ABSTRACT
Antigenic variation of the pilus is an essential component of
Neisseria gonorrhoeae
pathogenesis. Unidirectional recombination of silent pilin DNA into an expressed
pilin gene allows for substantial sequence variation of this highly immunogenic surface structure. While the RecA protein is required for pilin gene recombination, the factors
which maintain the silent reservoir of pilin sequences and/or allow
unidirectional recombination from silent to expression loci remain undefined.
We have previously shown that a conserved sequence at the 3
'
end of all pilin loci (the Sma/Cla
repeat) is required to be present at the expression locus for efficient
recombination from the silent loci. In this study, the binding of gonococcal
proteins to this DNA sequence was investigated. Gel mobility shift assays and
competition experiments using deletion derivatives of the repeat, show that
multiple activities bind to different regions of the Sma/Cla
repeat and define the boundaries of the binding sequences. Moreover, only the
pathogenic
Neisseria
harbor proteins which specifically bind to this repeat, suggesting a
correlation between the expression of these DNA binding proteins and the
potential to cause disease.
Gonococcal (Gc) pili are essential for bacterial colonization of the human
epithelium (
1
,
2
). The pilus, along with other gonococcal surface constituents (opacity proteins
and lipooligosaccharide), undergo structural variation (
3
-
6
), which produces numerous antigenic forms of the bacterium. Antigenic variation
is important in gonococcal pathogenesis since it promotes evasion of host
immune responses (
7
,
8
), and may also change functional characteristics of the bacterial cell (
9
-
11
).
Pilus antigenic variation occurs via changes in the pilin amino acid sequence (
12
-
16
). The chromosome of the gonococcal strain MS11 harbors either 1 or 2 complete,
expressed pilin genes (
pilE
) and 17 transcriptionally silent copies of partial pilin information (
pilS
) (
12
-
18
). Antigenic variation of the pilin protein occurs after segments of silent-copy pilin DNA recombine into the expressed pilin gene (
13
-
15
). The transfer of an entire silent copy is unusual. Instead, recombination of 30-150 silent copy nucleotides is more common, exponentially increasing the number of potential
expressed pilin gene sequences (
13
-
15
,
19
). Although recombination between
pilS
and
pilE
is facilitated by the RecA protein (
20
), gonococcal pilin gene recombination reactions predominantly generate non-reciprocal recombination products (
14
,
15
). The mechanism(s) responsible for promoting non- reciprocal recombination between
pilS
and
pilE
are unknown, although two models have been proposed. The extracellular model proposes that the non-reciprocal nature of the recombination reactions can result from homologous recombination of
pilS
sequences donated from lysed neighbor cells into the
pilE
locus of a live cell, through the natural exchange mechanism of DNA
transformation (
21
-
24
). The intracellular recombination model postulates that the
pilS
and
pilE
sequences which recombine to produce a variant pilin gene are present within
the same cell. The intracellular model necessitates more complex recombination mechanisms, requiring additional factors to ensure the unidirectional transfer of
pilS
DNA (
25
,
26
). Regardless of which model is utilized to generate recombinant
pilE
, we postulate that proteins in addition to RecA may function to: (i) limit the
extent of recombination tracts within
pilE
; (ii) limit
pilS
to
pilS
recombination to maintain a stable reservoir of variant pilin DNA; and/or (iii) ensure the non-reciprocal nature of
pilS
to
pilE
recombination.
All pilin loci possess a conserved DNA sequence at their 3' end (the Sma/Cla
repeat). We have previously shown that the Sma/Cla
repeat is important for pilin gene recombination since removal of this DNA
sequence from
pilE
reduces
pilS
to
pilE
recombination (
27
). This finding suggests that either the deletion blocks the action of one or
more proteins that act by binding this repeat, or the deletion reduces the
amount of flanking homology available for
pilS
to
pilE
recombination. The first hypothesis is favored for two reasons. First, specific
binding of multiple gonococcal proteins to the Sma/Cla
repeat has been observed (
27
). Second, several silent loci contain multiple silent copies (e.g.
pilS1
of strain MS11 has six copies), yet studies that examine the source of variant
pilS
DNA show no preference for silent DNA adjacent to the Sma/Cla
repeat (
14
,
28
-
30
). To determine how this DNA sequence functions in antigenic variation, we need
to know more about the specificity of protein binding to the repeat, and which
portions of the repeat are important in allowing efficient recombination.
The DNA sequence encoded by the Sma/Cla
repeat shows a high degree of sequence similarity to recombination protein binding sites from other systems. These include: (i) the recognition site (
hixR
) for the
Salmonella
recombination protein Hin; (ii) the recognition sequence (Tn
3
IR) for the Tn
3
transposon transposase and (iii) the recognition sequence (Tn
3
res) for the Tn
3
transposon resolvase (
31
; Fig.
1
). In these systems, DNA binding proteins function to cut or nick double
stranded DNA to promote subsequent recombination. The sequence similarities
between the Sma/Cla
repeat and these defined recombination protein binding sites may indicate a
conserved function, lending further support to the hypothesis that proteins involved in pilin gene recombination bind to the Sma/Cla
repeat.
Neisseria gonorrhoeae:
VD300-
recA6
(P
+
, [Delta]
pilE2, recA6
) (Seifert H.S. 1997, in press). Other
Neisseriae: N.meningitidis
(strain FAM18) (
32
);
N.cinerea
(strain NNRL9);
N.lactamica
(strain NNRL30011); and
N.sicca
(strain NRL9989). All commensal strains were a generous gift from J. Cannon.
Escherichia coli:
DH5[alpha]MCR (Gibco).
Neisseriae
were grown either: (i) on Gc medium agar base (GCB) (Difco) plus Kellogg
supplements (
33
) at 37oC in 5% CO
2
; or (ii) in Gc liquid growth medium (GCBL) (1.5% Protease Peptone #3; 0.4% K
2
HPO
4
; 0.1% KH
2
PO
4
; 0.1% NaCl) plus Kellogg supplements and 0.042% NaHCO
3
at 37oC in a shaking water bath.
Deletions of the 66 bp
Sma
I-
Cla
I
fragment were produced in plasmid pUCSC1, which has a
Sma
I-
Hin
dIII fragment from pNG1312 (
14
) containing the Sma/Cla
repeat inserted between the same sites of pUC18. All enzymes were from New
England Biolabs except as noted. Enzyme were used according to manufacturers'
directions except as noted.
Deletions from the 5' end of Sma/Cla
were generated by treating 8 [mu]g aliquots of
Sma
I digested pUCSC1 with 2 U of Exonuclease III at 30oC for various times. The reactions were stopped by addition of cold S1
nuclease reaction mixture (final concentration: 1 U S1 nuclease [USB], 30 mM
NaOAc pH 5.0, 50 mM NaCl, 1 mM ZnSO
4
). Cleavage of single stranded overhangs with S1 nuclease was performed at room
temperature for 30 min. S1 nuclease was inactivated by addition of 0.1
vol of 300 mM Tris (pH 8.0), 50 mM EDTA, and incubation at 70oC for 30 min. The DNA was purified from buffers and proteins by spin
dialysis through Sepharose CL-6B (Sigma). After
Not
I linker ligation, the plasmid DNA was digested with
Not
I, spin-dialyzed through CL-6B, self-ligated with T4 DNA ligase, and transformed into DH5[alpha]MCR. Deletions were sized by digesting resultant
plasmid with
Not
I and
Cla
I and analyzing the smaller DNA fragment on 4% Metaphor (FMC)-TBE gels, and further defined by fMol (Promega) cycle sequencing using
standard M13 primers.
Deletions from the 3'-end of Sma/Cla
were constructed as follows. Aliquots of 7 [mu]g
Cla
I digested pUCSC1 were treated with 1 U of BAL31 nuclease at 22oC for various times and the reactions stopped with 20 mM EGTA. This was
followed by spin-dialysis through Sepharose CL-6B, repair of ends with DNA polymerase I large fragment, ligation of
Not
I linkers, digestion with
Not
I, self-ligation with T4 DNA ligase, and transformation into DH5[alpha]MCR.
For use as competitors in mobility shift experiments, the Sma/Cla
deletion subclones were digested with
Sma
I and
Not
I, each fragment was separated from the vector by electrophoresis through 4.0%
Metaphor agarose-TBE gels, and purified using [beta]-agarase I followed by phenol/chloroform extraction.
Concentration of each fragment was determined by running 2-fold dilutions on a 4% Metaphor gel and comparing band intensities to
known standards.
A quantity of 5.4 or 10.8 [mu]g total protein was mixed with binding cocktail (
27
), 50 [mu]g/ml p[d(I
.
C)] carrier DNA, and increasing molar concentrations (0, 30-, 100-, 300- and 900-fold molar excess) of competitor DNA. The reactions were
incubated at 4oC for 10 min on a platform shaker. An aliquot of 0.5 ng of isolated,
undeleted
Sma
I-
Not
I fragment (labeled at the
Not
I site with [[alpha]-
32
P]dGTP and [[alpha]-
32
P]dCTP using DNA polymerase I large fragment) was added to the binding reaction and incubated for an additional 10 min at 4oC. The binding reactions were resolved through 6% PAGE as described (
27
).
Densitometry was performed on the dried radioactive PAGE gels using the Fuji BAS
2000 phosphoimager using whole band counting analysis and subtracting
background from an area of the gel not containing any radioactive counts.
Autoradiographs were scanned using a BioRad Imaging Densitometer and Molecular
Analyst 1.1.1 for Windows Software (BioRad). Scanned images were processed
using Photoshop 2.5 for Window (Adobe) and Corel Draw 6 (Corel).
Our preliminary analysis of protein binding to the Sma/Cla
repeat showed that different salt fractions from a DEAE-column contained Gc proteins that retarded the mobility of the Sma/Cla
repeat. To more specifically examine these Sma/Cla
binding activities, a late log-phase liquid Gc culture was partially fractionated over DEAE-anion exchange resin. Examination of eluted proteins on SDS-PAGE showed changes in the protein profile with increasing
salt concentrations indicating successful fractionation (Fig.
2
D), although each fraction contained numerous protein species. The protein
fractions were assayed for Sma/Cla
repeat binding activity in the presence of p[d(I
.
C)] carrier DNA, and were resolved through PAGE to separate the bound and
unbound DNA (Fig.
2
B). Results from this analysis identified several DNA binding proteins present
in these fractions and five distinct mobility shifts were defined. The first
DNA binding activities were designated IE (relative mobility I, early
fractions) and IL (relative mobility I, late fractions). Two additional
activities that retarded the fragment's migration to different but similar
intermediate levels were identified in the first salt fractions IIE (relative
mobility II, early fractions) and in the later salt fractions IIL (relative
mobility II, late fractions). The final activity eluted at relatively high salt
concentrations and showed the least change in mobility relative to the unbound
fragment (relative mobility III). Selected fractions that harbored the
representative binding activity(s) from the 200 and 800 mM fractions were
chosen and combined, and the proteins present in these fractions were desalted
and concentrated such that equal amounts of total protein could be used in each
binding reaction.
To define the region of the Sma/Cla
repeat recognized by the different gonococcal proteins, exonuclease-derived, deletion subclones were constructed that contained different
portions of the Sma/Cla
repeat. Four subclones containing either the first 33 or 27 nucleotides of the 5' end of the Sma/Cla
repeat (5-33 and 5-27, respectively) or 31 or 26 nucleotides of the 3' end of the repeat (3-31 and 3-26, respectively) were used for this study (Fig.
1
). In mobility shift assays using radiolabeled deletion fragments, each of these
fragments was bound by one or more proteins present in the Gc lysate fractions
and produced gel mobility shifts. However, it was not possible to correlate
these changes in binding pattern between the mobility shifts of intact and
deleted Sma/Cla
fragments (data not shown).
To determine the DNA sequences of the repeat responsible for the protein
binding, fractionated, concentrated and desalted protein samples were incubated
with increasing molar-excess amounts of the four deletion derivatives in the presence of
radiolabeled, full-length Sma/Cla
fragment. Results from this analysis showed that the different protein species isolated from DEAE fractionation had bound to different regions of the Sma/Cla
repeat. To quantitate the results from the competition experiments, several
repeats were performed and the averaged data reported as the percent of binding
of radiolabeled Sma/Cla
DNA in the presence of competitor relative to the binding of radiolabeled
Sma/Cla
DNA in the absence of competitor (Fig.
3
).
Horizontal genetic exchange is significant in the epidemiology and evolution of
Neisseria
species. Evidence exists to suggest that a continuous flow of genetic material
affects the chromosomal composition of both pathogenic and commensal
Neisseria
(
34
,
35
). To assess the distribution of Sma/Cla
binding proteins in the
Neisseria
genus, protein fractions were generated from the pathogenic
N.meningitidis,
and the non-pathogenic
N.cinerea, N.lactamica
and
N.sicca,
and tested in mobility shift experiments. Experiments that examined protein binding to either a radiolabeled Sma/Cla
repeat or a radiolabeled control DNA fragment showed specific binding in only
the pathogenic
Neisseriae
. As was observed with the gonococcus,
N.meningitidis
(Mc, Fig.
4
) produced a complex pattern of DEAE-fractionated proteins binding to the Sma/Cla
repeat as measured by gel mobility shifts. The proteins present in the Mc lysate
produced shifts of similar elution profile and r.m. to the Gc shifts, IL and
IIL and III. The proteins producing shifts IE and IIE were not detected in the
Mc lysates. Consistent to what was observed with Gc (
27
), the Mc proteins did not produce detectable binding to the radiolabeled
control DNA fragment.
Binding was also observed with both the Sma/Cla
repeat and the control DNA in the commensal species; however, it was largely non-specific (Fig.
4
).
Neisseria sicca
harbored proteins which bound to both the Sma/Cla
repeat and the non-Sma/Cla
control fragment (indicated by * and +, Fig.
4
D), showing that the proteins were not specific Sma/Cla
binding proteins. Similarly,
N.lactamica
also expressed proteins which bound to both DNAs (indicated by > and #, Fig.
4
C). While low level binding to the Sma/Cla
repeat was detected in a single fraction from both
N.cinerea
and
N.lactamica
(Fig.
4
B,
*
Sma/Cla-lane 4, and Fig.
4
C,
*
Sma/Cla-lane 5), the low abundance suggests that these activities do not
efficiently recognize the Sma/Cla
repeat. Interestingly, comparative enzyme electrophoresis experiments have shown
that both
N.cinerea
and
N.lactamica
are more closely related to the pathogenic
Neisseria
species than any other commensal species (
36
).
Southern blot analysis demonstrated that of the species tested, only pathogenic
species harbored DNA that hybridized to both conserved pilin probes and probes
specific for the Sma/Cla
repeat (data not shown). Also, an
E.coli-
K-12, DEAE-fractionated cell lysate, exhibited no specific Sma/Cla
binding activities (data not shown). Finally, Southern blot analysis showed that
no DNA was present in the
E.coli
genome that hybridized with either Sma/Cla
or conserved pilin oligonucleotide probes (data not shown).
The discovery that only pathogenic
Neisseria
harbor specific Sma/Cla
binding is consistent with a function that is important for pathogenesis such as
antigenic variation. Since Southern blot analysis showed that only Mc and Gc
harbored pilin and Sma/Cla
DNA, one could argue that the presence of both the Sma/Cla
repeat and its binding proteins have been conserved through evolution. It is
possible that the commensal species have DNA binding proteins specific for
other repeat sequences different from the Sma/Cla, but since there is no
evidence for a multi-locus antigenic variation system in the nonpathogenic species, it is
unlikely they have need for such a system.
The gonococcus expresses four distinct activities that fractionate
differentially and/or bind to different sequences of the Sma/Cla
repeat. The meningococcus expresses detectable amounts of three of the four
activities, while the comensal species tested do not express similar
activities. The identification and definition of these Sma/Cla
binding activities implies a complex series of protein-DNA interactions that influence the transfer of DNA sequences between
pilin loci in Gc and Mc.
This work was supported by NIH grant AI33493. J.F. was supported by NRSA AI09048
and H.S.S. was a recipient of an American Cancer Society, Junior Faculty Award.
*To whom correspondence should be addressed. Tel: +1 312 503 9788; Fax: +1 312
503 1339; Email: h-seifert@nwu.edu
REFERENCES
Return