Metabolic effectors secreted by bacterial pathogens: essential facilitators of plastid endosymbiosis?

Abstract

Under the endosymbiont hypothesis, over a billion years ago a heterotrophic eukaryote entered into a symbiotic relationship with a cyanobacterium (the cyanobiont). This partnership culminated in the plastid that has spread to forms as diverse as plants and diatoms. However, why primary plastid acquisition has not been repeated multiple times remains unclear. Here, we report a possible answer to this question by showing that primary plastid endosymbiosis was likely to have been primed by the secretion in the host cytosol of effector proteins from intracellular Chlamydiales pathogens. We provide evidence suggesting that the cyanobiont might have rescued its afflicted host by feeding photosynthetic carbon into a chlamydia-controlled assimilation pathway.

Figure 1.

Chlamydial Genes in Archaeplastida. (A) Distribution of BLASTx hits to proteins encoded on chlamydial genomes. The vast majority of proteins are, as expected, of bacterial affiliation, whereas the second largest class of hits is to Archaeplastida. (B) The sister group relationships of Chlamydiales-derived proteins in Archaeplastida at bootstrap cutoff ≥90%, ≥75%, and when these numbers are added, including the four genes shared exclusively by Chlamydiales and Archaeplastida (see Table 1)

Figure 2.

Phylogenetic Analysis of ADP-Glc–Specific Starch Synthases of Bacteria and Archaeplastida. Unrooted GlgA (glycogen starch synthase) tree using the best-fit protein model (LG + G) and 1000 RAxML bootstrap replicates (alignments are given in Supplemental Data Sets 2 to 4 online). To identify the source of the ADPGlc-using glucan synthase involved in endosymbiosis, the phylogeny of archaeplastidal starch synthases compared with bacterial glycogen synthases was determined using maximum likelihood analysis with 1000 replica bootstrap analysis. Extant archaeplastidal candidates whose ancestors may have triggered the establishment of the symbiotic link are those enzymes that can be linked phylogenetically to ADP-Glc–requiring enzymes conserved in green algae and land plants (in green), which group into three classes known as granule-bound starch synthase (GBSS in the bottom part of figure), the soluble starch synthase (SS) SSI/SSII class (in the bottom part of the figure), and the SSIII/SSIV class (in the top and middle parts of the figure; reviewed in Ball and Morell (2003). The high bootstrap nodes unifying the GBSS-SSI-SSII and the SSIII-IV clades are highlighted in bold. Rhodophyceae starch synthase progenitors are not present in this tree because those organisms have lost the ability to polymerize starch from ADP-Glc and have retained only the very different host-derived soluble glycogen (starch) synthase that uses UDP-Glc (reviewed in Ball et al., 2011). However, GBSSI was transmitted to red algae (in red) and Glaucophyta (in dark blue) where it polymerizes amylose from UDP-Glc in the cytosol. These proteins still display the ability to use ADP-Glc. The source of the GBSSI, SSI, and SSII clade is proposed to be cyanobacterial and is displayed by an arrow (for reasons sustaining this source, see text and Deschamps et al., 2008a; Ball et al., 2011). As to the SSIII/SSIV clade, SSIII/SSIV was previously reported to result from an HGT from environmental chlamydia-like amoeba parasites P. amoebophila and P. acanthamoeba (Moustafa et al., 2008). However, the phylogeny displayed here suggests the presence of several LGT events not previously evidenced. Because of the importance of the issue, we devoted the Supplemental Methods 1 online to the detailed analysis of these results.

Figure 3.

Chlamydial Enzymes Encode Functional TTS Signals. The N-terminal 30 codons (see Supplemental Table 4 online) of the indicated genes were cloned upstream of cya and expressed in the ipaB and mxiD S. flexneri strains. The ipaB strain constitutively secretes TTS substrates (labeled TTS+), whereas the mxiD strain is defective for TTS (labeled TTS−) (Subtil et al., 2001). Exponential cultures expressing the fusion protein were fractionated into supernatants (S) and pellets (P). Samples were resolved by SDS-PAGE gels, transferred to a polyvinylidene difluoride (PVDF) membrane, and probed with anti-Cya to detect the chimera. Probing the membrane with anti-IpaD showed that TTS in the ipaB background was not impaired by transformation of the various constructs. Antibodies against the cAMP receptor protein (CRP) were used to control for bacterial lysis during fractionation. A GlgA construct in which the first 10 amino acids were deleted (Δ10GlgA/Cya) was included as a negative control. When expressed in the ipaB strain, the Δ10GlgA/Cya chimera was not secreted in the supernatant, in contrast with the GlgA/Cya chimera. In the mxiB (TTS defective) background, none of the chimeric proteins were recovered in the culture supernatant, demonstrating that their secretion in the ipaB background occurred by a TTS mechanism.

Figure 4.

Glycogen Metabolism Putative Effectors in Extant and Ancient Chlamydiales. The infection cycle of a typical eukaryote by a chlamydia-like organism is shown. Chlamydiales are internalized through phagocytosis. The pathogen modifies the surrounding vesicle into an inclusion membrane (shown in blue). These bacteria undergo two successive developmental forms: the infectious elementary body (EB) and the replicative reticulate body (RB). At the end of the cycle, reticulate bodies differentiate back to elementary bodies that are eventually released from the cell. Both elementary bodies and reticulate bodies possess a type III secretion apparatus that spans the two bacterial membranes and the inclusion membrane, allowing for the secretion of bacterial proteins directly into the host cytosol. In the enlargement, the inclusion membrane is depicted in blue, type III secretion apparatuses in pink, the eukaryote cytosol in yellow, and the inside of the chlamydial cell in white. The Glc residues of the host glycogen used as primers for synthesis are symbolized by black dots. Red dots represent accessible Glc, whereas the Glc-1-P generated by glycogen phosphorylase is represented by red dots with an attached “P.” The sequential nature of the biochemical reactions is depicted as numbers 1 to 5. The enlargement displays the particular type of putative effector combination present in extant P. acanthamoeba and possibly in the ancient parasite that would have predisposed a host cell to endosymbiosis. Briefly, carbon is diverted to the glycogen pools through the chlamydial ADP-Glc pyrophosphorylase putative effector (GlgC) making use of the high host cytosolic ATP and Glc-1-P pools at the beginning of the infection cycle. Polymerization into glycogen from ADP-Glc occurs through the chlamydial glycogen synthase (GlgA). The host glycogen synthase only uses UDP-Glc and is not represented. Branching could occur through the action of either the host or parasite branching enzyme, although we did not find clear evidence for a type III secretion signal in the two parasite branching enzymes tested (GlgB), indicating that branching might solely involve host enzymes. When the orthophosphate concentration rises and the cytosolic ATP and Glc-1-P decreases as a consequence of the infection, parasite phosphorylase (GlgP) will recess the outer chains of glycogen, terminating four residues away from each branch. This will allow the action of GlgX, whose possible substrates are restricted to such chains. The maltotetraose generated by GlgX is normally not metabolized by eukaryotes in the cytosol and could therefore be a substrate for import and catabolism within the parasites. The transporter for the import of MOS in the parasite is presently unknown. We omitted MalQ from the drawing since we do not know if the chlamydial version of this enzyme would behave more like a maltase or disproportionate type of α-1,4 glucanotransferase.

Figure 5.

Ménage à Trois. The interdependent symbiosis between the cyanobiont, its host, and a chlamydial parasite is shown, displaying an effector combination akin to that hypothesized in Figure 4. Enzymes are colored with respect to their phylogenetic origin: in orange for host enzymes of eukaryotic glycogen metabolism, in pink for chlamydial effectors, and in blue for cyanobacterial enzymes. Only those enzymes of chlamydial origin that are required to establish the symbiotic flux and feed carbon into the parasite or the host are displayed. The other enzymes depicted in Figure 4 do not interfere with this biochemical flux. The GlgX chlamydial debranching enzyme is represented here by the name “iso” to denote its present function in Archaeplastida (isoamylase). The cyanobiont in blue exports the bacterial specific metabolite ADP-Glc by recruiting a family III nucleotide sugar translocator (colored in orange) from the host endomembrane system (colored in yellow) as recently proposed (Weber et al., 2006; Deschamps et al., 2008a; Colleoni et al., 2010). A possible primitive TOC (translocon of the outer chloroplast membrane) targeting system is drawn in gray. The ADP-Glc would be funneled to glycogen by the chlamydial effector glycogen synthase, whereas the host would still be able to polymerize Glc into glycogen from UDP-Glc. The dramatic increase of the cytosolic glycogen pools would have benefitted the host through the increased production of Glc and Glc-1-P that can be further metabolized. However, it would have equally benefitted the parasite by increasing the supply of MOS that may only be metabolized by the latter. BE, branching enzyme; iDBE, indirect debranching enzyme; ISO, glgX type of direct debranching enzyme; PHO, phosphorylase; SS-ADP, ADP-specific glycogen synthase; SS-UDP, UDP-specific glycogen synthase.

Figure 6.

Phylogenetic Analysis of Bacterial and Archaeplastidal Glycogen Debranching Enzymes. Starch debranching enzymes of Chloroplastida are known to play an important role both in polysaccharide synthesis and starch degradation. The maximum likelihood tree of these enzymes shows the three types of Chloroplastida (land plants and green algae) subunits (filled green triangles). Isoamylase1 (Isa1) is responsible for trimming misplaced preamylopectin branches that if unprocessed prevent polysaccharide crystallization, leading to glycogen formation, rather than starch. Isa2 is a noncatalytic subunit assembled in heteromultimers together with Isa1. Isa3 is involved in the breakdown of branches during starch degradation. The homologous enzymes of red algae (red filled triangle) and Glaucophyta (blue branches) are also shown in this tree. Analogous specialized roles in starch synthesis and degradation remain to be determined in these organisms. The novel glaucophyte sequences were isolated using RT-PCR. Alignments are shown in Supplemental Data Sets 3 to 5 online, and the full tree is displayed in Supplemental Figure 2 online.