Genome analysis of three Pneumocystis species reveals adaptation mechanisms to life exclusively in mammalian hosts

Abstract

Pneumocystis jirovecii is a major cause of life-threatening pneumonia in immunosuppressed patients including transplant recipients and those with HIV/AIDS, yet surprisingly little is known about the biology of this fungal pathogen. Here we report near complete genome assemblies for three Pneumocystis species that infect humans, rats and mice. Pneumocystis genomes are highly compact relative to other fungi, with substantial reductions of ribosomal RNA genes, transporters, transcription factors and many metabolic pathways, but contain expansions of surface proteins, especially a unique and complex surface glycoprotein superfamily, as well as proteases and RNA processing proteins. Unexpectedly, the key fungal cell wall components chitin and outer chain N-mannans are absent, based on genome content and experimental validation. Our findings suggest that Pneumocystis has developed unique mechanisms of adaptation to life exclusively in mammalian hosts, including dependence on the lungs for gas and nutrients and highly efficient strategies to escape both host innate and acquired immune defenses.

Figure 1. Conservation of Pneumocystis genome structure.

(a) Conserved synteny among three Pneumocystis genomes. Shared syntenic regions are depicted with grey boxes. Scaffold numbers are listed on the x-axis, and red dots indicate the location of msg genes. (b) Genome-wide SNP frequency between the P. jirovecii isolates from the United States (RU7) and Switzerland (SE8) with the same scaffold order as in top panel. The region beyond 8 Mb with a high number of SNPs is composed primarily of small scaffolds containing msg genes not assembled into the 20 large scaffolds. Using our genome assembly (RU7) as a reference, we identified a total of 24,902 SNPs, or 1 every 337 bases, between these 2 isolates, which are over-represented in subtelomeric regions where msg genes are found.

Figure 2. Protein domains depleted or enriched in Pneumocystis.

Significantly enriched (top panel) or depleted (lower panel) Pfam domains (Fisher's exact test, q value<0.05) are included in the heat map if the domains appear at least twice in the following comparisons: Pneumocystis versus Schizosaccharomyces, Pneumocystis versus Schizosaccharomyces and T. deformans, Pneumocystis versus S. cerevisiae and C. albicans, Pneumocystis versus E. cuniculi and E. intestinalis, Pneumocystis versus all others shown. Broader functional categories of proteins are indicated on the left, while specific Pfam domains are listed on the right. The number of proteins containing each domain is indicated within each box for each species. The heat map is colour coded based on a Z score, as indicated by the key at the bottom right. Fungal species are ordered based on their phylogenetic relationship as indicated at the bottom.

Figure 3. The Msg superfamily in three Pneumocystis species.

(a) Phylogeny of 384 Msg proteins identified in P. murina (blue squares), P. carinii (pink circles) and P. jirovecii (green diamonds). They are classified into five families of Msg-A, -B, -C, -D and -E, as indicated by the vertical bars on the right side. The Msg-A family is further classified into three subfamilies of Msg-A1 (classical Msg genes), Msg-A2 (Msr genes) and Msg-A3 (other Msg-associated genes). (b) Schematic representations of conserved domains in five Msg families. (c) Sequence logos showing the frequency of amino acid composition in Msg domains. Previously identified Pfam MSG and Pfam Msg2_C domains are included for comparison. Additional information on the Msg domain analysis is provided in Supplementary Note 4.

Figure 4. Reduction of carbohydrate and lipid metabolism in Pneumocystis.

A condensed version of pathways highlights retained (green arrows) and lost (grey arrows) pathways. Enzymes and membrane transporters absent in all three Pneumocystis species are highlighted in red font; those retained in all three species are highlighted in blue. Yellow and grey boxes indicate metabolites present and absent, respectively. Some metabolites are included more than once as they interact with multiple pathways, in which case the yellow or grey colouring refers to their role in different pathways. Enzyme Sur2 (in pink) is present in only P. murina but not P. carinii or P. jirovecii. Enzyme Mct1 (in pink) is present in both P. murina and P. carinii but not P. jirovecii. The boxed question mark (‘?') leading to inositol indicates a hypothetical enzyme. The names of enzymes, transporters and metabolites follow the standard abbreviated names for S. cerevisiae. The enzyme and transporter names containing two or more digits represent duplicated enzymes and transporters.

Figure 5. Summary of mechanisms of adaptation to host lungs by Pneumocystis.

Different mechanisms are highlighted by different colours. Potential mechanisms of uptake of nutrients (which cannot be synthesized de novo) include the use of plasma membrane-localized transporters (indicated by T), conversion of other metabolites scavenged from hosts (indicated by C), endocytosis (indicated by E) and unknown (indicated by U). ^§Potential uptake of haem or haemoglobin from lungs by endocytosis mediated by CFEM domain-containing proteins. *Cholesterol biosynthesis pathway is retained in P. jirovecii but lost in P. murina and P. carinii. ^¶β-glucan is present in cysts and absent in trophic forms. β-glucan as well as chitin and mannan in other fungal pathogens are known pathogen-associated molecular patterns (PAMP) involved in host immune recognition; none of these components is detected in Pneumocystis organisms except for the presence of β-glucan in the cyst form (Figs 6 and 7; Supplementary Fig. 13). Pneumocystis is the only fungus identified to date that cannot synthesize chitin.

Figure 6. Analysis of chitin in Pneumocystis and related fungi.

(a) Enzymes and accessory proteins involved in chitin metabolism in fungi. Pneumocystis genomes do not encode any chitin synthase or chitinase, which are present in other fungi, but retain genes encoding four accessory proteins (Supplementary Data 17). (b) Gas chromatograms of partially methylated alditol acetates of P. carinii and S. cerevisiae (control) cell walls. Terminal and 4-linked N-acetylglucosamine signals (T-GlcNAc and 4-GlcNAc in red font) were detected in S. cerevisiae but not in P. carinii. Glucose (Glc) and mannose (Man) signals were detected in both species. (c–e) Detection of chitin with recombinant chitin-binding domain (Alexafluor 488) using P. murina-infected lung tissue (c) and C. albicans-infected kidneys (e) as a positive control. Pneumocystis organisms are demonstrated in d by dual staining with anti-Msg (red), which labels both trophic forms and cysts, and a dectin-Fc construct (green), which labels β-1,3-glucan in cysts. Chitin staining is absent in P. murina but readily detected in C. albicans, while β-1,3-glucan is easily seen in P. murina. Original magnification, × 400; scale bar, 10 μm.

Figure 7. Lack of hyper-mannose (mannan) glycosylation in Pneumocystis.

(a) Diagram of N-linked mannan structure in C. albicans, based on ref. . (b) Diagram of N-linked glycans in Pneumocystis, which lack the α-1,6-linked mannose backbone as well as α-1, 2- and α-1,3- linked mannose outer chains seen in C. albicans (square brackets). (c,d) Representative results of tandem mass spectrometry (MS/MS)-higher energy collisional dissociation (HCD) and electron transfer dissociation (ETD) analysis of an N-linked glycopeptide carrying Hexose5 HexNAc2 (M5N2) from one Msg isoform (T552_03736) in P. carinii. (c) MS/MS-HCD spectrum of glycopeptides showing the detection of glycan oxonium ions in the low mass region at m/z 163.0603, 204.0868, 366.1398 and 528.1929 (indicated in red font). A series of fragment ions dues to neutral loss of the glycan moiety were observed as the main fragment ions in the HCD spectrum. Trace amounts of y-type and b-type peptide fragment ions were detected, confirming the sequence of the peptide backbone. (d) MS/MS-ETD spectra of peptide fragment ions with minimal neutral loss of glycan moiety. All expected peptide c-type and z-type fragment ions were detected except c8 fragment ion, confirming the peptide sequence with high confidence, as well as the site and mass of the glycosylation modification.