The Needle Length of Bacterial Injectisomes Is Determined by a Molecular Ruler

See allHide authors and affiliations

Science  05 Dec 2003:
Vol. 302, Issue 5651, pp. 1757-1760
DOI: 10.1126/science.1091422


Size determination represents a fundamental requirement for multicomponent biological structures. Some pathogenic bacteria possess a weapon derived from the flagellum. Like the flagellum, this type-III secretion apparatus, called the injectisome, has a transmembrane basal body, but the external component is a needle-like structure instead of a hook and a filament. Here, we provide evidence that the length of this needle is determined by the size of a protein, YscP, acting as a molecular ruler.

Yersinia pestis and Y. enterocolitica, the infectious agents of bubonic plague and gastroenteritis, respectively, share a common plasmid-encoded type-III secretion system consisting of the Ysc (Yop secretion) injectisome and the Yops (Yersinia outer proteins) that are secreted by this apparatus (1). The injectisome, made of 27 Ysc proteins, is thought to resemble those of Salmonella enterica and Shigella flexneri. These injectisomes, or “needle complexes,” appear as two pairs of rings that are anchored to the inner and outer membranes of the bacterial envelope, joined by a central rod and supporting a hollow needle about 10 nm thick and 60 nm long (24). It is thought that the injectisome serves as a hollow conduit through which the secreted proteins travel across the two bacterial membranes and the peptidoglycan in one step.

Several Ysc proteins that are anchored in the inner membrane and form the core of the secretion apparatus are similar to proteins from the basal body of the flagellum, suggesting a common evolutionary origin (5). Not surprisingly, the Salmonella and Shigella injectisomes resemble the flagellar basal body (6) except that they are topped by a needle instead of a hook and a flexible filament. The length of the flagellar hook (55 nm) is genetically controlled. Mutations in the gene fliK give rise to hooks of indefinite length (7), but it is unclear how FliK exerts its control. The fact that all truncated FliK proteins engineered so far lead to longer hooks rather than shorter hooks is presented as an argument that FliK cannot act as a simple molecular ruler (8). In addition, certain mutations in genes that encode the cytoplasmic ring lead to shorter hooks (8), supporting an elegant model in which this structure controls the hook length by acting as “a quantized measuring cup,” storing the subunits before their export (8). In this model, the role of FliK would be to terminate hook elongation by changing the secretion mode from hook-monomer to flagellinmonomer secretion (8). As for the injectisome of Salmonella and Shigella, mutants affected in a gene called invJ or spa32, respectively, display needles of various lengths, ranging from normal (60 nm) to as long as 1 μm (2, 9, 10). Thus, InvJ and Spa32 behave as FliK homologs, although they do not share any substantial sequence homology with FliK.

Here, we address the question of what controls the injectisome needle length in Yersinia. We first examined by electron microscopy Y. enterocolitica E40 bacteria incubated under conditions that artificially induce secretion of the Yops (11). We observed many needle-like structures that were about 6 to 7 nm thick and 60 nm long (12). Many of these needles were detached from the bacterial body (Fig. 1). We purified these detached needles and confirmed that they were made of the 6-kD YscF protein (12). Because it is difficult to define the exact insertion point of needles on bacteria, we measured only the detached needles. The length was distributed with an average of 58 ± 10 nm (Fig. 1), suggesting that the needles either detached or broke at a precise point close to the bacterial surface. Next, we examined Y. enterocolitica with a large deletion (codons 97 to 465 out of 515) in yscP (yscPΔ97–465) (13). This gene is synthenic to spa32 and invJ, but its product has no substantial sequence identity with Spa32, InvJ, or FliK. The yscPΔ97–465 mutant bacteria produced needle-like structures with an indefinite length ranging from 45 nm up to 1570 nm (Fig. 1). When the yscPΔ97–465 mutation was complemented with the yscP+ allele, control of the length was restored (55 ± 8 nm), indicating that YscP played an essential role in length control.

Fig. 1.

YscP is required for needle-length control. Electron micrographs of Y. enterocolitica wild type (wt) (A) and yscPΔ97–465 mutant (C) showing the needles of the injectisomes. Detached needles were measured at the vicinity of at least 10 to 15 different bacteria. Histograms of lengths are given in (B) (for wild type) and (D) (for yscPΔ97–465 mutant). Note the altered distribution of lengths in the mutant. , mean of the lengths; N, number of needles measured.

YscP from Y. enterocolitica E40 (YscPentero) carries a duplication of 60 central residues (13) (Fig. 2). YscP from Y. pestis KIM5 (YscPpestisKIM5) is 90% identical in sequence to YscPentero, but it is shorter (455 residues) because of the lack of such repetition (14). To explore whether the two proteins lead to needles of the same length, we complemented the Y. enterocolitica yscPΔ97–465 mutation with the yscPpestisKIM5+ gene (15). The shorter Y. pestis gene restored length control but programmed shorter needles (41 ± 8 nm) (Fig. 2).

Fig. 2.

YscP is shorter in Y. pestis than in Y. enterocolitica and determines shorter needles. (A) Residues 222 to 236 and 261 to 306 from YscP are duplicated in Y. enterocolitica E40 and W22703 but not in Y. pestis KIM. The yscP gene from Y. pestis KIM and the yscP gene from Y. enterocolitica deprived of these repeats were cloned downstream from an arabinose-inducible promoter and expressed in the Y. enterocolitica yscPΔ97–465 mutant. The figure marked with an asterisk includes a few amino acids inserted to generate the deletion. (B) Histograms of the needle lengths from recombinant Y. enterocolitica bacteria.

To investigate whether the needle length reduction was a result of the shortening of YscP and not subtle residue changes, we complemented the yscPΔ97–465 mutation with the allele yscPenteroΔ222–306+ encoding YscPentero without its repeat. The truncated YscPenteroΔ222–306 programmed short needles (42.5 ± 8 nm) (Fig. 2), suggesting that the needle length indeed correlated with the size of YscP. To further investigate this hypothesis, we engineered a set of deletions within the cloned yscPentero gene and used them to complement the yscPΔ97–465 mutation (Fig. 3). Proteins truncated within the first 35 or the last 130 residues were unable to control the length, even though their expression levels were comparable to that of the wild type (Fig. 3). In contrast, YscP with deletions up to 126 amino acids between residues 36 and 306 retained length control but programmed shorter needles. Moreover, the length of the needles was proportional to the size of the YscP protein (Fig. 3). We then inserted a second copy of residues 307 to 381 or 222 to 381 after residue 49, thus generating YscP proteins containing the same 60-residue sequence three or four times (Fig. 3). These mutants programmed longer needles with lengths of 72 ± 10 nm and 88 ± 12 nm, respectively. Insertion of residues 147 to 265 from FliK at the same position also resulted in a functional protein but longer needles (75 ± 14.5 nm). Thus, a strict linear relationship existed between needle length and the number of amino acids in YscP, with 1.9 Å per YscP residue (Fig. 3). To rule out bias as a result of an inadequate gene dosage or gene expression of the complementation plasmids, we replaced the wild-type yscP allele on the virulence plasmid by two truncated alleles (yscPΔ46–96 and yscPΔ222–306). In this native genetic environment, the two alleles again programmed shorter needles than those of the wild type (47.5 ± 10 nm and 44 ± 8.5 nm, respectively). Thus, YscP appeared to serve as a ruler determining the length of the needle. This is unprecedented in bacteria but it evokes the molecular rulers controlling the length of bacteriophage tails (1618), which are structures resembling the needle in morphology, size, and even function. A difference between the two systems is that tails do not assemble in the absence of the ruler whereas needles are of undetermined length in the absence of YscP. The length per residue is also slightly different for tail rulers (1.5 Å per ruler residue).

Fig. 3.

The needle length is proportional to the number of residues in YscP. (A) Various deletions and insertions were introduced in yscPenteroE40, cloned downstream from the arabinose promoter. Insertions occurred in a restriction site inserted after residue 49 (Ω). The dotted arrow spans the region where deletions and insertions modify the needle length. (B) Needle length measurements. N. aa, number of amino acids; Nr, number. (C) Expression of the various constructs in Y. enterocolitica yscPΔ97–465, monitored by Western blotting. (D) Plot of the lengths versus the number of residues in YscP.

If YscP acts as a ruler measuring the growing needle, one might expect it to be associated with the needle, at least during the needle elongation stage. To test this, we purified needles from Y. enterocolitica incubated in conditions that are either nonpermissive or permissive for secretion (11). Under nonpermissive conditions, some YscP was found in the needle fraction as well as in the culture supernatant, whereas under secretion-permissive conditions YscP was found only in the culture supernatant and not in the needle fraction any more (Fig. 4). These data, fitting with previous reports on the localization of YscP (13, 14), show that YscP is associated with newly synthesized needles that are not secreting Yops but not with needles that are secreting Yops.

Fig. 4.

Association of YscP with the needles. Y. enterocolitica E40 deprived of the effectors (ΔHOPEMT) was grown in conditions that are (S+) or are not (S–) permissive for Yop secretion (11). Detached needles purified from 2 × 109 (S+) or 1010 (S–) bacteria were analyzed by silver stained SDS–polyacrylamide gel electrophoresis (SDS-PAGE) and Western blotting (WB) (left). Needle samples prepared from S+ bacteria contain YscF (bottom arrows) and flagellin (middle) as a contaminant but no YscP (top). In contrast, needle samples prepared from S– bacteria contain YscF, YscP (top arrow), flagellin, and unidentified contaminants (dots). Proteins from the supernatant from 108 bacteria from the same cultures were analyzed by Coomassie-stained SDS-PAGE and WB (right). YscP is visible as a faint band in both the S+ and the S– supernatants (arrows). The darkest band stained in the S+ supernatant is YopB+D. YscF is not visible by Coomassie staining. The prominent bands in the supernatant of S– are cellular contaminants.

We propose that YscP controls the length of the needle by acting as a molecular ruler during the stepwise assembly of the injectisome. Because deletions affecting both N- and C-termini of YscP lead to a loss of length control, we hypothesize that the two ends of YscP act as anchors. One end would be attached to the basal body, whereas the other would be connected to the growing tip of the needle. Whatever the anchor mechanism may be, when the needle reaches its mature length, YscP would be fully stretched and signal, via its internal anchor, to the secretion apparatus, which would stop exporting YscF and switch to other substrates. This model (fig. S1) does not contradict the switch function of YscP (8, 19, 20) but rather includes it in a more complex dual function, which may also exist in some phage tail rulers (21). Taking into account the length of 1.9 Å per residue, the ruler domain of YscP would consist of about 300 to 350 residues, leaving more than 150 residues for anchoring and signaling. The fact that YscP is secreted also fits the model, because an internal ruler would be expected to obstruct the 2- to 3-nm-wide secretion channel (22). This evokes again the phage tail rulers, which are thought to exit the tail before the tail exerts its function (21, 23). Given the similarity between all the type-III secretion systems (5) and the fact that Spa32 (9, 10) InvJ (24), and FliK (25) are also secreted proteins, it is likely that the mechanism demonstrated here for YscP may apply to the control of the needle length in the other bacteria as well as the length of the flagellar hook. The proposed organization of FliK in three regions— export, hinge(147–265), and switch (25)—is also compatible with this view. Finally, the fact that YscP, InvJ, Spa32, and FliK diverged more during evolution than other proteins from type-III secretion suggests that rulers are subjected to fewer constraints. They nevertheless have to share intrinsic properties still to be discovered.

Supporting Online Material

Materials and Methods

Fig. S1


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article