A mRNA Signal for the Type III Secretion of Yop Proteins by Yersinia enterocolitica

See allHide authors and affiliations

Science  07 Nov 1997:
Vol. 278, Issue 5340, pp. 1140-1143
DOI: 10.1126/science.278.5340.1140


Pathogenic Yersinia species have a specialized secretion system (type III) to target cytotoxic Yop proteins during infection. The signals of YopE and YopN sufficient for the secretion of translational reporter fusions were mapped to the first 15 codons. No common amino acid or peptide sequence could be identified among the secretion signals. Systematic mutagenesis of the secretion signal yielded mutants defective in Yop translation; however, no point mutants could be identified that specifically abolished secretion. Frameshift mutations that completely altered the peptide sequences of these signals also failed to prevent secretion. Thus, the signal that leads to the type III secretion of Yop proteins appears to be encoded in their messenger RNA rather than the peptide sequence.

Secretion of Yop proteins during the pathogenesis of human or animal infections allowsYersinia species to evade phagocytic killing by macrophages (1). After establishing contact with specific host cells,Yersinia target some Yop proteins directly into the eukaryotic cytosol where these virulence factors exert their cytotoxic functions (2-4). This type III secretion of Yop proteins is thought to occur as a continuous translocation of polypeptide across the inner and outer membranes of the bacterial envelope (5).Yersinia export 12 different Yop proteins by this pathway (1); however, no common secretion signal within the amino acid sequences of these polypeptides has been identified (6,7). This feature clearly distinguishes type III secretion from other export pathways in which the secretion signals of substrate proteins are readily apparent on the basis of common peptide sequences, structures, or physical properties (8-10). To determine whether Yop proteins are marked for secretion by a covalent posttranslational modification, we purified, sequenced, and measured the mass of secreted YopE. The results indicated that YopE is not modified upon export by the Yersinia type III machinery (11).

Several other Gram-negative pathogens also target their cytotoxic proteins into eukaryotic host cells (12). Bacterial contact with the target cell induces expression of the otherwise tightly regulated export machinery and secretion substrates (4). Many components of the type III machinery are highly conserved among Gram-negative bacteria (13). Substrate proteins from one organism can be exported by heterologous pathogens, suggesting a universal mechanism for secretion (14). The NH2-terminal 15 to 17 amino acids of Yop proteins have been proposed to function as a secretion signal; however, in the absence of a common peptide, it has been unclear how these signals can be universally recognized (7, 15). We therefore sought to characterize the secretion signal through genetic and biochemical means. We studied two type III secretion substrates, YopE and YopN, in Yersinia enterocolitica by analyzing translational fusions to cytoplasmic neomycin phosphotransferase (Npt) (16).

To identify the minimal secretion signal of YopN, we fused NH2-terminal coding sequences to Npt (17). Secretion of the hybrid proteins was measured by immunoblot analysis of the sedimented cells or medium of Yersinia cultures induced by temperature shift (37°C) and low calcium concentration (18). As reported for YopE (7, 15), the first 15 codons of YopN still allowed secretion of the fused reporter protein (19), whereas truncating this signal to 10 codons abolished secretion (Fig. 1). Relative rates of polypeptide synthesis were analyzed by pulse labeling and compared with those of another type III secretion substrate, YopH (20). Truncating the YopE signal to 10 codons caused a reduction in synthesis of the fusion protein. This reduction was apparently caused by an inhibition of translation, because the relative amount of mRNA for this fusion protein was similar to that observed for the Npt hybrid containing the first 15 codons of YopE (21). We investigated whether the secretion signal of YopE functioned when moved from the NH2-terminus by constructing a hybrid Npt protein that contained YopE fused to its COOH-terminal end. This hybrid was not secreted, indicating that the secretion signal is only functional when located at the translational start (Fig. 1).

Figure 1

Secretion signals located within codons 1 through 15 of YopE and YopN. Schematic diagram of hybrid proteins consisting of YopN, YopE, or their truncated derivatives with fused neomycin phosphotransferase (Npt). All constructs were expressed from their wild-type promoter (YopN or YopE) in Y. enterocolitica. Numbers refer to the respective codon positions. Secretion was measured by immunoblotting of the medium and cell sediment of induced Yersinia cultures and is reported as the percentage of total protein that is secreted. The relative synthesis of polypeptides [X]/[YopH] was analyzed by comparing immunoprecipitated substrates after pulse labeling with the amount of immunoprecipitated YopH. NT, not tested.

To determine if single amino acid residues of the YopE and YopN signals were critically important for secretion, we individually replaced codons 2 to 15 with GCA or GCU, which both specify alanine. The alanyl substitutions had little effect on secretion of the hybrid Npt proteins (Table 1). However, as measured by pulse labeling, YopE signal mutants with substitutions at positions 2 and 15 were synthesized at lower rates (<50%), and GCA replacement at codon 4 (YopE4S A-Npt) completely abolished translation. We also individually mutagenized codons 2 to 15 of the YopE signal by substitution with GAG encoding glutamic acid. Substitution of hydrophobic or positively charged amino acids with this strongly acidic residue did not affect secretion of the mutant proteins, but replacement of codons 2, 3, 4, 10, or 12 caused a reduction in polypeptide synthesis (<50%).

Table 1

Scanning mutagenesis of the secretion signal of YopE1–15-Npt (pDA46) and YopN1–15-Npt (pDA85). Individual codons of the wild-type sequence were replaced with those encoding either alanine or glutamic acid. The secretion data were collected by separating culture medium from sedimented cells and immunoblotting with anti-Npt. The relative amount of synthesized fusion protein [X] was measured by pulse labeling and is reported as the ratio to another type III secreted protein [YopH]. ND, not determined.

View this table:

We sought to identify mutations that abolished substrate recognition of the type III machinery by drastically modifying the polypeptide sequence of the secretion signals. We constructed frame-shift mutations by inserting or deleting nucleotides immediately after the AUG start codon. The correct reading frame was restored by reciprocal nucleotide insertions or deletions at the fusion site with Npt. The secretion signals of both YopE and YopN tolerated several frameshift mutations, and the altered polypeptides were still secreted (Fig.2). For YopE, deleting one nucleotide (−1) or adding two nucleotides (+2) did not prevent the secretion of hybrid proteins. In contrast, mutations shifting to the third reading frame (+1, −2) abolished secretion, and the Npt hybrids remained in the cytoplasm. This reading frame encodes a very hydrophobic NH2-terminal peptide, a physical property that may interfere with its secretion by the type III machinery (Fig. 2C). For YopN, the +1, −1, +2, and −2 reading frame mutants all allowed secretion. To test whether frameshift mutations resulted in altered amino acid sequences, we purified one mutant protein (YopE −1) from the medium of Yersinia cultures and confirmed the predicted sequence by Edman degradation (22).

Figure 2

Frameshift mutations of the secretion signals of YopN and YopE. Translational reading frameshifts were constructed by either deleting (−1, −2) or inserting nucleotides (A or G) (+1, +2) immediately after the AUG start codon of YopE1–15-Npt (A) or YopN1–15-Npt (B). The correct reading frame was restored by a reciprocal change at the fusion site with Npt. Secretion was measured by immunoblotting and is indicated as the percentage of secreted protein. Npt alone expressed from the YopE or YopN promoter was not secreted. [X]/[YopH] indicates relative levels of polypeptide synthesis as measured by pulse labeling and immunoprecipitation. The altered peptide sequences of the frameshift mutants are compared with those encoded by the wild-type secretion signals (C).

Because several frameshift mutations resulted in proteins that were secreted, we considered that the secretion signal might be located within the mRNA sequence. If this were true, nucleotide changes at the third position of codons that do not alter the protein sequence (23) might affect either secretion or translation of the hybrid Npt proteins. We tested codons 2 to 4 of YopN because these positions were sensitive to mutation in the secretion signal of YopE. Single-nucleotide changes at position 2, 3, or 4 of the YopN signal did not affect either translation or secretion (Table2). However, combined mutations at codons 2 and 4 reduced the amount of mRNA translation, which was restored to wild-type amounts when the mutant RNA contained all three altered codons (Table 2). The reduced concentration of the mRNA with mutations at codons 2 and 4 is likely caused by its increased degradation rather than by an effect on transcription (Table 2).

Table 2

Nucleotide changes at the third position of codon triplets that do not alter the protein sequence of the secretion signal were introduced into YopN1–15-Npt. Secretion was measured by immunoblotting of culture supernatants and cell sediment. The relative amount of synthesized fusion protein [X] was measured by pulse labeling and is reported as the ratio to another type III secreted protein [YopH]. [x]/[yopH] indicates relative levels of mRNA that were observed by RNA blot hybridization.

View this table:

Several mutations in the secretion signals of YopE and YopN either reduced or abolished synthesis of the recombinant proteins. These mutations may hinder Yop translation, for example, by interfering with ribosome binding or translational initiation. Alternatively, this mutational phenotype could represent a defect in the recognition of an mRNA signal that ultimately leads to the secretion of Yop proteins. Suppressor mutations that restore a translational defect of the YopE4S A-Npt mutant (pDA54, GCA replacing TCA at codon 4) should alter the mutant codon, whereas mutations that suppress a signal-recognition defect could also be located at other positions involved in contacting the secretion machinery. We selected spontaneous mutants by plating Yersinia enterocolitica harboring pDA54 on agar medium containing neomycin (24). Nine independent mutants were analyzed by immunoblotting; each of them synthesized and secreted the Npt fusion protein in a manner similar to that observed for the wild-type construct. These isolates were intragenic suppressors that contained mutations located at codons 2 through 6 and 12 (Table3). Transversion of the nucleotide at the third position of codon 12 (CCC to CCA) restored translation and thus secretion of the hybrid protein without an alteration of its amino acid sequence. This mutation was found in every suppressor isolate and was sometimes combined with mutations at codons 2, 4, or 5 or a deletion of codon 6. Although these results do not permit a definitive explanation, we think it is more likely that the mutational change at codon 4 abolished the recognition of an mRNA signal rather than causing a hindrance of translational initiation.

Table 3

Spontaneous suppressor mutations of YopE4S–A-Npt were selected by plating Y. enterocolitica harboring plasmid pDA54 (YopE4S–A-Npt) on tryptic soy agar plates with neomycin (50 μg/ml). Plasmid was purified from individual colonies and transformed into W22703, and plasmid transformants were selected on chloramphenicol plates. Individual isolates were tested for resistance to neomycin [minimal inhibitory concentration (MIC) for 105 cells], relative concentration of mRNA ([x]/[yopH]), synthesis ([X]/[YopH]), and secretion of hybrid Npt proteins. Mutational changes of the suppressors were determined by DNA sequencing. ND, not determined.

View this table:

Other secretion or protein-targeting signals do not tolerate such drastic mutational changes without a loss of function. The reason for this difference may reflect the mode of substrate recognition by the type III machinery. RNA may be the carrier of a signal that ultimately leads to the export of encoded Yop proteins. One possible mechanism is that the mRNA signals cotranslational secretion by the type III machinery. In support of this hypothesis, pulse-chase experiments ofY. enterocolitica cultures revealed that YopE was secreted during a short pulse with 35S-methionine but not after the addition of unlabeled methionine, suggesting that secretion occurred during the ribosomal synthesis of YopE (25). Yop translation might be inhibited by an intrinsic property of the mRNA that can be relieved by its interaction with the secretion apparatus. Most mutations that affect recognition of an RNA signal would therefore abolish both secretion and translation. An uncoupling of secretion from translation might result from larger deletions of the signal that destroy its structure. We have incorporated some of the mutations described here in predicted RNA structures of the YopE and YopN secretion signals (26) (Fig.3). Common to both structures is a stem loop that buries the AUG translational start in a base-paired duplex while positioning codons 2 to 4 within a loop. Mutations that abolished translation are located either within the predicted loop or its adjacent base pairs, that is, at positions typically recognized by RNA-binding proteins (27). Such an RNA structure would have to undergo dynamic changes because it would have to first assume an untranslatable fold, which could then be relieved by specific interaction with components of the secretion machinery.

Figure 3

Predicted RNA structures of the YopE and YopN secretion signals compared with that of the Npt mRNA. RNA sequences were subjected to folding analysis as described (26). The displayed structures show an area encompassing the Shine/Dalgarno ribosome binding site (filled squares), start codon (AUG, boxed), and downstream sequence of the YopE and YopN secretion signals [ΔG values (Gibbs energy) of −89.5 kj (YopE) and −102.9 kj (YopN)]. Nucleotides sensitive to mutation are circled. Mutations that abolished synthesis and secretion of reporter proteins are shadowed and their suppressors are indicated in bold.

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


View Abstract

Navigate This Article