Role of the Hrp Pilus in Type III Protein Secretion in Pseudomonas syringae

See allHide authors and affiliations

Science  21 Dec 2001:
Vol. 294, Issue 5551, pp. 2556-2558
DOI: 10.1126/science.1066397


Bacterial surface appendages called pili and needle-like filaments are associated with protein and/or DNA transfer to recipient plant, human, or bacterial cells during pathogenesis or conjugation. Although it has long been suspected that pili function as a conduit for protein or DNA transfer, direct evidence has been lacking. The Hrp pilus ofPseudomonas syringae is assembled by the type III secretion system. We used an in situ immunogold labeling procedure to visualize the extrusion of an effector protein, AvrPto, from the tip of the Hrp pilus, providing direct evidence that a bacterial pilus can function as a conduit for protein delivery.

Pili are nonflagellate filamentous surface appendages produced by virtually every species of Gram-negative bacterium. Pili are known to function as a means of attaching bacteria to various surfaces, often playing a crucial role in bacterial infection of humans and plants. Several pili are also required for delivery of effector proteins and DNA into recipient host or bacterial cells (1–4). Although the long and thin pili appear to be especially suited for long-distance protein and DNA delivery, direct evidence showing the extrusion of effector proteins or DNA from the tip of a pilus has never been shown.

The Hrp pilus of Pseudomonas syringae pv. tomatostrain DC3000 (hereafter Pst DC3000) is assembled by the hrp gene–encoded type III secretion system (4–8). This pilus is composed of a major subunit, the HrpA protein. The hrpA gene is required directly or indirectly for the secretion of effector proteins, such as AvrPto and HrpW, in culture (9). In situ immunogold labeling studies reveal that during active type III secretion, AvrPto and other effectors are localized specifically along the entire length of the Hrp pilus but not randomly in the extracellular space or along flagella (10, 11). Type III secretion thus occurs only at specific sites where the Hrp pilus is assembled. The deposition of AvrPto and other effectors along the entire length of the Hrp pilus suggests that secretion of effector proteins occurs while the Hrp pilus is being constructed, presumably because the expression of effector genes (such as avrPto) and type III secretion/Hrp pilus assembly genes (hrp/hrc) is co-regulated inPst DC3000, owing to the presence of a common “harp box” regulatory motif in the promoters of these genes (12–14). This co-regulation made it impossible to determine whether effector proteins exit from the tip or the base of a continuously growing pilus (11).

In this study, we uncoupled the expression of theavrPto gene from the assembly of the Hrp pilus by placingavrPto under the control of the P. aeruginosasalicylic acid (SA)–inducible PG promoter (15) in pAVRPTO (16). pAVRPTO was introduced into an avrPto deletion mutant, PstDC3000ΔavrPto (17). In PstDC3000ΔavrPto(pAVRPTO), AvrPto was produced and secreted in hrp-inducing medium only when SA was supplemented (Fig. 1A). As a periplasmic protein control, β-lactamase, expressed constitutively from the bla gene of pAVRPTO, was detected independently of SA in the cell fraction but was not released to the medium (Fig. 1B). Thus, in Pst DC3000ΔavrPto(pAVRPTO), AvrPto production was tightly controlled by SA, and the extracellular release of AvrPto was not due to nonspecific leakage.

Figure 1

Immunoblot analysis of AvrPto synthesis and secretion in Pst DC3000ΔavrPto(pAVRPTO). Bacteria were grown in hrp-inducing medium for 4 hours at 20°C without SA (lane 1), with 350 μM SA for 4 hours (lane 2), or with 350 μM SA only for the last 1 hour (lane 3). Cell (C) and supernatant (S) fractions were separated by centrifugation. Proteins were resolved in 12% SDS-PAGE gel and immunoblotted with AvrPto (A) and β-lactamase (B) antibodies according to the protocol described previously (25).

We next used in situ immunogold labeling to determine whether AvrPto is still localized specifically to Hrp pili under the SA induction condition (18). No significant level of gold particles (indicative of secreted AvrPto) was found when DC3000ΔavrPto(pAVRPTO) was grown inhrp-inducing medium without SA (Fig. 2A). In contrast, when PstDC3000ΔavrPto(pAVRPTO) was grown in thehrp-inducing medium supplemented with SA for 4 hours, AvrPto was abundantly secreted and was localized almost exclusively along a subpopulation of Hrp pili (Fig. 2B). We counted 95 Hrp pili that were attached to bacteria and labeled by the AvrPto antibody; they fell into two categories. Eighty-two Hrp pili were labeled along the entire length of the Hrp pilus. This category included those Hrp pili that were labeled uniformly or with more gold particles at the distal end, or occasionally at the proximal end. We also observed 13 Hrp pili with gold particles deposited along the Hrp pilus, except for the base (about one-tenth of the entire pilus length; occasionally close to one-third of the length). Thus, under the SA induction condition, AvrPto was still specifically localized to Hrp pili. However, the finding of the second category of labeling pattern, albeit at a low frequency, suggested that the assembly of at least some Hrp pili had already begun before secretion of AvrPto, even though bacteria were exposed to hrp-inducing medium and SA at the same time.

Figure 2

Representative EM images ofPst DC3000ΔavrPto (pAVRPTO), immunogold-labeled with the AvrPto antibody. In situ immunogold labeling (18) was done after bacteria were grown inhrp-inducing medium for 4 hours, supplemented with (A) no SA, (B) SA for 4 hours, or (Cand D) SA for the last 1 hour (chase induction). The Hrp pilus was not labeled in the absence of SA (A), was labeled throughout the Hrp pilus in the presence of SA for 4 hours (B), or was labeled along the top portion of the Hrp pilus under the chase induction condition (C and D). Dark dots are 15-nm gold particles in (A) and (C) and 10-nm gold particles in (B) and (D). Arrows indicate Hrp pili attached to rod-shaped bacteria (only a portion of the bacterium is shown, surrounded by dark stain). Scale bars, 100 nm.

We next examined the localization of AvrPto under conditions in which the Hrp pilus assembly was purposely induced first in thehrp-inducing medium for 3 hours, before a “chase” induction of AvrPto expression with SA for 1 hour (18). We reasoned that if AvrPto travels through the Hrp pilus and exits from the tip [the conduit model (11)], we should observe an increased number of Hrp pili with AvrPto localized near the tip. In contrast, if AvrPto exits from a hypothetical channel near the base of the Hrp pilus [the “guiding filament” model (11)], we should observe an increased number of Hrp pili with AvrPto localized near the base. We counted 100 Hrp pili that were attached to bacteria and labeled with the AvrPto antibody from four independent experiments. We found that the AvrPto labeling was drastically shifted to the tip portion of the Hrp pilus, as compared with that when the Hrp pilus assembly and AvrPto synthesis were co-induced (Fig. 2B). Seventy-two Hrp pili were labeled primarily along the top portion, among which were 42 Hrp pili with gold particles along the tip one-third portion of the Hrp pilus (Fig. 2C) and 30 Hrp pili with gold particles mainly along the top two-thirds of the Hrp pilus (Fig. 2D). Only 28 Hrp pili were labeled along the entire length. No Hrp pili with gold particles labeled exclusively at the base were found. Because each Hrp pilus was presumably at a distinct assembly stage at the time of SA induction, the uniform localization of AvrPto along the 28 Hrp pili suggested that these Hrp pili were assembled simultaneously with or after SA-induced AvrPto synthesis and secretion. In contrast, the assembly of Hrp pili that were labeled near the top portion must have begun before AvrPto synthesis and secretion. Consequently, the portion of the Hrp pilus assembled before AvrPto secretion was not labeled.

Under the chase immunogold labeling condition, the HrpA and HrpW antibodies still labeled the entire length of the Hrp pili (Fig. 3, A and B). This was best shown when HrpA and AvrPto antibodies were used in dual labeling experiments. Although the HrpA antibody bound to the entire length of the Hrp pilus, the AvrPto antibody labeled mainly along the tip portion of Hrp pili (Fig. 3C). This result is consistent with the fact that in PstDC3000ΔavrPto(pAVRPTO), the expression ofhrpA and hrpW genes was under the control of their native harp box–containing promoters, whereas the expression of AvrPto was controlled by SA.

Figure 3

Representative EM images of PstDC3000ΔavrPto (pAVRPTO), immunogold-labeled with the HrpA (A) or HrpW (B) antibody or dual-labeled with HrpA and AvrPto antibodies (C) under the SA chase induction condition (18). The HrpA and HrpW antibodies (indicated by 10-nm gold particles) labeled the entire Hrp pili (A and B). In (C), the HrpA antibody (indicated by smaller 10-nm gold particles) labeled the entire Hrp pilus, whereas the AvrPto antibody (indicated by larger 15-nm gold particles) labeled mainly the top portion of the pilus. Arrows indicate Hrp pili. Scale bars, 100 nm.

The drastic shift of AvrPto localization toward the tip of the Hrp pilus under the chase immunogold labeling condition directly supports the conduit model (11). The demonstrated role of the Hrp pilus as a conduit for protein secretion has important ramifications for further study of the type III secretion mechanism, because any future model must take into account the fact that effector proteins travel inside the Hrp pilus. The conduit function of the Hrp pilus also provides support for the postulated shared mechanism of type III secretion and flagellar assembly, which involves the export of partially folded flagellin through an internal channel of a growing flagellum filament before assembly at the tip (19,20).

  • * To whom correspondence should be addressed. E-mail: hes{at}


View Abstract

Navigate This Article