Saturating mutagenesis of an essential gene: a majority of the Neisseria gonorrhoeae major outer membrane porin (PorB) is mutable.

Saturating Mutagenesis of an Essential Gene: a Majority of the Neisseria gonorrhoeae Major Outer Membrane Porin (PorB) Is Mutable

Updated information and services can be found at: http://jb.asm.org/content/196/3/540 These include: SUPPLEMENTAL MATERIAL REFERENCES

CONTENT ALERTS

Supplemental material This article cites 39 articles, 26 of which can be accessed free at: http://jb.asm.org/content/196/3/540#ref-list-1 Receive: RSS Feeds, eTOCs, free email alerts (when new articles cite this article), more»

Information about commercial reprint orders: http://journals.asm.org/site/misc/reprints.xhtml To subscribe to to another ASM Journal go to: http://journals.asm.org/site/subscriptions/

Downloaded from http://jb.asm.org/ on September 30, 2014 by GEORGE MASON UNIV

Adrienne Chen and H. Steven Seifert J. Bacteriol. 2014, 196(3):540. DOI: 10.1128/JB.01073-13. Published Ahead of Print 15 November 2013.

Saturating Mutagenesis of an Essential Gene: a Majority of the Neisseria gonorrhoeae Major Outer Membrane Porin (PorB) Is Mutable Adrienne Chen, H. Steven Seifert ‹Department of Microbiology-Immunology, Northwestern University Feinberg School of Medicine, Chicago, Illinois, USA

T

he identification and study of essential genes in bacteria has been an experimental challenge. Gene products required for basic biological functions are significant in that their study informs important biochemical and enzymatic processes that occur within the cell and are also considered to be attractive targets for antimicrobial therapy; however, their essential nature renders them difficult to study. Recent developments in genomics have allowed for identification of essential genes expressed by an organism via comparative genomics or transposon-mediated genomewide gene inactivation, but we are still constrained in our ability to study phenotypes associated with these genes. Beyond classical genetic approaches using chemical or spontaneous mutation to isolate mutants with a biological phenotype, current strategies for functional studies of essential genes are limited to generating partial loss-of-function mutants by transposon mutagenesis, generating conditional mutants of the gene of interest, the use of inducible promoters for controlled depletion of gene products, or gene overexpression (1). However, not all of these strategies are amenable for use in all microorganisms; for instance, the strict human pathogen Neisseria gonorrhoeae exhibits greatly reduced growth at temperatures above 41°C (2), which makes the isolation of conditional temperature-sensitive mutants problematic. The major outer membrane porin (PorB) of N. gonorrhoeae is an essential protein that forms a voltage gated pore that mediates ion exchange between the organism and its environment (3, 4). It has also been directly implicated in bacterial pathogenesis by contributing to serum resistance via interactions with complement factors (5–10), mediating invasion of host cells (11–13), and modulating apoptotic signaling in infected cells (14–18). PorB exists in the bacterial outer membrane as a homotrimeric ␤-barrel; each monomer is 32 to 35 kDa and consists of 16 transmembranespanning segments and 8 extracellular loops. N. gonorrhoeae strains contain a single porB gene that is usually found in one of two allelic forms (P.IA or P.IB) which possess ⬃80% nucleotide similarity and have been associated with different infection phe-

540

jb.asm.org

Journal of Bacteriology

notypes (19, 20). The closely related organism Neisseria meningitidis possesses a porB gene that shares 60 to 70% amino acid similarity to gonococcal porB, and other commensal Neisseria species also encode porin genes with various degrees of homology to porB (21). The majority of the variation between different porin genes and porB alleles can be found in the extracellular loop domains, with more amino acid conservation within the transmembrane regions (21). N. meningitidis also expresses a related PorA porin that serves as a trimeric cation pore in the outer membrane (22); the gonococcal porA orthologue is a pseudogene (23). PorB’s essential function in maintaining N. gonorrhoeae viability has presented challenges in terms of investigating other biological functions of porin such as its role in modulating the immune response or its ability to affect the viability of infected cells. Since all mutations generated in porB need to preserve its outer membrane pore function, we are limited in the ability to manipulate porin to study its functional domains. Previous strategies to study porin have involved generating hybrid porins from PorB molecules exhibiting different biological phenotypes (6, 7, 24), or attempts at deleting the extracellular loop domains with variable success (12, 39). Thus, to facilitate future structure-function studies of PorB, there is a need to identify essential versus nonessential residues of the protein to determine domains that are more amenable to mutagenesis. Our attempts at constructing a library of random viable porin

Received 10 September 2013 Accepted 8 November 2013 Published ahead of print 15 November 2013 Address correspondence to H. Steven Seifert, [email protected]. Supplemental material for this article may be found at http://dx.doi.org/10.1128 /JB.01073-13. Copyright © 2014, American Society for Microbiology. All Rights Reserved. doi:10.1128/JB.01073-13

p. 540 –547

February 2014 Volume 196 Number 3


The major outer membrane porin (PorB) of Neisseria gonorrhoeae is an essential protein that mediates ion exchange between the organism and its environment and also plays multiple roles in human host pathogenesis. To facilitate structure-function studies of porin’s multiple roles, we performed saturating mutagenesis at the porB locus and used deep sequencing to identify essential versus mutable residues. Random mutations in porB were generated in a plasmid vector, and mutant gene pools were transformed into N. gonorrhoeae to select for alleles that maintained bacterial viability. Deep sequencing of the input plasmid pools and the output N. gonorrhoeae genomic DNA pools identified mutations present in each, and the mutations in both pools were compared to determine which changes could be tolerated by the organism. We examined the mutability of 328 amino acids in the mature PorB protein and found that 308 of them were likely to be mutable and that 20 amino acids were likely to be nonmutable. A subset of these predictions was validated experimentally. This approach to identifying essential amino acids in a protein of interest introduces an additional application for next-generation sequencing technology and provides a template for future studies of both porin and other essential bacterial genes.

Essential Residues in Gonococcal Porin

MATERIALS AND METHODS Generating mutations in the FA1090 porB gene. The plasmid construct used to introduce mutations into FA1090 porB (expressing a P.IB allele) has been described previously (39). Briefly, the majority of the porB gene (nucleotides [nt] 50 to 1047) and sequences from the downstream intergenic region (500 nt) were cloned into the pBluescript vector (Stratagene) using EcoRI and SacI restriction sites. As part of the cloning process, a synonymous C-to-T mutation was engineered at nt 1017 of porB to introduce an NheI restriction site. This mutation does not impact bacterial growth or other PorB-associated phenotypes. The EZ-Tn5具Kan-2典 insertion kit (Epicentre) was used to randomly insert a kanamycin resistance cassette into sequences downstream of porB to provide a selectable marker. The porB gene downstream of nt 64 was divided into six smaller 150to 200-nt domains to be targeted for random mutagenesis (nt 64 to 243, nt 241 to 438, nt 439 to 600, nt 598 to 759, nt 760 to 909, and nt 910 to 1043 of porB). The N-terminal signal sequence was not mutated to ensure proper protein localization. Random mutagenesis of these defined porB regions was done using the GeneMorph II EZClone domain mutagenesis kit (Stratagene) according to the manufacturer’s specifications. The targeted porB domains were individually amplified under mutagenic PCR conditions to generate a pool of megaprimers containing random mutations (see Table S1 in the supplemental material). The megaprimers were then used to amplify the original pBluescript plasmid containing porB. The PCR products were digested with DpnI for 2 h at 37°C and then transformed into XL10-Gold ultracompetent cells (Stratagene). Plasmid DNA from kanamycin-resistant clones containing a random pool of porB mutants was purified from the resulting pool of colonies. For each mutated domain, plasmids from 2,000 E. coli transformants were isolated in duplicate, and the resultant pool of DNA from ⬃4,000 clones was used for subsequent transformation into N. gonorrhoeae. For site-directed mutagenesis of individual amino acids in PorB, the porB gene was amplified using a QuikChange II XL site-directed mutagenesis kit (Stratagene) according to the manufacturer’s specifications. The primers used to incorporate the desired mutations are found in Table S3 in the supplemental material. The appropriate mutants were identified by nucleotide sequencing and transformed into FA1090 as described below.


FA1090 growth, transformation, and isolation of porB recombinants. N. gonorrhoeae strains were cultured on gonococcal medium base (GCB; Difco) agar plus Kellogg’s supplements (26) and were grown at 37°C with 5% CO2 for 18 to 20 h. Strains expressing mutations in porB were generated in the background of a piliated FA1090 strain expressing multiple Opa proteins. Mutations were introduced into the FA1090 chromosome at the porB locus (NGO1812, between truA [NGO1811] and oxyR [NGO1813]) by spot transformation as described previously (27, 28), and recombinants were selected by plating transformants on GCB agar containing 50 ␮g of kanamycin/ml. For each mutated domain, genomic DNA was isolated from 20,000 kanamycin-resistant FA1090 transformants in duplicate, and the resultant pool of DNA from 40,000 clones was used as the template for amplification of mutated domains for next-generation sequencing. For site-directed mutagenesis, individual random recombinant colonies were resuspended in colony lysis solution (1% Triton, 20 mM Tris [pH 8.3], 2 mM EDTA) and heated at 94°C for 15 min, after which the DNA in the lysates was screened for the presence of introduced mutations by PCR amplification and nucleotide sequencing. Since all of the sequencing was performed on either pools of DNA isolated from recombinant colonies in bulk, or crude lysates of individual colonies, we are not in possession of recombinant strains expressing specific mutations. Amplification and preparation of mutated porB regions for nextgeneration sequencing. Mutated regions of the porB gene were amplified using either input plasmid DNA pools isolated from E. coli or output genomic DNA pools isolated from recombinant FA1090 as the templates. A total of 20 amplicons spanning the porB gene (⬃200 nt in size, offset by ⬃50 nt) were amplified by using either KOD Hot Start DNA polymerase (Novagen) for genomic DNA or GoTaq Flexi DNA polymerase (Promega) for plasmid DNA and gel purified from a 2% agarose gel (QIAquick gel extraction kit; Qiagen) (for primer sequences see Table S2 in the supplemental material). Each PCR was performed in triplicate, resulting in three sets of amplicons for the input plasmid pool and three sets of amplicons for the output genomic DNA pool. The resultant gel-purified PCR products were analyzed for purity and quantity using the Experion automated electrophoresis system and an Experion DNA 1K kit (Bio-Rad). A total of 50 ng of each of the 20 amplified and purified domains was combined for a total of 1 ␮g of DNA. The NEBNext DNA Library Prep Master Mix Set for Illumina and the NEBNext Multiplex Oligos for Illumina (New England BioLabs) were used to prepare the pooled amplicons for Illumina sequencing according to the manufacturer’s specifications. Agencourt AMPure XP beads (Beckman Coulter) were used to clean up the reactions after each of the end repair, dA-tailing, and adapter ligation steps. Size-selection of dA-tailed, adapter-ligated amplicons (⬃300 nt) was done using AMPure XP beads, where progressively smaller DNA fragments are retained on the beads with increasing concentrations of polyethylene glycol (PEG)-salt found in the bead buffer. AMPure XP beads were first added to the DNA solution at 0.7⫻, after which the beads bound to larger (⬎300-nt) DNA fragments were discarded. A further 0.3⫻ volume of beads was then added to the supernatant to increase the PEG-salt concentration, and the beads containing the target DNA were kept for washing and DNA elution. The size-selected adapter-ligated DNA was then subjected to PCR enrichment using 20 ng of adapter-ligated DNA as the template followed by gel purification. Six different index primers (Index 2, 4, 5, 6, 7, and 12) from the NEBNext Multiplex Oligos for Illumina kit were used to barcode the amplicon pools (three input and three output pools). The resultant DNA was analyzed for purity and quantity using the Experion automated electrophoresis system and DNA 1K kit. The input and output amplicon pools were mixed together so that there was 10-fold more output DNA relative to input DNA, and the sample was subjected to sequencing on one lane of an Illumina HiSeq 2000 (Illumina) with 100-bp single-end reads at the Tufts University Core Facility Genomics Core (Boston, MA). Demultiplexing of the reads was done with CASAVA-1.8.2, and Bowtie2 was used to align the sequences to one of the 20 amplicons. Reads that mapped to

jb.asm.org 541


mutant strains have been met with mixed results; although we have successfully isolated mutations at many residues, the process is inefficient, and most of the recombinant strains contain wildtype porin sequence (39). Thus, we wanted to use a more highthroughput approach for identifying mutations that could maintain expression of a functional porin. Next-generation sequencing technology has revolutionized biomedical research by dramatically reducing the cost of sequencing such that the cost per megabase is now less than $0.10 (http://www.genome.gov /sequencingcosts/). The ability to sequence large numbers of samples in parallel is broadly applicable to a number of diverse biological questions, including but not limited to sequencing genomes, identifying genetic and epigenetic variation within and between individuals, characterizing transcriptomes, detection of mutants, and SNP annotation (25). In the present study, we performed saturating mutagenesis of the porB gene by independently mutating smaller regions of porB in a plasmid background and transforming pools of plasmids isolated from Escherichia coli into N. gonorrhoeae to select for allowable porin mutations. We utilized Illumina sequencing to sequence the libraries of mutants and compared mutations present in the E. coli pools versus the N. gonorrhoeae pools to identify putative essential residues in the gene. Our results will inform future studies on porin’s role in N. gonorrhoeae pathogenesis and provide a template for expanded studies on other essential bacterial genes.

Chen and Seifert

randomly inserted in the downstream intergenic region using the EZ-Tn5具Kan-2典 transposon (Kanr). Several pools of plasmids expressing random mutations in different porB domains (indicated by different colors) were generated using PCR-based mutagenesis. Each plasmid pool containing both mutated and wild-type plasmids was transformed into N. gonorrhoeae strain FA1090 and kanamycin-resistant recombinants expressing porin variants that supported bacterial viability were selected. Plasmid and genomic DNA was isolated from the input and output pools, respectively, for further analysis.

the complementary strand were discarded. A custom Perl script was used to determine the frequencies of base calls at each nucleotide position in each amplicon, as well as to discard bases with a Q score of ⬍30.

RESULTS

Generating pools of FA1090 strains expressing random mutations in porB. In order to identify essential and nonessential residues in porB, pools of strains expressing random mutations in the porB gene were constructed. The aim was to perform saturating mutagenesis of porB in an E. coli plasmid vector and then to transform the pool of mutants into the chromosome of N. gonorrhoeae strain FA1090, where the only surviving recombinants would be those possessing porB genes that produce functional porins. In this way, essential PorB residues would be identifiable by comparing the mutations present in the input plasmid pool to the mutations expressed by the recombinant gonococcal strains. To this end, a plasmid containing the majority of the parental porB gene linked to a kanamycin resistance marker in the downstream intergenic region was constructed (39). The porB gene was divided into six 150- to 200-nt domains that were individually subjected to random mutagenesis (Fig. 1). We chose not to mutate the entire gene at once since multiple mutations along the entirety of the gene would be more likely to result in a nonfunctional porin. The signal sequence was not targeted for mutation to ensure proper localization of the mature protein. Each individual porB domain was subjected to PCR-based mutagenesis, and the mutated plasmids were transformed into E. coli. An initial screen to determine mutagenesis frequency showed that for each individually mutated domain, 22 to 42% of the clones contained zero mutations in that domain, 26 to 45% contained one mutation, and 19 to 32% contained two or more mutations. Overall, at least 50% of the clones in a given mutant pool contained at least one nucleotide mutation in the targeted region; thus, to obtain a minimum of 5-fold coverage for each mutated domain, we estimated that at least 2,000 E. coli clones would be needed. Plasmid DNA from ⬎2,000 E. coli transformants per in-

542

jb.asm.org

dividual mutated domain (input pools) was isolated for subsequent transformation into N. gonorrhoeae. Plasmid DNA pools isolated from E. coli were transformed into N. gonorrhoeae strain FA1090, and kanamycin selection used to isolate clones that had recombined mutant porB genes into the chromosomal locus. Initial screening to determine recombination frequencies of random mutants in each domain yielded frequencies between 3.2 and 41.9% of the kanamycin-resistant clones containing at least one nucleotide mutation in the mutated domain of interest. For each individual mutated domain, 58 to 97% of the recombinant clones contained zero mutations in that domain, 0 to 19% contained one mutation, and 0 to 22% contained two or more mutations. We assessed the linkage frequencies of various defined and allowable porB mutations with the kanamycin resistance cassette and the results showed a large variation depending on the mutation, from 2 to ⬎50%, and did not correlate with the distance of the mutation from the selectable marker. Because it is difficult to predict which residues and/or domains would be more or less mutable in the PorB protein, we chose to screen at least 20,000 recombinant FA1090 transformants from each mutated domain. Genomic DNA was then isolated from ⬎20,000 kanamycin-resistant FA1090 colonies for deep sequencing (output pools). Deep sequencing of porB mutant pools. Plasmid and genomic DNA containing mutations in porB was amplified for sequencing using 20 sets of primers designed to amplify 200-nt fragments of porB spanning the entire gene, where the fragments were offset from each other by ⬃50 nt (Fig. 2). In order to minimize the presence of nonspecific amplification products, PCR products were gel purified and then analyzed with an automated electrophoresis system to confirm that only a single amplicon was present. Amplification from each of the input and output DNA pools was performed in triplicate, and for each replicate, the 20 amplicons spanning the porB gene were combined and ligated to Illumina TruSeq universal and indexed adapters such that each input/



FIG 1 Generating random mutations in the N. gonorrhoeae porB gene for sequencing. The porB gene was cloned next to a kanamycin resistance cassette that was


A

output replicate was assigned one barcode. All of the amplicon pools were then combined such that there was ⬃10-fold more amplified DNA from the output pools relative to the input pools to ensure greater sequence coverage of the output pools which would contain a lower frequency of mutation. The entire DNA library was sequenced on one lane of an Illumina HiSeq 2000, with 100-bp reads. Identification of putative essential residues in porB. After separating the sequences according to their adapter indices into the three input and three output pools, the frequency of base calls at each position in the gene was determined. Each sequence was assigned to an amplicon, and we only considered sequences in the forward orientation and discarded sequences in the reverse orientation (Fig. 2), with the exception of the two amplicons at the C terminus of the protein where the complementary sequence was read. Bases with a Phred quality score of ⬍30 were eliminated to minimize the chance of incorporating sequencing errors into the analysis. Indels were also removed since those should represent amplification or sequencing errors and in most cases would not be expected to produce a viable porin. For each nucleotide position the frequency of non-wild-type bases was determined relative to the total number of reads at that position. Replicates from the input plasmid and output genomic pools were paired at random to compare frequencies of each base in the output pool relative to the input pool, such that we had three replicates for each ratio at each position. The ratio of output frequency to input frequency was calculated for each base at every nucleotide subjected to mutagenesis in the porB gene and was used to determine whether a given nucleotide was mutable (see Table S4 in the supplemental material). To determine the appropriate cutoff ratio for separating mutable versus nonmutable nucleotides, we considered a random selection of 20 porB amino acids that had been mutated experimentally with a single point mutation (39) (Fig. 3A, a subset of underlined residues), as well as a random selection of codons in the middle of the porB gene that could create a stop codon by a single point mutation. We reasoned that an optimal cutoff ratio


°

°

°

21

VTlYgAIKaGVQTYrSVEHTDgKVSKvETGSEIADFGskIGFKGQEDLGNGLKAVWQLEQ

81

GASVAGTNTGWGNKQSFVGLKGGFgTIRAGSLNSPLKNTGANVNAWESGKFTGNVLEISG

*

*

* *

*

*

141 MAQREHRYLSVRYDSPEFAGFSGSVQYAPKDNSGSNGESYHVGLNYQNSGFfAQYAGLFQ 201 RYGEGTKKIEYDGQTYSIPSLFVEKLQVHRLVGGYDNNALYvSVAAQQQDAKLYGAMSgN

°

*

°

*

°

°

261 SHnSQTEvAaTaAYRFGNVTPRVSYAHGFKGTVDSANhDnTYDQVVVGaEYDFSKRTSAL 321 VSAGwLQGEKGADKIVSTASAVVLRHKF

B

Extracellular loop domains

FIG 3 Summary of mutations. (A) The amino acid sequence of the residues in PorB subjected to random mutagenesis is shown (amino acids 21 to 348). Amino acids shown in green uppercase are predicted to be mutable, amino acids shown in purple uppercase are possibly mutable, and amino acids shown in red lowercase are predicted to be nonmutable. Residues shown in boxes represent the 8 extracellular loop domains. The underlined residues have been experimentally mutated in a previous study (39). Asterisks (*) indicate residues where putative stop codons are predicted to be allowable, and degree signs (Œ) indicate residues where mutations are missing in the input plasmid pool that could affect the predicted mutability of the amino acid. Arrows underneath residues indicate that they were targeted for site-directed mutagenesis in the present study (red, experimentally nonmutable; green, experimentally mutable; blue, putative allowable stop codons that were not isolated experimentally). (B) The predicted nonmutable residues were mapped onto the structure of the FA1090 PorB monomer as determined by I-TASSER modeling (36, 37). Visualization was done using the DeepView SwissPdbViewer (v4.1) (38). The structure is shown in two orientations for better visualization, and the locations of the extracellular loop domains are indicated. The predicted nonmutable residues that have not been experimentally tested are shown in red, the predicted nonmutable residues which have been previously mutated experimentally are shown in blue, and the predicted nonmutable residues that are experimentally nonmutable are shown in yellow. Whether amino acids localized in close proximity were predicted to interact was determined using the DeepView Swiss-PdbViewer (data not shown).

jb.asm.org 543


FIG 2 Sequencing strategy for identifying mutable residues in PorB. Mutated porin sequences were amplified from input plasmid and output genomic DNA pools to generate 200-bp-long amplicons offset by 50 nt along the length of the porB gene. Each amplicon contained both mutated and unmutated sequences (represented by colored and black line segments, respectively), and amplifications were done in triplicate. After purification and quantitation, the amplicons were pooled and ligated to Illumina TruSeq and indexed adapters (represented by the gray line segments) for sequencing. Sequences in the reverse orientation were discarded in the data analysis, with the exception of the two amplicons at the C terminus of the protein, where only the reverse complement sequence was read.

Chen and Seifert

544

jb.asm.org

TABLE 1 Recombination frequencies for targeted porB mutants

Targeted residue and position

Mutated to:

Predicted mutability

Recombination frequency (no. of recombinant colonies/total no. of colonies)a

Arginine 35 Phenylalanine 62 Serine 175 Phenylalanine 192 Tyrosine 274 Serine 318 Tryptophan 325

Alanine Alanine Stop codon Alanine Stop codon Alanine Alanine

Not mutable Mutable Allowable Not mutable Allowable Mutable Not mutable

0 (0/95) 0.13 (6/47) 0 (0/95) 0 (0/95) 0 (0/95) 0.56 (27/48) 0 (0/95)

a The recombination frequencies for the mutable amino acids were significantly different than the recombination frequencies of nonmutable amino acids as determined by the Fisher exact test (P ⬍ 0.005).

given nucleotide/codon is mutable or not. Of the possible 2,949 nonwild-type nucleotides present in the analyzed region (983 mutated nucleotide positions, 3 non-wild-type bases per position), 81 (2.7%) are not present in the input pool (i.e., they have an input frequency of 0) and thus could not be present in the output pool. These missing mutations from the input pool resulted in 8 amino acid residues where the incomplete data set could have affected our conclusions regarding their predicted mutability (Fig. 3A). Overall, ⬎90% of individual amino acids in the tested PorB region (308 of 328) are predicted to be either likely or possibly mutable (Fig. 3A, uppercase green and purple amino acids). Surprisingly, the allowable mutations were not enriched in the variable extracellular loop regions; the majority of transmembrane residues that are more highly conserved between Neisseria strains were also predicted to be mutable by our analysis. The predicted essential residues are also found in both loop and transmembrane regions, although many of the predicted nonmutable extracellular loop amino acids had been previously mutated experimentally (39). The nonmutable residues in the transmembrane regions do not localize to a particular region in the protein, and when these residues were mapped in the context of a predicted FA1090 PorB structure, no significant interactions were predicted between any of the essential amino acids that might constitute a structural domain, even for residues localized in close proximity to each other in the structure (Fig. 3B). Experimental validation of predicted essential and nonessential porB residues. To further validate the predictions from the sequencing analysis, we chose a selection of PorB residues for targeted mutation via site-directed mutagenesis (Fig. 3A). The F62 and S318 residues are predicted to be mutable, R35, F192, and W325 are predicted to be nonmutable, and S175 and Y274 are residues where putative nonsense mutations are predicted to be allowable. We mutated the F62, S318, R35, F192, and W325 amino acids to alanine in a plasmid vector and S175 and Y274 to stop codons and transformed N. gonorrhoeae strain FA1090 with the mutated plasmids. Kanamycin-resistant recombinants were isolated and screened for the presence of the introduced mutation by PCR amplification of the region of interest and sequencing. The recombination frequencies for each of the targeted mutations ranged from 0 to 0.56 (Table 1). As predicted, we were able to mutate F62 and S318 to alanine and were unable to isolate alanine substitution mutations at R35, F192, and W325. We were also unable to recover any recombinants containing premature stop



should allow identification of most or all of the previously mutated amino acids as mutable, and should not identify putative nonsense mutations as allowable. Based on these criteria we determined that an output frequency/input frequency ratio of 0.1 would allow most of the experimentally mutated amino acids to pass, while the majority of nucleotide changes resulting in putative stop codons possessed a ratio of ⬍0.1. At each nucleotide position, we considered whether that base could be mutated to each of the other three non-wild-type bases. If the ratio of the output frequency to input frequency was ⬎0.1 in three of three technical replicates for that base, we considered that mutation to be present in the output pool. If the output/input ratio was ⬎0.1 in two of three technical replicates, we considered the change to have a high probability of occurring. The putative amino acid changes resulting from each possible nucleotide change at every position was then determined. Any given amino acid in the PorB protein can be mutated to 5 to 7 other amino acids with a single point mutation depending on the nature of the codon sequence. For example, lysine 72 (CTG) can be mutated to methionine, valine, proline, arginine, or glutamine by a single nucleotide change. Conversely, isoleucine 53 (ATC) can be mutated to leucine, valine, phenylalanine, asparagine, serine, threonine, or methionine by a single nucleotide change (see Table S4 in the supplemental material). Once we identified all of the putative amino acid changes that were possible, whether the nucleotide mutations that are likely to have occurred (described above) would result in a concomitant amino acid change was determined. We considered a given amino acid to be mutable if at least two independent nonsynonymous substitutions at a given codon occurred in all three technical replicates (e.g., lysine 72 was mutated to valine, proline, arginine, and glutamine, and those mutations occurred in all three replicates). If only one nonsynonymous substitution occurred in all three technical replicates, or if one or two nonsynonymous substitutions occurred in two of three replicates, that codon was considered to be possibly mutable. Amino acids where no nonsynonymous substitutions occurred were considered to be nonmutable. The mutability of 328 amino acids in the mature PorB protein (amino acids 21 to 348 of the porB open reading frame, minus the N-terminal signal sequence) was assessed. Based on the criteria outlined above, 251 of 328 amino acids are predicted to be mutable, 57 are possibly mutable, and only 20 amino acids are predicted to be nonmutable (Fig. 3). We have previously experimentally mutated several residues in PorB through targeted and random mutagenesis (Fig. 3A, underlined residues) (39), and the majority of those amino acids were predicted to be mutable by this analysis, with the exception of 6 (of 115 known mutable amino acids, 5.2%) that were predicted to be nonmutable (Fig. 3A, residues in lowercase red and underlined; Fig. 3B, blue residues). We also examined the presence of nonsense mutations in the output pool; since porB is an essential gene in N. gonorrhoeae, stop codons in the middle of the coding region should not be allowable. There are 90 codons in the porB gene that could be mutated to a stop codon by a single point mutation, and in our analysis the majority of the putative mutations (81/90, 90%) were predicted to be not allowable (see Fig. 3A for amino acids where nonsense mutations are predicted to be allowable). We also considered the possibility that mutations missing from the input pool could have an impact on determining whether a


codons at either position S175 or position Y274. Overall, these results validate the predictions obtained from the sequencing analysis. DISCUSSION


jb.asm.org 545


The N. gonorrhoeae outer membrane porin PorB plays an important biological role for the organism not only in terms of its ability to transport small molecules into the periplasm but also because it has important functions for immune modulation and infection of host cells. The essential nature of the gene has made it challenging to dissect its functional domains since it was unknown which regions and/or residues of the protein are necessary for its outer membrane pore function. The development of next-generation sequencing technology has been of great utility for identifying mutations that are both naturally occurring such as those present in cancer genomes, as well as introduced mutations to identify virulence factors in pathogenic microorganisms. In the present study, we used saturating mutagenesis and deep sequencing to identify essential residues in PorB. The results of the sequencing analysis correlated well with experimental data and will inform future structure-function studies of PorB, as well as provide a blueprint for analyzing essential residues in other bacterial proteins that are required for viability. Our goal for the initial mutagenesis efficiencies of the input plasmid pool was to have every possible point mutation represented. Since the probability of obtaining a nonfunctional protein increases with multiple simultaneous mutations, we chose to err on the side of using a lower mutation frequency of one mutation per recombinant clone. We succeeded in generating 97.3% of the possible mutations using 20-fold coverage of each individually mutated domain, where approximately half of the clones did not contain any mutations. For future studies requiring higher coverage, either the mutation frequency or the number of clones screened could be increased easily given the large capacity of nextgeneration sequencing platforms. A more complete representation of potential mutants could be provided by using amplification primers where non-wild-type bases have been incorporated at a defined frequency at each nucleotide position, since the errorprone polymerases we used for PCR-based mutagenesis may have inherent biases in terms of which types of mutations are generated. The results of our analyses suggested that most residues in the mature PorB protein (308 of 328) can be mutated on an individual basis. Many of the allowable mutations resulted in at least one nonconservative substitution; we found evidence that ca. 90% of the 308 mutable amino acids could be changed to residues with different biochemical properties (data not shown; see also Table S4 in the supplemental material). Our predictions regarding mutability of specific residues correlated well with experimental data, though there were six experimentally mutated amino acids that were predicted to be nonmutable in the analysis (Fig. 3). It is unclear why these particular experimentally mutable residues were predicted to be nonmutable. Likely not all mutations are equally tolerated at all positions, and our method of random mutagenesis necessarily meant that each individual amino acid could only be mutated to seven others at most; thus, we are not observing the full spectrum of mutations that could potentially occur. This limitation in the scope of possible amino acid changes could also contribute to why some amino acids were considered to be “possibly mutable” versus “likely mutable.”

We also predicted several allowable stop codons in the middle of the open reading frame, which appeared incompatible with the idea that porin is essential for bacterial viability. One possible explanation for the detection of allowable stop codons in the output pools is the presence of contaminating plasmid DNA containing a wild-type porB gene carried over from the input pools, but since there were two overnight growth steps between introducing the transforming plasmid DNA and recovering recombinant genomic DNA, the likelihood of retaining significant amounts of contaminating DNA is low. Another possibility is the presence of errors occurring during sequencing; the removal of low-quality reads from the analysis was intended to minimize the inclusion of inaccurate base calls, although it is possible that some errors were still present. The incorporation of deleterious mutations could also be allowed if porB merodiploids were generated during recombination such that all or a portion of wild-type porB remained in the chromosome (29), but it is not clear whether all regions of the genome are prone to duplication. Based on our previous experience where we attempted to generate numerous targeted mutations in porB with variable success and our inability to recover strains expressing certain mutations in the present study, we do not believe this region of the genome readily undergoes duplication. Alternatively, translational readthrough of introduced stop codons could be occurring at a low frequency (30); however, we did not detect any evidence of this occurring after introducing premature stops using site-directed mutagenesis and screening ⬃95 recombinant clones by sequencing. The presence of allowable stop codons could also be an artifact of the way the analysis was performed, which is further discussed below. The fact that most of the amino acids analyzed were mutable was surprising given the essential nature of the protein and the higher degree of conservation in porin transmembrane domains between different Neisseria species relative to the extracellular loop sequences (21). However, we only considered single point mutations in this analysis, not multiple mutations or deletions, and the former are less likely to significantly alter the protein structure and may be better tolerated. Although our study provided an assessment of which residues could be mutated on an individual basis, we also did not determine whether any of the mutations were linked to each other. Our goal for mutagenesis was to generate an average of one mutation per 150- to 200-nt domain; thus, the likelihood of more than one mutation per codon would be small. Nevertheless, it remains possible that multiple mutations within a single codon could result in compensatory mutations, for example, those that remove a stop codon in the middle of the open reading frame. It is also possible that mutations at one residue that adversely affects pore function could be compensated for by a mutation at a neighboring residue, which we have not determined in our analysis. Porin is a highly abundant protein in the bacterial outer membrane, and it has been considered a vaccine candidate for both N. gonorrhoeae (31, 32) and N. meningitidis (33). PorB stimulates immune responses in humans and animals in the absence of adjuvants and can itself function as an adjuvant to enhance protection against Francisella infection in mice (34, 35). However, despite the fact that PorB performs an essential function for bacterial survival and has not been reported to undergo phase or antigenic variation, our studies suggest that a significant amount of variation can be tolerated throughout the length of the protein and, as such, it may not be an ideal vaccine target.

Chen and Seifert

In summary, we have designed here a strategy to identify essential residues in a bacterial outer membrane protein that is required for viability by utilizing next-generation sequencing technology to screen large pools of bacterial recombinants where the frequency of obtaining viable mutant strains is low. Our results will facilitate future structure/function studies of PorB, and this approach could be easily adapted to any number of other essential bacterial genes to potentially identify important structural or functional domains. ACKNOWLEDGMENTS

REFERENCES 1. Msadek T. 2009. Grasping at shadows: revealing the elusive nature of essential genes. J. Bacteriol. 191:4701– 4704. http://dx.doi.org/10.1128/JB .00572-09. 2. Larribe M, Taha MK, Topilko A, Marchal C. 1997. Control of Neisseria gonorrhoeae pilin gene expression by environmental factors: involvement of the pilA/pilB regulatory genes. Microbiology 143(Pt 5):1757–1764. http://dx.doi.org/10.1099/00221287-143-5-1757. 3. Haines KA, Yeh L, Blake MS, Cristello P, Korchak H, Weissmann G. 1988. Protein I, a translocatable ion channel from Neisseria gonorrhoeae, selectively inhibits exocytosis from human neutrophils without inhibiting O2⫺ generation. J. Biol. Chem. 263:945–951. 4. Rudel T, Schmid A, Benz R, Kolb HA, Lang F, Meyer TF. 1996. Modulation of neisseria porin (PorB) by cytosolic ATP/GTP of target cells: parallels between pathogen accommodation and mitochondrial endosymbiosis. Cell 85: 391– 402. http://dx.doi.org/10.1016/S0092-8674(00)81117-4. 5. Ram S, McQuillen DP, Gulati S, Elkins C, Pangburn MK, Rice PA. 1998. Binding of complement factor H to loop 5 of porin protein 1A: a molecular mechanism of serum resistance of nonsialylated Neisseria gonorrhoeae. J. Exp. Med. 188:671– 680. http://dx.doi.org/10.1084/jem.188.4 .671. 6. Ram S, Sharma AK, Simpson SD, Gulati S, McQuillen DP, Pangburn MK, Rice PA. 1998. A novel sialic acid binding site on factor H mediates serum resistance of sialylated Neisseria gonorrhoeae. J. Exp. Med. 187:743– 752. http://dx.doi.org/10.1084/jem.187.5.743. 7. Ram S, Cullinane M, Blom AM, Gulati S, McQuillen DP, Monks BG, O’Connell C, Boden R, Elkins C, Pangburn MK, Dahlback B, Rice PA. 2001. Binding of C4b-binding protein to porin: a molecular mechanism of serum resistance of Neisseria gonorrhoeae. J. Exp. Med. 193:281–295. http: //dx.doi.org/10.1084/jem.193.3.281. 8. Ngampasutadol J, Ram S, Blom AM, Jarva H, Jerse AE, Lien E, Goguen J, Gulati S, Rice PA. 2005. Human C4b-binding protein selectively interacts with Neisseria gonorrhoeae and results in species-specific infection. Proc. Natl. Acad. Sci. U. S. A. 102:17142–17147. http://dx.doi.org/10.1073 /pnas.0506471102. 9. Madico G, Ngampasutadol J, Gulati S, Vogel U, Rice PA, Ram S. 2007. Factor H binding and function in sialylated pathogenic neisseriae is influenced by gonococcal, but not meningococcal, porin. J. Immunol. 178: 4489 – 4497. 10. Ngampasutadol J, Ram S, Gulati S, Agarwal S, Li C, Visintin A, Monks B, Madico G, Rice PA. 2008. Human factor H interacts selectively with Neisseria gonorrhoeae and results in species-specific complement evasion. J. Immunol. 180:3426 –3435. 11. van Putten JP, Duensing TD, Carlson J. 1998. Gonococcal invasion of epithelial cells driven by P.IA, a bacterial ion channel with GTP binding properties. J. Exp. Med. 188:941–952. http://dx.doi.org/10.1084/jem.188.5.941. 12. Bauer FJ, Rudel T, Stein M, Meyer TF. 1999. Mutagenesis of the Neisseria gonorrhoeae porin reduces invasion in epithelial cells and enhances phagocyte responsiveness. Mol. Microbiol. 31:903–913. http://dx.doi.org/10 .1046/j.1365-2958.1999.01230.x.

546

jb.asm.org



Traditional sequencing services were performed at the Northwestern University Genomics Core Facility. Illumina HiSeq 2000 sequencing was performed at the Tufts University Core Facility Genomics Core. We thank Albert Tai for bioinformatics support and Mark Anderson, Laty Cahoon, and Carl Gunderson for critical reading and editing of the manuscript. This study was supported by grants R01 AI044239 and R37 AI033493 to H.S.S. A.C. was partially supported by American Heart Association grant 10POST2550017.

13. Kuhlewein C, Rechner C, Meyer TF, Rudel T. 2006. Low-phosphatedependent invasion resembles a general way for Neisseria gonorrhoeae to enter host cells. Infect. Immun. 74:4266 – 4273. http://dx.doi.org/10.1128 /IAI.00215-06. 14. Muller A, Gunther D, Dux F, Naumann M, Meyer TF, Rudel T. 1999. Neisserial porin (PorB) causes rapid calcium influx in target cells and induces apoptosis by the activation of cysteine proteases. EMBO J. 18: 339 –352. http://dx.doi.org/10.1093/emboj/18.2.339. 15. Muller A, Gunther D, Brinkmann V, Hurwitz R, Meyer TF, Rudel T. 2000. Targeting of the proapoptotic VDAC-like porin (PorB) of Neisseria gonorrhoeae to mitochondria of infected cells. EMBO J. 19:5332–5343. http://dx.doi.org/10.1093/emboj/19.20.5332. 16. Muller A, Rassow J, Grimm J, Machuy N, Meyer TF, Rudel T. 2002. VDAC and the bacterial porin PorB of Neisseria gonorrhoeae share mitochondrial import pathways. EMBO J. 21:1916 –1929. http://dx.doi.org/10 .1093/emboj/21.8.1916. 17. Binnicker MJ, Williams RD, Apicella MA. 2004. Gonococcal porin IB activates NF-␬B in human urethral epithelium and increases the expression of host antiapoptotic factors. Infect. Immun. 72:6408 – 6417. http: //dx.doi.org/10.1128/IAI.72.11.6408-6417.2004. 18. Kozjak-Pavlovic V, Dian-Lothrop EA, Meinecke M, Kepp O, Ross K, Rajalingam K, Harsman A, Hauf E, Brinkmann V, Gunther D, Herrmann I, Hurwitz R, Rassow J, Wagner R, Rudel T. 2009. Bacterial porin disrupts mitochondrial membrane potential and sensitizes host cells to apoptosis. PLoS Pathog. 5:e1000629. http://dx.doi.org/10.1371/journal .ppat.1000629. 19. Brunham RC, Plummer F, Slaney L, Rand F, DeWitt W. 1985. Correlation of auxotype and protein I type with expression of disease due to Neisseria gonorrhoeae. J. Infect. Dis. 152:339 –343. http://dx.doi.org/10 .1093/infdis/152.2.339. 20. Cannon JG, Buchanan TM, Sparling PF. 1983. Confirmation of association of protein I serotype of Neisseria gonorrhoeae with ability to cause disseminated infection. Infect. Immun. 40:816 – 819. 21. Derrick JP, Urwin R, Suker J, Feavers IM, Maiden MC. 1999. Structural and evolutionary inference from molecular variation in Neisseria porins. Infect. Immun. 67:2406 –2413. 22. Tommassen J, Vermeij P, Struyve M, Benz R, Poolman JT. 1990. Isolation of Neisseria meningitidis mutants deficient in class 1 (porA) and class 3 (porB) outer membrane proteins. Infect. Immun. 58:1355–1359. 23. Feavers IM, Maiden MC. 1998. A gonococcal porA pseudogene: implications for understanding the evolution and pathogenicity of Neisseria gonorrhoeae. Mol. Microbiol. 30:647– 656. http://dx.doi.org/10.1046/j .1365-2958.1998.01101.x. 24. Carbonetti NH, Simnad VI, Seifert HS, So M, Sparling PF. 1988. Genetics of protein I of Neisseria gonorrhoeae: construction of hybrid porins. Proc. Natl. Acad. Sci. U. S. A. 85:6841– 6845. (Erratum, 86:1317, 1988.) 25. Shendure J, Lieberman Aiden E. 2012. The expanding scope of DNA sequencing. Nat. Biotechnol. 30:1084–1094. http://dx.doi.org/10.1038/nbt.2421. 26. Kellogg DS, Jr, Peacock WL, Deacon WE, Brown L, Pirkle CI. 1963. Neisseria gonorrhoeae. I. Virulence genetically linked to clonal variation. J. Bacteriol. 85:1274 –1279. 27. Stein DC, Petricoin EF, Griffiss JM, Schneider H. 1988. Use of transformation to construct Neisseria gonorrhoeae strains with altered lipooligosaccharides. Infect. Immun. 56:762–765. 28. Stohl EA, Seifert HS. 2001. The recX gene potentiates homologous recombination in Neisseria gonorrhoeae. Mol. Microbiol. 40:1301–1310. http://dx.doi.org/10.1046/j.1365-2958.2001.02463.x. 29. Taha MK, So M, Seifert HS, Billyard E, Marchal C. 1988. Pilin expression in Neisseria gonorrhoeae is under both positive and negative transcriptional control. EMBO J. 7:4367– 4378. 30. Baranov PV, Gesteland RF, Atkins JF. 2002. Recoding: translational bifurcations in gene expression. Gene 286:187–201. http://dx.doi.org/10 .1016/S0378-1119(02)00423-7. 31. Zhu W, Thomas CE, Sparling PF. 2004. DNA immunization of mice with a plasmid encoding Neisseria gonorrhoeae PorB protein by intramuscular injection and epidermal particle bombardment. Vaccine 22:660 – 669. http://dx.doi.org/10.1016/j.vaccine.2003.08.036. 32. Zhu W, Thomas CE, Chen CJ, Van Dam CN, Johnston RE, Davis NL, Sparling PF. 2005. Comparison of immune responses to gonococcal PorB delivered as outer membrane vesicles, recombinant protein, or Venezuelan equine encephalitis virus replicon particles. Infect. Immun. 73:7558 – 7568. http://dx.doi.org/10.1128/IAI.73.11.7558-7568.2005. 33. Wright JC, Williams JN, Christodoulides M, Heckels JE. 2002. Immuniza-


tion with the recombinant PorB outer membrane protein induces a bactericidal immune response against Neisseria meningitidis. Infect. Immun. 70: 4028 – 4034. http://dx.doi.org/10.1128/IAI.70.8.4028-4034.2002. 34. Wetzler LM. 2010. Innate immune function of the neisserial porins and the relationship to vaccine adjuvant activity. Future Microbiol. 5:749 – 758. http://dx.doi.org/10.2217/fmb.10.41. 35. Chiavolini D, Weir S, Murphy JR, Wetzler LM. 2008. Neisseria meningitidis PorB, a Toll-like receptor 2 ligand, improves the capacity of Francisella tularensis lipopolysaccharide to protect mice against experimental tularemia. Clin. Vaccine Immunol. 15:1322–1329. http://dx.doi.org/10 .1128/CVI.00125-08.

36. Zhang Y. 2008. I-TASSER server for protein 3D structure prediction. BMC Bioinformatics 9:40. http://dx.doi.org/10.1186/1471-2105-9-40. 37. Roy A, Kucukural A, Zhang Y. 2010. I-TASSER: a unified platform for automated protein structure and function prediction. Nat. Protoc. 5:725– 738. http://dx.doi.org/10.1038/nprot.2010.5. 38. Guex N, Peitsch MC. 1997. SWISS-MODEL and the Swiss-PdbViewer: an environment for comparative protein modeling. Electrophoresis 18: 2714 –2723. http://dx.doi.org/10.1002/elps.1150181505. 39. Chen A, Seifert HS. 2013. Structure-function studies of the Neisseria gonorrhoeae major outer membrane porin. Infect. Immun. 81:4383– 4391. http://dx.doi.org/10.1128/IAI.00367-13.



jb.asm.org 547

The class 3 outer membrane protein (PorB) of Neisseria meningitidis: gene sequence and homology to the gonococcal porin PIA.

The major anaerobically induced outer membrane protein of Neisseria gonorrhoeae, Pan 1, is a lipoprotein.

The surface of Neisseria gonorrhoeae: isolation of the major components of the outer membrane.

The major outer membrane protein of Acidovorax delafieldii is an anion-selective porin.

Neutral sphingomyelinase 2 is a key factor for PorB-dependent invasion of Neisseria gonorrhoeae.

Role of pilA, an essential regulatory gene of Neisseria gonorrhoeae, in the stress response.

Crystal structure of the open state of the Neisseria gonorrhoeae MtrE outer membrane channel.

Pleiotropic transport mutants of Escherichia coli lack porin, a major outer membrane protein.

Cell envelope of Neisseria gonorrhoeae: outer membrane and peptidoglycan composition of penicillin-sensitive and-resistant strains.

Up-regulation of human neutrophil receptors for Neisseria gonorrhoeae expressing PII outer membrane proteins.

Altered outer membrane protein in different colonial types of Neisseria gonorrhoeae.

Isolation of Neisseria meningitidis mutants deficient in class 1 (porA) and class 3 (porB) outer membrane proteins.

Association of tetrapyrrole intermediates in the bacteriochlorophyll a biosynthetic pathway with the major outer-membrane porin protein of Rhodobacter capsulatus.

Neisseria meningitidis Lacking the Major Porins PorA and PorB Is Viable and Modulates Apoptosis and the Oxidative Burst of Neutrophils.

The serological classification of Neisseria gonorrhoeae. I. Isolation of the outer membrane complex responsible for serotypic specificity.

Transcriptional landscape and essential genes of Neisseria gonorrhoeae.

Penetrability of the outer membrane of Neisseria gonorrhoeae in relation to acquired resistance to penicillin and other antibiotics.

Bacterial entry and intracellular processing of Neisseria gonorrhoeae in epithelial cells: immunomorphological evidence for alterations in the major outer membrane protein P.IB.

Phase variation of the opacity outer membrane protein controls invasion by Neisseria gonorrhoeae into human epithelial cells.

Escherichia coli expressing a Neisseria gonorrhoeae opacity-associated outer membrane protein invade human cervical and endometrial epithelial cell lines.

Interactions of Neisseria gonorrhoeae with human neutrophils: studies with purified PII (Opa) outer membrane proteins and synthetic Opa peptides.

Neisseria gonorrhoeae acquire a new principal outer-membrane protein when transformed to resistance to serum bactericidal activity.

OT.

Phylogeny and nucleotide sequence of a 23S rRNA gene from Neisseria gonorrhoeae and Neisseria meningitidis.