. The new method was successfully used for gene deletion and insertion, seamless 94
gene deletion, site-directed mutagenesis, and gene replacement. The new method could be 95
used to modify both nonessential and essential genes. 96
97
MATERIALS AND METHODS 98
Strains, plasmids and growth conditions 99
Strains and plasmids used in this study are in Table 1. E. coli strains MG1655 and BL21 100
(DE3) were used for genetic modification experiments; strain DH5� was used for recombinant 101
DNA manipulations. All E. coli strains were cultured in LB (Luria-Bertani) medium [1% tryptone 102
(Oxoid or Angel Yeast), 0.5% yeast extract (Oxoid or Angel Yeast), 1% NaCl] supplemented 103
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
6
with 100 mg/L ampicillin, 50 mg/L kanamycin, 100 mg/L spectinomycin, 100mg/L apramycin or 104
1 mmol/L S-adenosylmethionine (SAM), as required. 105
Oligonucleotides used are in Table S1. PCR was performed using Taq (Fermentas), 106
KOD-plus-neo or KOD-FX (Toyobo) polymerase. I-SceI endonuclease and restriction enzymes 107
were purchased from Takara or Fermentas. 108
Modular plasmids for donor and templates 109
A 1.5 kb NotI fragment containing the kanamycin -resistance gene kanMX from pUG6 (22) 110
was cloned into pBluescript II KS (-) (23) yielding pKS-K. A 1.7 kb fragment containing the 111
kanMX gene was amplified from pKS-K using primer pairs T3_IsceI/T7_IsceI and digested 112
with BssHII. The resulting fragment was used to replace a 0.2 kb BssHII region containing the 113
multiple cloning sites (MCS) of pBluescript II KS (-). The resulting ampicillin-resistance and 114
kanamycin-resistance plasmids were pKSKI-1 and pKSKI-2 and differed in insert direction. 115
pKSKI-2 was digested with NotI to excise kanMX and the plasmid backbone was recircularized 116
with T4 DNA ligase, yielding pKSI-1. 117
Apramycin-resistance or spectinomycin-resistance genes on 1.5 kb fragments were 118
amplified from pIJ773 and pIJ778 (24), and T-A cloned into pMD18-T simple (Takara) for 119
plasmids pMDIAI and pMDISI. 120
Bifunctional helper plasmids 121
The rhaB promoter fragment and the I-SceI encoding region fragment from pUC19RP12 122
(19) were joined by overlapping PCR and cloned into pKD46 (3) at the NcoI site for plasmid 123
pREDIA, which was similar to previous reported pREDI plasmids (16) with some differences in 124
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
7
the I-SceI-coding region. 125
The ampicillin-resistance gene of helper plasmids was replaced with the 126
kanamycin-resistance gene from pPIC3.5K (Invitrogen). Bifunctional helper plasmids 127
pREDTAI and pREDTKI with the trc promoter were constructed using the process used for the 128
rhaB promoter. The trc promoter was amplified from pTRC99a (25). (See Supplementary 129
Material for detail process of plasmid construction, available online) 130
Modifications in nonessential genes 131
A linear fragment with resistance genes flanked by I-SceI recognition sites was introduced 132
by electroporation into MG1655 using conventional Red-mediated recombination (3) to obtain 133
an intermediate strain. A bifunctional helper plasmid or the Red helper plasmid pKD46 were 134
used for this step. Transformants were tested by colony PCR to confirm integration of the 135
resistance gene and I-SceI recognition sites. After transformation, the bifunctional helper 136
plasmid remained for further recombination and the conventional Red helper plasmid pKD46 137
was cured by 42°C growth (3). 138
Donor plasmid based on the pKSI-1 vector or conventional T-vectors (Takara) harboring 139
modifications were constructed by traditional restriction cloning, overlapping PCR, or Gibson 140
assembly (26). (See Supplementary Material for detail process of donor plasmid construction, 141
available online) Donor plasmids were transformed into intemediate strains; or donor 142
plasmids and bifunctional helper plasmids were co-transformed into helper-plasmid-free 143
intermediate strains. 144
Resulting colonies were incubated in LB medium with 0.5% glucose and 50 μg/mL 145
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
8
ampicillin and kanamycin (selection for both the donor and the helper plasmid) for 6–8 hours or 146
overnight; 40 µl seed culture was used to inoculate a test tube with 4 mL LB with 10 mmol/L 147
L-arabinose and 50 μg/mL ampicillin or kanamycin (selection for the helper plasmid only). After 148
30 ºC and 200 rpm for ~2 h, L-rhamnose or IPTG (isopropyl β-D-1-thiogalactopyranoside) was 149
added to 20 mmol/L with cultivation at 30 ºC for 6–8 hours or overnight. From this culture, 40 µl 150
were used to inoculate 4 mL LB medium with 20 mmol/L L-rhamnose or IPTG, 10 mmol/L 151
L-arabinose, and 50 μg/mL ampicillin or kanamycin (selection for the helper plasmid only), 152
which was cultivated at 30 ºC for 6–8 hours or overnight. Cells were harvested, suspended in 153
sterile water, diluted (10-1 to 10-4) and spread on LB plates with 20 mmol/L L-rhamnose or IPTG, 154
10 mmol/L L-arabinose and 50 μg/mL ampicillin or kanamycin (selection for the helper plasmid 155
only). Controls were LB plates with apramycin or spectinomycin, L-rhamnose or IPTG, 156
L-arabinose, and ampicillin or kanamycin spread at the same time. (Alternatively, after the first 157
~2 h L-arabinose inducing Red enzymes, arabinose-free medium with only L-rhamnose or 158
IPTG for inducing I-SceI and antibiotic for maintaining helper plasmid could be used, to avoid 159
the potential deleterious effects caused by the prolonged expression of the lambda Red 160
functions; (27) but a slightly decreased efficiency would be obtained. ) After overnight at 30 ºC, 161
colonies were analyzed for apramycin or spectinomycin sensitivity and final clones were 162
confirmed by colony PCR. (See Supplementary Material for a step-by-step protocol, available 163
online) 164
Modifications in metK 165
Using Red-mediated recombination(3), an apramycin resistance gene and I-SceI 166
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
9
recognition sites (IsceI-aprR-IsceI cassette) was inserted into the yqgC site, and a 167
spectinomycin resistance gene and I-SceI recognition sites (IsceI-spcR-IsceI cassette) was 168
inserted into galP site, by successive electroporation steps for strain A10. yqgC and galP are 169
nonessential genes next to the essential metK gene. 170
A 2.5 kb PCR fragment containing part of the speA gene, full-length yqgB-yqgC-metK 171
genes and part of the galP gene was cloned into pKSI-1 for pKSI-metK. Site-directed 172
mutagenesis of metK followed the protocol in QuikChange Site-Directed Mutagenesis kits 173
(Agilent) to obtain the donor plasmid pKSI-metKmut (28). A 1 kb fragment with the gene for the 174
SAM transporter (29, 30) was synthesized and replaced the metK coding region in pKSI-metK, 175
yielding pKSImetKSamT. 176
Mutagenesis or replacement of metK was the same as for the nonessential genes, with 1 177
mmol/L SAM added to LB medium if required. For replacing metK gene with the Rickettsia 178
SAM transporter gene, clones were also analyzed for SAM auxotrophy. 179
180
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
10
181
RESULTS 182
Scarless modification method strategy 183
This method included a conventional initial step and a final recombination step (Figure 184
1AB). In the initial step, a resistance gene flanked by I-SceI recognition sites was inserted into 185
the chromosome by conventional λ-Red PCR targeting (Figure 1A); meanwhile, a fragment 186
carrying the desired modification was inserted into donor plasmid with I-SceI recognition sites. 187
In the final recombination step, expression of I-SceI endonuclease induces DSBs within both 188
the genome and the plasmid. The fragment with modification is excised from a donor plasmid, 189
and then incorporated into the genome at the DSB locus via λ-Red recombination. (Figure 1B) 190
Three groups of plasmids (bifunctional helper plasmid, templates plasmid and modular 191
donor plasmid) were constructed (Table 1, Figure 2) for performing this method. The plasmids 192
have been distributed to Addgene (http://www.addgene.org/) with IDs 51625~51628, 193
51652~51655 and 51725 . 194
The method could be initiated by transformation of a host strain with a bifunctional helper 195
plasmid (Figure 1A)., which were constructed from the λ-Red helper plasmid pKD46 (3). The 196
λ-Red genes gam, bet and exo were driven by the araBAD promoter and inducible with 197
L-arabinose (Figure 2A). These genes mediated recombination between target regions and 198
homologous arms. Plasmids with the I-SceI gene under the control of the trc or rhaB promoter 199
were inducible with IPTG or L-rhamnose. The temperature sensitivity of the pSC101 200
replication origin maintained plasmids at low copy number and for easy curing by growth at 201
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
11
42°C. Two versions of the helper plasmids were differentiated by use of either the kanamycin 202
or ampicillin resistance markers. This design made the plasmids compatible with different 203
donor plasmids (Figure 2A). 204
Plasmids pMDIAI and pMDISI were used as PCR templates to amplify 205
apramycin-resistance or spectinomycin-resistance genes flanked by I-SceI recognition sites. 206
(Figure 2C). The small construct sizes allowed simple and reliable integration of resistance 207
marker genes into any location using established λ-Red PCR-targeting recombination 208
methods.(3) (Figure 1A). 209
To facilitate donor plasmid construction, several modular vectors were constructed based 210
on the high-copy vector pBluescript II KS (-) (23). Plasmid pKSI-1 consisted of a 211
pBluescript II KS (-) backbone, an MCS, and two I-SceI recognition sites (Figure 2B). The 212
MCS could be used for further subcloning steps. Plasmids pKSKI-1 and pKSKI-2 were similar, 213
with a kanamycin- resistance gene kanMX in the MCS. Alternatively, donor plasmids could be 214
constructed with common high-copy T-vectors such as pMD19-T simple and pBackZero-T 215
(Takara). In this case, I-SceI recognition sites were introduced by a PCR step using longer 216
primers. Donor plasmids based on pKSI-1 or common T-vectors showed high recombination 217
frequencies (Table 2). 218
Markerless knockout and knock-in of selected genes 219
knock-out of cadA 220
An apramycin-resistance gene cassette flanked by I-SceI recognition sites was integrated 221
into the cadA locus by λ-Red PCR targeting to obtain an intermediate strain. Positive clones 222
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
12
were screened by PCR (Figure 3A). PCR products were digested with I-SceI, to check for 223
functionality of the inserted I-SceI restriction sites (data not shown). 224
To construct a donor plasmid, truncated cadA segments with I-SceI recognition sites was 225
cloned into a pBackZero-T vector. The donor plasmid was transformed into the intermediate 226
strain for the final recombination step. In the recombination step, λ-Red enzymes, then I-SceI 227
were induced, leading to cleavage of both the donor plasmid and the chromosome. Once 228
DSBs were generated by I-SceI, recombination between the resulting donor fragment and the 229
DSB chromosomal locus was mediated by λ-Red enzymes (Figure 1B). 230
After growth in IPTG or L-rhamnose to induce I-SceI expression with L-arabinose to induce 231
expression of λ-Red enzymes, most surviving cells that were apramycin sensitive were 232
confirmed as having the desired deletion. The deletion was confirmed by PCR, restriction 233
endonuclease analysis and sequencing (Figure S2A). 234
Meanwhile, in the negative control experiments without donor plasmid, the frequency of 235
apramycin sensitivity of the surviving cells was lower than 5% of the total, indicating the 236
knock-out of cadA would occur by repair with the donor DNA, not by non-homologous 237
end-joining of the double stranded breaks. 238
Knock-in of gdhA 239
Results for knock-in of the gdhA gene were similar to results for knocking out cadA. In the 240
recombination step, I-SceI were induced, leading to cleavage in the chromosome at gntT gene 241
locus and evict spectinomycin-resistance gene. Once DSBs were generated at gntT locus, 242
recombination happened between the donor fragment with gdhA and the DSB chromosomal 243
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
13
locus, mediated by λ-Red enzymes. After recombination, the gdhA gene fragment replaced the 244
spectinomycin-resistance gene. The resulting strain had the desired modification of gdhA 245
integrated into the chromosome (Figure S2B). 246
Multiple modifications 247
We also inserted the gdhA gene into the chromosome of a cadA-disrupted E. coli strain. 248
Similar modification efficiency was obtained (data not shown). The bifunctional helper plasmid 249
for cadA disruption was retained, and directly used for the PCR targeting experiments to 250
integrate spectinomycin-resistance gene into gntT locus. These results demonstrate the 251
protocol could be used repeatedly, to make multiple modifications in a strain. 252
253
Seamless deletion of pepD 254
To determine if the new method could be used for seamless deletion of a selected gene, 255
we deleted the pepD gene following a procedure similar to the process described above. An 256
apramycin-resistance gene cassette flanked by I-SceI recognition sites was integrated into the 257
pepD locus, replacing 600 bp in the pepD ORF (Figure 3A). 258
To construct a donor plasmid, upstream and downstream fragments for the pepD gene 259
were joined and subcloned into pKSI-I. The fusion fragment was suitable for seamless deletion 260
of pepD (Figure 3B). The apramycin-resistance gene fragment and remaining 1.0 kb pepD 261
fragments were replaced by “nothing” (Figure 3B). The 1.6 kb pepD ORF was seamlessly 262
deleted from the ATG start codon to the TAA terminator codon (Figure 3C). 263
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
14
Modifications in the essential metK gene 264
To use the new method to modify an essential gene locus, a novel strategy for an 265
essential gene was designed. Two pairs of I-SceI sites with 2 different resistance genes were 266
inserted into the loci next to the target gene by consecutive λ-Red PCR targeting. Meanwhile, 267
donor plasmid with fragment carrying the desired modification was constructed as same 268
process for the non essential genes. In the final recombination step, the wild-type target gene 269
and the two resistance genes were removed from the chromosome. Then the fragment with 270
modification is excised by I-SceI from a donor plasmid and incorporated into the genome at the 271
gene locus via recombination. (Figure 4) 272
To determine if the new method could be used to modify an essential gene locus, we used 273
the method to modify the metK gene. E. coli metK is essential because it produces the only 274
enzyme for synthesis of SAM, which is involved in many metabolic reactions, and E. coli 275
cannot take up SAM from medium(31). As an essential gene, metK could not be directly 276
deleted, but mutations can be made in its coding region. The metK gene can be replaced by a 277
Rickettsia SAM transporter gene and the resulting cells take up extracellular SAM and survive 278
with SAM in the medium (29, 30). 279
Site-directed mutations in metK 280
The intermediate strain A10 was constructed with two resistance genes with 2 pairs of 281
I-SceI sites inserted into the loci next to metK (Figure 4A). A donor plasmid was constructed 282
with mutagenesis fragments containing metK sequences with three synonymous mutations. 283
In the recombination step, I-SceI cut at the 2 pairs of I-SceI sites next to metK, and the 284
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
15
wild-type metK gene and the two resistance genes were removed from the chromosome. 285
Subsequently, the mutation fragment from the donor plasmid replaced the wild-type fragment 286
via homologous recombination, introducing the mutations into the metK ORF without scar 287
fragments (Figure 4B). Mutations were confirmed by PCR, restriction endonuclease analysis 288
and sequencing (Figure 4C). 289
Replacement of metK with Rickettsia SAM transporter 290
A donor plasmid containing the SAM transporter coding region, the metK promoter and 291
the metK terminator was constructed. The plasmid was transformed into the A10 intermediate 292
strain. Using the same procedure, the SAM transporter fragment was used to replace wild-type 293
metK via homologous recombination (Figure 4 B,C). The resulting strains had SAM 294
auxotrophy and grew in medium supplemented with SAM (Figure S1). 295
Curing plasmids 296
Donor plasmids are efficienty cured by I-SceI cleavage (14). After recombination, at least 297
half of colonies with the desired genotype were sensitive to ampicillin or kanamycin (Table 2), 298
showing efficient curing of donor plasmids. For the colonies with the desired genotype and the 299
donor plasmid remaining, prolonged culture in IPTG or L-rhamnose medium to induce I-SceI 300
expression resulted in donor plasmid curing (data not shown). Helper plasmids can be cured 301
by 42°C growth (3). 302
303
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
16
Method efficiency 304
Preparatory steps followed an established λ-Red PCR targeting method(3). Integration 305
frequencies (colonies with expected PCR bands per tested spectinomycin/apramycin resistant 306
colonies) were greater than 50%, depending on the size of the homologous regions and the 307
targeted genomic locus (data not shown). 308
For recombination for modification and dominant marker recycling, IPTG induction of the 309
trc promoter to stimulate I-SceI expression gave higher recombination frequencies and donor 310
plasmid curing frequencies than rhaB promoter. (Table 2). 311
312
313
314
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
17
315
DISCUSSION 316
The scarless genetic modification strategy described here was based on a simple 317
two-plasmid procedure using λ-Red recombination and endonuclease I-SceI cleavage. A 318
fragment with I-SceI recognition sites was inserted into the target site using λ-Red 319
recombination and positively selected by conventional λ-Red PCR targeting (2, 3). I-SceI 320
cleaved both the donor plasmid and the chromosome. Incorporation of the insertion fragment 321
was enhanced by the expression of λ-Red enzymes and led to integration of the donor 322
fragment and scarless genetic modification (Figure 1). Generally, the strategy took about 1.5-2 323
weeks, with donor plasmid construction and λ-Red PCR targeting procedure performed 324
simultaneously. We used this strategy in both E. coli K-12 strain MG1655 and the B strain 325
BL21(DE3). This strategy should be applicable in any bacterial strains that can be modified 326
with λ-Red or other PCR targeting recombination systems. As endonuclease I-SceI directly 327
cleavage at the chromosome, this method would be applicable for host strains with any 328
genotypes, including those with genotypes not suitable for published counterselectable 329
markers. 330
Our method use two separated rounds of recombination, and this approach is also used 331
by some similar previously reported methods (14, 15, 20). This approach avoids the 332
integration of a duplicated sequence in a first recombination to be a substrate for the second 333
recombination (16, 17) and gives flexibility in designing and performing experiments. As shown 334
by our pepD seamless deletion, in the first PCR-targeting step, short arms homologous to 335
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
18
pepD ORF were used for the first recombination, deleting a pepD coding region of 336
approximately 600 bp by replacement with a resistance marker gene and leaving pepD coding 337
region of approximately 1 kb. Subsequently, fragments homologous to upstream and 338
downstream regions of pepD were used for the second recombination, resulting in complete 339
deletion of the truncated pepD segments and the resistance genes from the start codon to the 340
terminator codon (Figure 3). 341
Our method shows some similarity to a method described previously by Kuhlman and Cox 342
for integrating large fragments into the E coli genome (14, 15). Kuhlman and Cox‘s method 343
requires integration of 25bp exogenous sequences into the genome to generate a “landing 344
pad”, so it is not a seamless method. Our method uses the endogenous fragment for 345
recombination in both initial and the final recombination steps. No redundant sequences are 346
introduced into the genome, so seamless modification is possible. 347
Our method could be used for both nonessential and essential genes. Of the total 4288 E. 348
coli genes, 3985 are nonessential and 303 genes are essential candidates including 37 with 349
unknown function (1). To modify nonessential genes, I-SceI recognition sites and resistance 350
gene could be inserted directly into the gene (Figure 1). For modifying essential genes, the 351
I-SceI recognition site and resistance genes could be inserted upstream and downstream of 352
the essential gene (Figure 4). In both cases, subsequent recombination was accurate as 353
expected. 354
As previously reported, in vivo cleavage of donor plasmids by I-SceI expression has 355
advantages over introduction of linear DNA fragments by direct transformation(14). Repair and 356
editing mechanisms during chromosomal and plasmid replication result in high fidelity. The 357
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
19
higher transformation efficiencies of supercoiled plasmids facilitates the experiments. High 358
plasmid copy numbers are important for maintaining a high concentration of intracellular 359
fragments (14). The gene gorging method is reported to result in chromosomal modification 360
frequencies as high as 1–15%, even without cutting chromosomes (21). 361
I-SceI expression must be considered. In our experiments, using the rhaB promoter for 362
I-SceI expression resulted in about a half of the colonies with the resistant phenotype. PCR 363
confirmed that no mutations occurred within or next to the resistance genes and the integration 364
site (data not shown). The trc promoter gave a higher efficiency for fragment replacement and 365
donor plasmid curing, probably because of differences in expression levels from the different 366
promoters (32). 367
Our method could be a simple and easy way to move mutations from one strain into 368
another. For example, generating a strain with a desired phenotype generated by random 369
mutagenesis or evolution require introducing every mutation into a wild-type strain to analyze 370
the effect on function (33). Such mutations can be identified by high-throughput genome 371
sequencing and confirmed by PCR and dideoxy chain-termination sequencing. The pKSKI 372
series could be made into T-vectors by simple digestion (34-36), so donor plasmids could be 373
constructed by direct T-A cloning. A resistance gene cassette and I-SceI recognition sites 374
could be integrated into the gene locus of the recipient strain by λ-Red PCR targeting. The 375
fragment from the donor plasmid harboring the mutation could be integrated into the genome 376
of recipient strains to replace the wild type allele. 377
Another advantage of our new method is that different modifications could be performed 378
on a single intermediate strain. By using different donor plasmid, a target gene of interest could 379
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
20
be modified differently. Any methods could be used to construct the donor plasmid, even by 380
artificial whole gene synthesis. The only requirement is inclusion of upstream and downstream 381
homologous regions and I-SceI recognition sites. According to our experiments, homologous 382
regions in 300~500bp length could benefit both plasmid construction and recombination 383
efficiency. The designed sequence on the donor plasmid will be the exact sequence introduced 384
in the final strains. 385
The system could potentially be improved. For making multiple modifications, a different 386
set of resistance marker genes flanked by I-SceI recognition sites could be integrated 387
successively into target sites. Donor fragments for each modification could be assembled 388
together with I-SceI sites between the fragments. The assembled fragments could be cloned 389
into a single donor plasmid. Upon I-SceI cleavage of the chromosome and donor plasmid, the 390
linear dsDNA fragments from the donor plasmid could introduce a series of genetic 391
modifications into the chromosome. Recent studies demonstrated that Cas/CRISPR 392
RNA-guided targeting methods can be used to edit bacterial genomes (37). These methods 393
appear to be more efficient than previously reported TALENs method (38). Two-plasmid 394
recombination methods could be further modified by replacing I-SceI cleavage with 395
Cas/CRISPR and combining two-step recombination into one step. 396
Our data demonstrated that the new method presented here are a useful tool for 397
unmarked genetic modification of E. coli. This effective method can be used with both 398
essential and nonessential genes modification, and will benefit basic and applied genetic 399
research. 400
401
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
21
402
ACKNOWLEDGEMENTS 403
We thank Dr. Daibang Nie and Dr. Jian Li for thoughtful discussions and critical reading of 404
the manuscript. We thank Prof. Benjamin S Glick (University of Chicago) providing SnapGene 405
software for producing plasmid maps. We also thank the editor and the anonymous referees 406
for their comments and constructive suggestions. 407
This work was supported by The National Basic Research Program (973 Program) of 408
China (2014CB745100, 2011CBA00800). 409
Conflict of interest statement: None declared. 410
411
412
SUPPLEMENTAL MATERIAL 413
Supplementary Table S1, Supplementary Figures S1~S2, Supplementary Methods, 414
Protocol. 415
416
REFERENCES 417
418
1. Baba T, Ara T, Hasegawa M, Takai Y, Okumura Y, Baba M, Datsenko KA, 419
Tomita M, Wanner BL, Mori H. 2006. Construction of Escherichia coli K-12 in-frame, 420
single-gene knockout mutants: the Keio collection. Mol Syst Biol 2:2006 0008. 421
2. Yu D, Ellis HM, Lee EC, Jenkins NA, Copeland NG, Court DL. 2000. An 422
efficient recombination system for chromosome engineering in Escherichia coli. Proc 423
Natl Acad Sci U S A 97:5978-5983. 424
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
22
3. Datsenko KA, Wanner BL. 2000. One-step inactivation of chromosomal 425
genes in Escherichia coli K-12 using PCR products. Proc Natl Acad Sci U S A 426
97:6640-6645. 427
4. Murphy KC. 1998. Use of bacteriophage lambda recombination functions 428
to promote gene replacement in Escherichia coli. J Bacteriol 180:2063-2071. 429
5. Marx CJ, Lidstrom ME. 2002. Broad-host-range cre-lox system for 430
antibiotic marker recycling in gram-negative bacteria. Biotechniques 33:1062-1067. 431
6. Blomfield IC, Vaughn V, Rest RF, Eisenstein BI. 1991. Allelic exchange 432
in Escherichia coli using the Bacillus subtilis sacB gene and a temperature-sensitive 433
pSC101 replicon. Mol Microbiol 5:1447-1457. 434
7. Warming S, Costantino N, Court DL, Jenkins NA, Copeland NG. 2005. 435
Simple and highly efficient BAC recombineering using galK selection. Nucleic Acids 436
Res 33:e36. 437
8. Wong QN, Ng VC, Lin MC, Kung HF, Chan D, Huang JD. 2005. Efficient 438
and seamless DNA recombineering using a thymidylate synthase A selection system 439
in Escherichia coli. Nucleic Acids Res 33:e59. 440
9. Stringer AM, Singh N, Yermakova A, Petrone BL, Amarasinghe JJ, 441
Reyes-Diaz L, Mantis NJ, Wade JT. 2012. FRUIT, a scar-free system for targeted 442
chromosomal mutagenesis, epitope tagging, and promoter replacement in Escherichia 443
coli and Salmonella enterica. PLoS One 7:e44841. 444
10. Springer B, Kidan YG, Prammananan T, Ellrott K, Bottger EC, Sander P. 445
2001. Mechanisms of streptomycin resistance: selection of mutations in the 16S rRNA 446
gene conferring resistance. Antimicrob Agents Chemother 45:2877-2884. 447
11. Nefedov M, Williamson R, Ioannou PA. 2000. Insertion of 448
disease-causing mutations in BACs by homologous recombination in Escherichia coli. 449
Nucleic Acids Res 28:E79. 450
12. Stoddard BL. 2005. Homing endonuclease structure and function. Q Rev 451
Biophys 38:49-95. 452
13. Monteilhet C, Perrin A, Thierry A, Colleaux L, Dujon B. 1990. 453
Purification and characterization of the in vitro activity of I-Sce I, a novel and highly 454
specific endonuclease encoded by a group I intron. Nucleic Acids Res 18:1407-1413. 455
14. Kuhlman TE, Cox EC. 2010. Site-specific chromosomal integration of 456
large synthetic constructs. Nucleic Acids Res 38:e92. 457
15. Kuhlman TE, Cox EC. 2010. A place for everything: chromosomal 458
integration of large constructs. Bioeng Bugs 1:296-299. 459
16. Yu BJ, Kang KH, Lee JH, Sung BH, Kim MS, Kim SC. 2008. Rapid and 460
efficient construction of markerless deletions in the Escherichia coli genome. Nucleic 461
Acids Res 36:e84. 462
17. Tischer BK, von Einem J, Kaufer B, Osterrieder N. 2006. Two-step 463
red-mediated recombination for versatile high-efficiency markerless DNA manipulation 464
in Escherichia coli. Biotechniques 40:191-197. 465
18. Gerlach RG, Blank K, Wille T. 2013. Site-Directed Mutagenesis Using 466
Oligonucleotide-Based Recombineering. In Figurski D (ed.), Genetic Manipulation of 467
DNA and Protein - Examples from Current Research. InTech. 468
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
23
19. Posfai G, Kolisnychenko V, Bereczki Z, Blattner FR. 1999. Markerless 469
gene replacement in Escherichia coli stimulated by a double-strand break in the 470
chromosome. Nucleic Acids Res 27:4409-4415. 471
20. Blank K, Hensel M, Gerlach RG. 2011. Rapid and highly efficient method 472
for scarless mutagenesis within the Salmonella enterica chromosome. PLoS One 473
6:e15763. 474
21. Herring CD, Glasner JD, Blattner FR. 2003. Gene replacement without 475
selection: regulated suppression of amber mutations in Escherichia coli. Gene 476
311:153-163. 477
22. Guldener U, Heck S, Fielder T, Beinhauer J, Hegemann JH. 1996. A new 478
efficient gene disruption cassette for repeated use in budding yeast. Nucleic Acids Res 479
24:2519-2524. 480
23. Alting-Mees MA, Short JM. 1989. pBluescript II: gene mapping vectors. 481
Nucleic Acids Res 17:9494. 482
24. Gust B, Challis GL, Fowler K, Kieser T, Chater KF. 2003. PCR-targeted 483
Streptomyces gene replacement identifies a protein domain needed for biosynthesis 484
of the sesquiterpene soil odor geosmin. Proc Natl Acad Sci U S A 100:1541-1546. 485
25. Amann E, Ochs B, Abel KJ. 1988. Tightly regulated tac promoter vectors 486
useful for the expression of unfused and fused proteins in Escherichia coli. Gene 487
69:301-315. 488
26. Gibson DG, Young L, Chuang RY, Venter JC, Hutchison CA, 3rd, Smith 489
HO. 2009. Enzymatic assembly of DNA molecules up to several hundred kilobases. 490
Nat Methods 6:343-345. 491
27. Sawitzke JA, Thomason LC, Costantino N, Bubunenko M, Datta S, 492
Court DL. 2007. Recombineering: in vivo genetic engineering in E. coli, S. enterica, 493
and beyond. Methods in enzymology 421:171-199. 494
28. Hogrefe HH, Cline J, Youngblood GL, Allen RM. 2002. Creating 495
randomized amino acid libraries with the QuikChange Multi Site-Directed Mutagenesis 496
Kit. Biotechniques 33:1158-1160, 1162, 1164-1155. 497
29. Tucker AM, Winkler HH, Driskell LO, Wood DO. 2003. 498
S-adenosylmethionine transport in Rickettsia prowazekii. J Bacteriol 185:3031-3035. 499
30. Driskell LO, Tucker AM, Winkler HH, Wood DO. 2005. Rickettsial 500
metK-encoded methionine adenosyltransferase expression in an Escherichia coli 501
metK deletion strain. J Bacteriol 187:5719-5722. 502
31. Wei Y, Newman EB. 2002. Studies on the role of the metK gene product of 503
Escherichia coli K-12. Mol Microbiol 43:1651-1656. 504
32. Terpe K. 2006. Overview of bacterial expression systems for heterologous 505
protein production: from molecular and biochemical fundamentals to commercial 506
systems. Appl Microbiol Biotechnol 72:211-222. 507
33. Hayashi M, Tabata K. 2013. Metabolic engineering for L-glutamine 508
overproduction by using DNA gyrase mutations in Escherichia coli. Appl Environ 509
Microbiol 79:3033-3039. 510
34. Marchuk D, Drumm M, Saulino A, Collins FS. 1991. Construction of 511
T-vectors, a rapid and general system for direct cloning of unmodified PCR products. 512
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
24
Nucleic Acids Res 19:1154. 513
35. Holton TA, Graham MW. 1991. A simple and efficient method for direct 514
cloning of PCR products using ddT-tailed vectors. Nucleic Acids Res 19:1156. 515
36. Cha J, Bishai W, Chandrasegaran S. 1993. New vectors for direct cloning 516
of PCR products. Gene 136:369-370. 517
37. Jiang W, Bikard D, Cox D, Zhang F, Marraffini LA. 2013. RNA-guided 518
editing of bacterial genomes using CRISPR-Cas systems. Nat Biotechnol 31:233-239. 519
38. Joung JK, Sander JD. 2013. TALENs: a widely applicable technology for 520
targeted genome editing. Nat Rev Mol Cell Biol 14:49-55. 521
522
523
524
525
526
527
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
25
528
Figure Legends 529
530
Figure 1. Diagram of this two-plasmid method. 531
A: Antibiotic cassette fragment with I-SceI recognition sites integrated at a target via λ Red 532
-mediated recombination creating an intermediate strain. B: Intermediate strain with mutation(s) 533
and I-SceI recognition sites was transformed with donor plasmid. Expression of I-SceI was 534
induced by L-rhamnose or IPTG (isopropyl β-D-1-thiogalactopyranoside). I-SceI recognition 535
sites in donor plasmid and chromosome were cleaved. Integration of donor fragment at the 536
cleaved site of chromosome was mediated by λ Red recombination. 537
538
Figure 2. Plasmids used for this method. 539
A: pREDTKI and pREDTAI (helper plasmids) with arabinose-inducible (araB promoter ) 540
λ-Red recombinase functions, IPTG-inducible (trc promoter ) I-SceI expression. B: pKSI-1 541
(modular plasmids for donor) based on the high-copy vector pBluescript II KS (-), with multiple 542
cloning site (MCS) and two I-SceI recognition sites. C: Part of plasmid pMDIAI (template 543
plasmid) with an apramycin-resistance gene flanked by FRT (Flippase Recognition Targe) 544
sites and I-SceI recognition sites. For pMDISI, apramycin-resistance was replaced with 545
spectinomycin-resistance genes. Pink arrow: binding sites for primers MDF and MDR. 546
547
Figure 3. Seamless pepD deletion and PCR analysis 548
A: 40 bp short arms homologous to pepD ORF; 600 bp pepD segment replaced by 549
marker. B: Arms with upstream (U) and downstream (D) homology to pepD, with truncated 1.0 550
kb pepD segments and apramycin (apr) resitance genes seamlessly deleted from ATG to TAA. 551
C: PCR of pepD. BL21(DE3) (wild-type pepD), 2.5 kb; intermediate strain (pepD::apr), 3.3 kb; 552
final strain (ΔpepD), 0.9 kb. 553
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
26
554
Figure 4. Modifications to metK gene and PCR analysis. 555
A: IsceI-apr-IsceI cassette inserted into yqgC and IsceI-spc-IsceI cassette inserted into 556
galP. YqgC and galP are nonessential genes next to metK. B: Wild-type metK and resistant 557
genes replaced by mutated metK or SAM transporter. galP fragment and speA-yqgB-yqgC 558
fragment were homologous arms. C: PCR analysis of metK. MG1655 (wild-type metK), 2.8 kb; 559
intermediate strain A10 (yqgC::apr, galP::spc), 5.9 kb; final strain with mutated metK, 2.8 kb, 560
the same length as wild-type metK; final strain with metK replaced by SAM transporter, 2.6 kb. 561
562
563
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
1
orfA
marker
marker
marker
mut
mut
E. coli Chromosome
orfA 3'orfA 5' I-SceII-SceI
helper
helper
helper
helperorfA 5' orfA 3'
Double Crossover Recomination
Curing
Double Crossover Recomination
PCR-Targeting
B
A
mut
orfA 3'orfA 5'
I-SceI I-SceI
Donor
I-SceI Cutting + Recombination
Transformation
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
��������������
�.5�k�
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
������
��
=��5��
��� ��
��� ����
����
����
�����
����������
�����
����
����
�����
��������������
�������
��������� ���
����
����
������
���������!��
���"�#$�
%��&��
'���
��� ��
=��5��
%( ����
�'��)�
�.��k� on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
Mark
er
Metk(
WT)
yqgC::a
pr
&galP::s
pc
SAM tr
ansporto
r
metK
(mut)
Mark
er
5.9 k
2.8 k
2.6 k3.0 k
6.0 k
1.0 k
2.5 k2.8 k
C
A
B
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
C
Marker pepD pepD::apr pepD
3.2 k
2.5 k
0.9 k
3.0 k
0.5 k
1.0 k
A
B
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
Table 1: Strains and plasmids used in this study
Strain or plasmid Characteristics Source
Strains
DH5alpha General cloning host strain Takara
MG1655 The "wild type" E. coli K-12 strain CGSC
W3110 Another E. coli K-12 strain CGSC
BL21(DE3) An E. coli B strain with DE3, suitable for
protein expression. Invitrogen
Helper plasmids Characteristics
pREDIA ori SC101(ts), AmpR, araBAD promoter for
λ-Red, rha promoter for I-SceI
This work
pREDKI ori SC101(ts), KanR, araBAD promoter for
λ-Red, rha promoter for I-SceI
This work
pREDTAI ori SC101(ts), AmpR, araBAD promoter for
λ-Red, trc promoter for I-SceI
This work
pREDTKI ori SC101(ts), KanR, araBAD promoter for
λ-Red, trc promoter for I-SceI
This work
Template plasmids
pMDIAI AprR, AmpR, Apramycin resistance gene flanked
by FRT and I-SceI sites. pMD18-T simple
backbone
This work
pMDISI SpcR, AmpR, Spectinomycin resistance gene
flanked by FRT and I-SceI sites. pMD18-T
simple backbone
This work
Modular plasmids for donor
pKSKI-1 AmpR, KanR, pBluescript II KS(-) backbone with
I-SceI site - kanMX - I-SceI site, differed in insert
direction with pKSKI-2
This work
pKSKI-2 AmpR, KanR, pBluescript II KS(-) backbone with
I-SceI site - kanMX - I-SceI site, differed in insert
direction with pKSKI-1
This work
pKSI-1 AmpR, pBluescript II KS(-) backbone with I-SceI
site - multiple cloning sites - I-SceI site
This work
Other plasmid
pMD18-T simple AmpR, cloning vector Takara
pMD19-T simple AmpR, cloning vector Takara
pBackzero-T KmR, cloning vector Takara
pUG6 KmR vector EUROSCARF,
Guldener, et al., 1996
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
pPIC3.5K AmpR, P. pastoris expression plasmid Invitrogen
pBluecriptII KS(-) Cloning vector Alting-Mees and Short,
1989
pKS-K AmpR, KanR, pBluescript II KS(-) backbone
with kanMX This work
pKD46 λ-Red helper plasmid Datsenko and Wanner,
2000,
pKD46K λ-Red helper plasmid with kanamycin
resistance This work
pMDcadA. cadA fragment subcloned into pMD18-T
simple This work
pMDcadA-ISceI-Apr Plasmid for integrating spectinomycin
resitance gene into cadA locus This work
pMDcadAL pMD18-T simple with truncated cadA
fragment This work
pBackZero-ISceI-cadA Donor plasmid for cadA deletion, pBackzero-T
backbone This work
pKSIpepD0.7k Donor plasmid for pepD deletion, derived from
pKSI-1 This work
pMD19-gntT gntT fragment subcloned into pMD19-T
simple This work
pMD19-gntT-spc Plasmid for integrating spectinomycin
resitance gene into the gntT locus This work
pMD19-gntT-gdhA Donor plasmid for integrating gdhA gene This work
pKSImetK metK fragment cloned into pKSI-1 This work
pKSImetKmut Donor plasmid for site-directed mutagenesis
of the metK This work
pKSImetKSamT Donor plasmid for replacing metK with SAM
transporter This work
AmpR: Ampicillin resistance; Km
R: Kanamycin resistance, AprR: Apramycin resistance; SpcR:
Spectinomycin resistance
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from
Table 2: Rates for marker gene eviction, donor plasmid curing and correct modification
Modification Helper plasmid/
promoter for
I-SceI expression
Donor plasmids
backbone
Marker eviction rate
(Apr and/or Spc
sensitive colonies per
tested colonies)
Donor plasmid curing rate
(Amp or Kan sensitive
colonies per tested
colonies)
Correct modification rate,
confirmed by PCR
(colonies with expected PCR
bands per tested colonies)
Correct modification
rate, confirmed by
PCR
fragment sequencing
cadA deletion pREDIA / rhaB pBackZero-T 4/10 3/4 3/3 3/3
cadA deletion pREDTKI / trc pMD19-T 20/20 18/20 18/20 4/4
gntT integration pREDTKI / trc pMD19-T 8/10 7/8 7/7 4/4
PepD seamless
deletion pREDKI / rhaB
pKSI-1 3/8 2/3 2/2 2/2
PepD seamless
deletion pREDTKI / trc
pKSI-1 6/8 6/6 6/6 6/6
MetK mutation pREDKI / rhaB pKSI-1 9/30 5/9 9/9 4/4
MetK mutation pREDTKI / trc pKSI-1 18/18 18/18 16/18 4/4
MetK
replacement pREDTKI / trc
pKSI-1 2/4 2/2 2/2 2/2
Amp: Ampicillin; Apr: Apramycin ; Spc: Spectinomycin.
on March 3, 2021 by guest
http://aem.asm
.org/D
ownloaded from