Type III Secretion: More Systems Than You Think

Paul Troisfontaines, Guy R. Cornelis


The type III secretion (T3S) pathway allows bacteria to inject effector proteins into the cytosol of target animal or plant cells. T3S systems evolved into seven families that were distributed among Gram-negative bacteria by horizontal gene transfer. There are probably a few hundred effectors interfering with control and signaling in eukaryotic cells and offering a wealth of new tools to cell biologists.

What Is Type III Secretion?

Exotoxins, in particular A-B toxins, which allow bacteria to destroy or disturb animal cells at a distance from the site of infection or colonization, have been known for a very long time. In contrast, type III secretion (T3S), which allows direct communication between bacteria and animal or plant cells, was only discovered in the 1990s. With T3S, a bacterium docked at the surface of a cell or even residing within a vacuole injects effector proteins into the cytosol across the cellular or vacuolar membrane and modulates the cell functions to its advantage (30, 31, 51, 66, 74, 112, 124). The T3S apparatus, called the “injectisome,” consists of two pairs of rings spanning the two bacterial membranes, linked by a rod, and a ~60-nm-long needle protruding outside the bacterial body (10, 81, 84) (FIGURE 1). The length of the needle is genetically controlled (78) and is adjusted in relation to the length of other macromolecules present at the bacterial and cellular surfaces (95). The injection process requires the presence of proteins called “translocators” (9, 15, 16, 36, 61, 75, 113, 129), which form a pore into the target cell or vacuolar membrane. Whether the translocators are present at the tip of the needle before contact with the target cell or whether they are secreted just before the effectors is not yet clear. The requirement for translocators shows that the primary function of the needle is to contact target cellular membranes rather than to pierce these membranes (95).


A: a resting T3S injectisome with the 2 rings spanning the membranes and the needle protruding outside the bacterium. Effectors and translocators are stored. B: a T3S injectisome in operation. The translocators form a pore into the target cell membrane, and the effectors are translocated into the cytosol of the target cell. C: electron micrograph of the surface of Yersinia enterocolitica with protruding needles. Image courtesy of L. Journet (Biozentrum der Universität Basel).

The injectisome and the translocators are generally encoded within a single gene cluster located on a mobile genetic element—plasmid or pathogenicity island (60)—that is absent from the genome of related nonpathogenic bacteria. The recent sequencing of many bacterial genomes has shown that T3S is quite common among proteobacteria and Chlamydiae (101).

The aim of this review is to shed light on the many T3S systems that have been recently discovered by genome mining and to show how they relate to the archetypal systems described in Yersinia, Salmonella, Shigella, and Xanthomonas campestris.

How Did T3S Evolve?

The injectisome has a common evolutionary origin with the bacterial flagellum, as suggested by the fact that the basal bodies of both nanomachines are similar and include many proteins with a significant similarity. Moreover, there are also functional similarities, such as the use of specific chaperones (5, 45, 104, 108, 138) or of molecular systems regulating the size of the external appendage (the hook in the case of flagellum and the needle in the case of the injectisome) (93). Although the flagellum is a motility organelle with a rotary motor powered with the electron-motive force, it contains a built-in secretion apparatus that serves to export its own distal elements such as the hook and filament subunits before they polymerize. Interestingly enough, in some cases, the flagellum has been shown to secrete proteins that are not flagellar, showing that its secretion apparatus can easily be reassigned to pure secretory tasks (83, 85, 142). Like the flagellum, the injectisomes export their own distal components (needle subunits) before they secrete the substrates that represent their raison d’être. Thus the flagellum and the injectisome are evolutionarily related nanomachines sharing a common secretion apparatus and a similar assembly process. In the strictest sense, the secretion pathway that is common to the flagellum and injectisomes is called T3S. However, the term is often used to qualify the injectisome plus its substrates. Here we will refer to the injectisomes and their substrates as “nonflagellar T3S systems” (NF-T3S).

Since injectisomes are dedicated to trans-kingdom communication, they must have appeared after the first eukaryotes. In contrast, the urge to move toward nutrients in the primitive soup was a more basic need; hence injectisomes are likely to be derived from the flagellum. In phylogenetic studies (49, 56, 96), the flagellar components always cluster on a branch distinct from their homologs from the injectisomes, showing that no exchanges have occurred between the two systems since they diverged.

The phylogenetic analyses carried out on different constituents of the injectisome lead to the same tree, whatever the component used for computing, showing that all of the genes encoding injectisomes have evolved as large, intact genetic blocks. These trees identify seven families of injectisomes (FIGURE 2B, Table 1). Within the seven families, not only are the genes conserved, but so is the genetic organization of the loci (syntheny). The most striking feature of this evolutionary tree is that it is completely different from the bacterial evolutionary tree based on 16S rRNA (59) (FIGURE 2A). This implies that T3S systems evolved by lateral genetic transfers. Several bacteria have two T3S systems, and when two systems coexist in one bacterium, the loci are generally composed by xenologs. Hence the concomitant presence of two T3S in a species is never the result of gene duplication inside the species but rather the result of two successive horizontal gene transfers.

View this table:
Table 1.

Bacteria, their ecology, and their T3S systems


Phylogentic tree of bacteria T3S systems
A: rRNA tree made with aligned sequences from the Ribosomal Database Project II ( The main phylogenetics groups are mentioned on the branches. B: relationship phylogram of the ATPases of injectisomes and FliI, the ATPase of the flagellum of E. coli. The 7 T3S families are represented in 7 different colors; the same colors are used in the rRNA tree to illustrate the lack of correlation between the 2 trees. The bacteria that are shown with mixed colors in the rRNA tree posses 2 or 3 injectisomes from different families. The Eds T3S of Edwardsiella tarda (Ssa-Esc family) and the Cpi-1 T3S of Chromobacterium violaceum (Inv-Mxi-Spa group) are not represented in the tree because of the lack of sequences when the tree was drawn.

As for the effectors of T3S, they are often encoded outside of the genetic block encoding the injectisome, and they vary significantly from system to system, even within families of T3S. They sometimes share functional similarity with eukaryotic proteins, suggesting that they may have been acquired from the eukaryotic cells (Table 2). The list of effectors is far from being complete, but their diversity already appears to be very large, offering a wealth of new tools to the cell biologist.

View this table:
Table 2.

Nonexhaustive list of T3S effectors

The Main Families of T3S Systems

The Ysc family (Ysc, Psc, Lsc, Asc, Vsc, Dsc, Bsc)

The Ysc family is named after the archetypal Ysc (for Yop secretion) injectisome from Yersinia spp. (Table 1 and FIGURE 2B). It includes the Psc system of Pseudomonas aeruginosa, the Lsc system of Photorhabdus luminescens, the Asc system of Aeromonas spp., and the Vsc system of Vibrio parahaemolyticus. The Bsc system of Bordetella spp. and the Dsc T3S system of Desulfovibrio vulgaris could form a subgroup within this family, but this is a matter of debate (56, 101)

The plasmid-encoded Ysc injectisome is common to the Yersinia spp. Y. pestis, the agent of bubonic plague, and Y. pseudotuberculosis and Y. enterocolitica, two agents of gastroenteritis and mesenteric lymphadenitis. The role of this system in pathogenesis has been extensively studied over the past ten years. The Ysc injectisome delivers six effectors, YopE, YopH, YopO (YpkA), YopT, YopM, and YopJ (YopP), which inhibit the cytoskeleton dynamics (and hence phagocytosis) and prevent the onset of the inflammatory response (Table 2) (for a review, see Ref. 29).

P. aeruginosa is a versatile pathogen, virulent for nematodes, plants, insects, and mammals (11, 76, 94, 106, 128). For humans, it is an opportunistic pathogen causing fatal lung infection in cystic fibrosis and immune-compromised patients and causing severe infections of burn wounds. The Psc T3S system is only one weapon among a large arsenal encountered in P. aeruginosa. The four effectors, ExoS, ExoT, ExoY, and ExoU (Table 2) lead to disruption of the actin cytoskeleton, cell cytotoxicity, and apoptosis. They also interfere with cell-matrix adherence (23, 47, 91, 117, 139, 140).

P. luminescens is an entomopathogen that lives in a mutualistic association with nematodes that are pathogenic for insects. Once the nematode has invaded a potential host, the bacteria are released from the nematode’s gut into the blood-circulatory system (hemocoel) of the insect, where it produces a large variety of virulence factors. The Lsc T3S system allows bacteria to survive in the insect hemocoel and resist phagocytosis by insect macrophages (13). LopT, a homolog of YopT (Table 2) was recently identified as an effector of the Lsc system (13).

Aeromonas salmonicida is an important fish pathogen causing a fatal disease in Salmonids, and its Asc system has been shown to play a crucial role in pathogenesis (14). Although A. salmonicida is exclusively a fish pathogen, A. hydrophila is an opportunistic pathogen for a variety of aquatic and terrestrial animals, including humans, where it causes septicemia, gastroenteritis, and wound infections. The Asc T3S has been shown to be involved in cytotoxicity for fish cells and to be required for full virulence in the gourami fish (143). The Asc T3S is also present in A. veronii and A. caviae (20).

The Vsc T3S system is found in several marine Vibrio spp. that are found in association with plankton and juvenile shrimps, prawns, lobsters, and fishes (Table 1) (87, 102). V. parahaemolyticus is an emerging human pathogen causing gastroenteritis, particularly in areas where seafood consumption is prominent. The Vsc T3S system is found in environmental as well as in human pathogenic strains. Like P. luminescens, Vibrio may use its T3S system in the hemocoel of marine arthropods.

D. vulgaris is an anaerobic, sulfate-reducing bacterium, more studied for its bioremediation ability than for its potential pathogenicity. It is found in the human digestive tract and has been reported to be responsible for the development of ulcerative colitis (62).

Bordetella spp. B. pertussis, B. parapertussis, and B. bronchiseptica, which are responsible for mammalian respiratory tract infections, share a highly similar T3S system called Bsc. At least in B.bronchiseptica, the Bsc system is involved in long-term colonization of the respiratory tract epithelium and downregulation of the immune response (98, 122, 144).

Finally, the nonpathogenic Myxococcus xanthus also has genes that encode a T3S system that would be a member of this subgroup (101).

The Ysc family of T3S systems is thus encountered in α-proteobacteria (Bordetella spp.), γ-proteobacteria (Yersinia spp., P. aeruginosa, Aeromonas spp., and Vibrio spp.) and δ-proteobacteria (D. vulgaris). At least in α-proteobacteria, these T3S systems confer resistance to the innate immune response through resistance to phagocytosis and triggering of apoptosis in macrophages. Consequently, this cascade of actions leads to an extracellular localization of the pathogen. The Bsc T3S system also allows B. bronchiseptica to avoid phagocytosis and to reduce the proinflammatory response (144). Although the role of the three other T3S systems of this family is not described yet, we would suggest that the presence of a T3S system of the Ysc family is indicative of an extracellular pathogen.

The Inv-Mxi-Spa family

This family is named after the Inv-Spa T3S system of Salmonella enterica and the Inv-Mxi T3S system of Shigella spp. (FIGURE 2B). Both represent paradigms of T3S systems triggering bacterial uptake by nonphagocytic cells (24, 115). In Shigella spp., the T3S system is dedicated to the entry of the bacterium in epithelial cells and the defense against macrophages. (For a review, see Ref. 115). The Inv-Spa T3S system, encoded by Salmonella pathogenicity island 1 (SPI-1) of S. enterica, is associated with cell invasion as well. New members of this family include the Ysa T3S from Y. enterocolitica biogroup 1B and the T3S systems of Yersinia ruckeri, Sodalis glossinidius (Inv-Spa), Escherichia. coli (Eiv-Epa), Burkholderia spp. (Inv-Spa), and Chromobacterium violaceum (Inv-Spa).

S. glossinidius is an intracellular endosymbiont of the tsetse fly. Its genome encodes two different T3S systems, both of the Inv-Mxi-Spa family. The first one, encoded by the Sodalis symbiosis region 1 (SSR-1) is dedicated to the invasion of eukaryotic cells (32, 34). The second one, encoded on the SSR-2 locus, is required for the intracellular survival of S. glossinidius (32). The latter seems to have undergone modifications due to the obligate intracellular life of S. glossinidius (32): it has no functional needle and no translocator genes (32). The SSR-2 T3S system could be the first known T3S system to secrete proteins into the environment that surrounds the bacterium and not across membranes.(32). Similarly to S. glossinidius, the primary endosymbiont of Sitophilus zeamais (SZPE) also has a T3S system of the Inv-Mxi-Spa family for cell invasion (33).

A similar T3S system involved in cell invasion is found in Burkholderia spp. (Table 1). B. pseudomallei causes melioidosis in humans. Like Shigella flexneri, B. pseudomallei invade nonphagocytic cells, escape from the endocytic vacuole into the cytosol where they multiply, and spread from cell to cell by inducing actin protrusions (115, 127). The T3S system is active during these intracellular stages because T3S-defective mutants are unable to escape from endocytic vacuoles and cause the actin rearrangements (127). B. pseudomallei also survives and multiplies within phagocytes (127). Burkholderia mallei, an obligate equine pathogen causing glanders disease and occasionally a human pathogen (133), and Burkholderia thailandensis (4) also have this T3S system.

These four T3S systems in the Inv-Mxi-Spa family thus trigger phagocytosis of nonphagocytic cells by triggering actin polymerization. The genes responsible for this phenotype (sicAsipBCDA in Salmonella SPI-1) form the so-called “entry locus” (115). When a complete entry locus and an Inv-Mxi-Spa locus appears in a genome, it is likely that the bacterium has an invasive phenotype. This might be the case for C. violaceum, a versatile soil- and water-borne bacterium found in a variety of tropical and subtropical ecosystems. Although it is generally considered to be nonpathogenic, it is responsible for sporadic cases of lethal human septicemia, mainly in young children and in immunodeficient individuals (42).

The Ysa system of Y. enterocolitica (48, 62) is present only in the high-virulence strains of Y. enterocolitica, the so called American strains (biotype 1B), but until now, no role could be attributed to this T3S system (49, 62, 136). A slightly different T3S system is found in Y. ruckeri, the agent of the red mouth disease in trout (58), but little is known about its implication in pathogenesis.

Finally, a large panel of E. coli strains encode a system called Eiv-Epa (99), but it seems that this system is functional only in a few strains (64, 88, 110). The locus encoding this system (ETT2) is highly variable from strain to strain and has suffered mutational attrition in most cases (101, 110). In the few strains where the ETT2 locus seems to encode a functional T3S, the entry locus (eip locus) is also intact (110), suggesting a role in invasion, but it is unclear of what cell or organism (110).

The Ssa-Esc family

Whereas the SPI-1-encoded T3S is involved in the early stages of S. enterica infection, the SPI-2-encoded T3S, called Ssa, allows S. enterica to survive in macrophages by preventing endocytic trafficking and phagosome maturation (for a review, see Ref. 137). Edwardsiella tarda, a fish pathogen that also causes gastro- and extraintestinal infections in humans, and C. violaceum have a similar T3S (8, 12a, 109).

Surprisingly, the chromosome of Y. pestis, Y. pseudotuberculosis, and Y. enterocolitica from serotype 3 also encode a T3S of this family (55, 103, 105). It is not yet clear whether it plays a role in pathogenesis. At least in Y. pestis, it might be nonfunctional, since some open reading frames are disrupted by frame shifts (101, 103) and no translocator genes are found. This T3S thus offers another example of the massive attrition that occurred in the genome of Y. pestis (103).

The archetypal Esc T3S system of the enteropathogenic E. coli and enterohemorrhagic E. coli strains allows both to adhere to enterocytes and cause the effacement of the microvilli located at the contact point with the bacterium. The Esc T3S has also been shown to cause invasion of the nonphagocytic cells and to allow survival within the host (19, 57, 77, 89, 134). In the Esc injectisome, the needle is prolonged by a long, flexible filament (35, 82).

The Hrc-Hrp 1 and Hrc-Hrp 2 families

All of the T3S systems from plant pathogens can be grouped into two families on the basis of the genetic loci organization and regulatory systems (1, 2). This subdivision also appears in the phylogenetic trees (48, 56, 96). In both families, the injectisome has a long, flexible pilus instead of a needle. This adaptation results from the need to traverse the thick plant cell wall. Although the injectisome is well conserved in many plant pathogens, the set of effectors is exceptionally variable. This diversity is reflected in the numerous pathovars within one species, each one specialized in the infection of one plant species (111). In host plants the bacteria multiply in the intercellular spaces, whereas in nonhost plants the bacterial colonization of the tissue is rapidly stopped by the induction of the hypersensitive response, which consists of a programmed cell death of the cells that are at the site of infection and a callose deposition in the cell wall of the cells around the site of infection. The effectors are the elements that are recognized by the nonhost plant cells, which means that, in this case, they betray the bacterium (see Refs. 1, 26, and 27 and see Ref. 2 for a review). This recognition game led to a huge diversity of effectors that somehow participates in the definition of pathovars (54, 111).

Surprisingly, V. parahaemolyticus, which is a marine bacterium and an agent of human gastroenteritis, possesses a T3S system that belongs to the Hrc-Hrp 1 family, in addition to its Ysc-like T3S. What is intriguing is that this plant-pathogen-like T3S is exclusively present in strains isolated from patients (87).

A range of Burkholderia spp. bacteria possess the Inv T3S that is typical of animal pathogens and the Hrc T3S that is typical of plant pathogens (Table 1 and FIGURE 1B). This duality is not too surprising for B. pseudomallei, which is a highly versatile soil saprophyte (41) responsible for melioidosis in humans. B. pseudomallei even possesses two different Hrc-Hrp 2 T3S systems, suggesting a large diversity of cell-cell interactions (107). This duality is more surprising for a bacterium like B. mallei, which is considered to be an obligate animal pathogen restricted to equine species.

The Rhizobiales family

T3S loci were identified in a number of Rhizobium species, in Bradyrhizobium japonicum, and in Mesorhizobium loti. Rhizobiales establish symbiotic relations that result in the formation of nodules on the roots of leguminous plants. Inside these nodules, the bacteria change their metabolism and start to reduce nitrogen into a form that is absorbable for the plant while the plant supplies the bacteria with carbohydrates. This symbiotic relationship requires an intensive communication between the plant and the bacterium. Depending on the species, the T3S can play a positive role and favor the formation of the nitrogen-fixing nodule or can favor a defensive reaction from the plant and prevent the bacteria from colonizing the roots (90). The tree shown in FIGURE 2B misleadingly suggests that T3S systems from Rhizobiales could be considered to be part of the Hrc-Hrp 2 family, but the bootstrap value of the upper node is low and trees made with other T3S components would place this clade in relation to other families.

The Chlamydiales family

Chlamydiales are the only nonproteobacteria to harbor a T3S. They represent a coherent group that diverged from the other Gram-negative bacteria ~2 billion years ago (71). Chlamydiae are strictly intracellular pathogens that grow inside a transformed vacuole called inclusion. The bacterium enters the target cell in a metabolically inactive form called the developmental body and then differentiates into a metabolically active form that becomes strictly dependent on the cell to continue its life cycle. The Chlamydiales group is extremely interesting in terms of host-pathogen interactions but unfortunately is not yet amenable to genetic studies. T3S in Chlamydia was discovered by genome sequencing in 1998, but it was only in 2003 that this system was shown to be functional in vivo and involved in the maintenance of a permissive state of the eukaryotic host (46). The more genomes of Chlamydiae are sequenced, the more it appears that T3S is common in this order (101), including in the environmental species that infect amoeba (71), which suggests that Chlamydiae might posses the most ancient NF-T3S system. This view, which is in agreement with the phylogenetic trees (48, 56, 71, 80, 96), is also reinforced by the fact that in Chlamydiae, the genes encoding the T3S component are spread on different loci in the genome and not localized within a single large locus as often seen in proteobacteria (80, 125).

Concluding Remarks

Four remarks come to mind when looking at the new data generated by entire genome-sequencing programs. First, NF-T3S is far more widespread than was anticipated when the concept was developed from a few archetypes. It is found not only in animal and plant pathogens but also in insect and amoeba pathogens, and it is essential to some symbionts. Second, there is nothing in common between the evolutionary tree of NF-T3S systems and the evolutionary tree of bacteria based on 16S RNA sequences. This implies that T3S systems did not evolve with their present host bacteria but spread between bacteria by horizontal gene transfer. Two families, the Chlamydiales and the Rhizobiales, form a homogenous clade in the phylogenetic trees (FIGURE 2), showing that their T3S systems have not recently spread to bacteria from other phylogenic groups. The fact that both groups live in a very peculiar niche may account for this lack of exchanges with other bacteria.

Third, NF-T3S systems occur in species that interact with eukaryotes and not in closely related nonpathogenic, nonsymbiotic species. This fits perfectly well with the idea that they spread by horizontal transfer, conferring “by chance” new properties to a symbiotic bacterium or to some bacteria that are accidentally present in a potential host. In addition, although many pathogenic Gram-negative bacteria have no NF-T3S system, some have more than one and generally from different families. This suggests that bacteria that already have one system may be more prone to gain a selective advantage when they possess a second one. This could result from the fact that there is some promiscuity within the effectors. Thus a bacterium that already has effectors for one system may be more capable of benefiting from the acquisition of a new system. Interestingly, there are no examples of bacteria having both a T3S and a classical type IV secretion system (T4S). T4S are structurally and functionally different from T3S; they evolved from bacterial conjugation systems, but they basically fulfill the same function as T3S systems (for a review, see Ref. 18).

Finally, apart from the few archetypes, we have almost no idea of what these systems do. Genomes bring more questions than answers. The clustering of T3S systems in distinct families might reflect the variety of host-bacterium relationships in which T3S systems are involved, and the archetypal systems of the family might help us to understand the role of newly discovered systems. This is a very simple idea that is unlikely to apply in every case, but it might suggest new avenues for research


We thank Jaime Mota and Hwain Shin for critical reading of this manuscript and helpful suggestions and thank Laure Journet for the gift of the electron micrograph from FIGURE 1.

Our work on T3S is financed by the Swiss National Science Foundation (grant 32-65393.01). P. Troisfontaines was funded by the Belgian Fonds pour la Formation à la Recherche dans l’Industrie et dans l’Agriculture.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 12a.
  14. 13.
  15. 14.
  16. 15.
  17. 16.
  18. 17.
  19. 18.
  20. 19.
  21. 20.
  22. 21.
  23. 22.
  24. 23.
  25. 24.
  26. 25.
  27. 26.
  28. 27.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
  57. 57.
  58. 58.
  59. 59.
  60. 60.
  61. 61.
  62. 62.
  63. 63.
  64. 64.
  65. 65.
  66. 66.
  67. 67.
  68. 68.
  69. 69.
  70. 70.
  71. 71.
  72. 72.
  73. 73.
  74. 74.
  75. 75.
  76. 76.
  77. 77.
  78. 78.
  79. 79.
  80. 80.
  81. 81.
  82. 82.
  83. 83.
  84. 84.
  85. 85.
  86. 86.
  87. 87.
  88. 88.
  89. 89.
  90. 90.
  91. 91.
  92. 92.
  93. 93.
  94. 94.
  95. 95.
  96. 96.
  97. 97.
  98. 98.
  99. 99.
  100. 100.
  101. 101.
  102. 102.
  103. 103.
  104. 104.
  105. 105.
  106. 106.
  107. 107.
  108. 108.
  109. 109.
  110. 110.
  111. 111.
  112. 112.
  113. 113.
  114. 114.
  115. 115.
  116. 116.
  117. 117.
  118. 118.
  119. 119.
  120. 120.
  121. 121.
  122. 122.
  123. 123.
  124. 124.
  125. 125.
  126. 126.
  127. 127.
  128. 128.
  129. 129.
  130. 130.
  131. 131.
  132. 132.
  133. 133.
  134. 134.
  135. 135.
  136. 136.
  137. 137.
  138. 138.
  139. 139.
  140. 140.
  141. 141.
  142. 142.
  143. 143.
  144. 144.
  145. 145.
  146. 146.
  147. 147.
View Abstract