P. massiliensis was isolated from a cooling tower in the Vaucluse region of France, as previously described (La Scola et al., 2000; Pagnier et al., 2015), using Vermamoeba vermiformis as a cell support for amoeba co-culture. For production, after being rinsed with PAS (Page's amoeba saline) and suspended in a starvation medium (Bou Khalil et al., 2016) with a final concentration of 10⁶ amoebas/ml, V. vermiformis was infected with a P. massiliensis suspension at a multiplicity of infection (MOI) of 10, in two 75-cm² culture flasks at 30°C. Washing with PAS buffer was performed after 1 h (H0) of incubation to eliminate non-internalized bacteria. Next, 10 ml of infected cultures was incubated at 30°C in new culture flasks alongside a negative control consisting of a flask containing only amoeba for further analysis.

Of these 10 ml, 500 μl of culture suspension was used for the preparation of five slides through cyto-centrifugation in order to perform Gram staining, Gimenez staining and DAPI nucleic acid labeling (Molecular probes, Life Technologies USA) at the following stages respectively: H0, H2, H4, H6, H8, H12, H18, H24, H30, H36, H42, and H48. In addition, 500 μl of the prepared culture suspension was used for DNA extraction and molecular analysis. In short, bacterial growth and count were calculated using real time PCR assays, in order to assess the correlation between bacterial concentration and cycle threshold (Ct.), 200 μl of each co-culture at every infection time point of the cycle (H0–H48) were used for DNA extraction by EZ1 DNA Tissue Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions on a CFX96TM thermocycler (BioRad Laboratories Inc.). The real-time PCR primers used were the following: Forward: 5′GCTCCGATTCAGCGAATACT 3′, Reverse: 5′GTCTGCTCTTCCATTCCCATAA3′, and Probe: 5′CTGCTAAGCATGTTGCTAAGCTTGGC 3′. Amoebal quantification was performed on counting slides (kovaslides, HYCOR biomedical Inc., 90 California, USA).

The remaining 9 ml was used for transmission electron microscopy after pellet fixation. For transmission electron microscopy processing, all specimen fixation, embedding, cutting, observation, and analysis procedures were performed as previously described by Bou Khalil et al. (2016). For the host range, the supernatant of a 1-week-old culture flask of V. vermifromis infected with Chlamydiae was collected and filtered through a 5 μm pore-size filter and washed three times with modified PAS. The later was used for inoculation onto A. castellanii (strain Neff 30010) and A. polyphaga (strain Linc-AP1) at an MOI of 10 in a 24-well microplate seeded with 10⁶ Amoebae and containing 1 ml of PAS. The microplates were then incubated at 32°C (A. castellanii) for 5 days following centrifugation at 1,500 × g for 30 min. Amoebas were observed daily for lysis and 500 μl from the co-cultures was taken daily for DNA extraction and assessment of the number of chlamydial DNA copies using real-time quantitative PCR.

Quantitative replication values were statistically compared using the Mann whitney test and a p-value < 0.05 was considered to reflect a significant difference.

The genomic DNA of P. massiliensis was sequenced using a MiSeq Technology sequencer (Illumina, Inc., San Diego, CA). Four Illumina short-insert paired-end libraries were performed with an average insert size of 251 bps along with an Illumina long-insert paired-end library with an insert size of 520. Libraries and sequencing were performed as previously described (Lo et al., 2015). Reads were then assembled using the de novo method and the SPAdes-3.0.0 software (Bankevich et al., 2012). The non-coding genes and miscellaneous features were generated using RNAmmer (Lagesen et al., 2006), ARAGORN (Laslett and Canback, 2003), Rfam (Griffiths-Jones et al., 2003), and Infernal (Nawrocki et al., 2009). Coding DNA Sequences (CDSs) were predicted using Prodigal (Hyatt et al., 2010) and functional annotation was achieved using BLAST+ (Camacho et al., 2008) and HMMER3 (Eddy, 2011) against the UniProtKB database. Data for P. massiliensis were submitted to the EMBL database and were assigned Bio-projects number PRJEB6590; the accession numbers for the genome in the EMBL database are CCJF01000001-CCJF01000005.

Phylogenetic analyses were performed for the Chlamydiae genes and the corresponding gene sequences available on the NCBI database. Multiple sequence alignments were performed using MUSCLE (Edgar, 2003) and curated using Gblocks (Talavera and Castresana, 2007). Phylogenetic trees were generated using the PhyML Maximum Likelihood algorithm and visualized using MEGA v5 (Tamura et al., 2011).

The bacterium is strictly intracellular and unable to grow outside amoeba. To date, all attempts to grow this bacterium on agar media and nutritive broths (PYG, Trypticase soy broth) under several axenic conditions (aerobic, anaerobic, and microaerophilic atmospheres) have failed. For the replicative cycle, bacterial embedded bodies (EBs) are phagocytized and internalized in a vacuole inside the amoebal cytoplasm and can be seen near the nucleus (Figure 1A). At H4, H6, H8 post-infection, bacteria begin to lose their electron-density, progressively increase in size, and form reticulate bodies (RBs) (Figures 1B,C). Between H8 and H12, bacterial bodies continue to grow and form large accumulations of material, and binary division of the bacteria can be observed (Figures 1C,D). Between H24 and H36, the division continues to increase and the bacteria occupy the entire amoeba cytoplasm (Figures 1E,F). After a period of growth and division of a typical asynchronized cycle of Chlamydiae, the RBs reorganize between, approximately, H36 and H42 p.i., condensing to form infectious and pre-mature EBs (Figure 1E) that mature into highly condensed ones, leading to cell burst (Figure 1F). The P. massiliensis grew in V. vermiformis, A. polyphaga and A. castellanii, and a cytopathic effect was seen 48 h after infection. Complete lysis was observed within 96 h. In contrast, the non-infected cultures of each host cell, used as negative controls, showed no detectable loss of amoebae over the same experimental timeframe.

Moreover, the culture and relative quantification by real-time PCR showed an increase in bacterial multiplication in all three amoebae strains at H24, but particularly at H72 and H96, with an approximate three-log increase in the number of bacteria per milliliter in 4 days (Figure 2). No significant statistical difference was found in the quantitative replication values between the three tested hosts, where the realized Mann whitney test showed a p-value > 0.05. When comparing the replicative cycle of P. massiliensis in A. castellanii and in V. vermiformis at the same time points of infection, no inclusion vacuoles could be detected in the case of V. vermiformis and bacteria were directly identified in the cytoplasm (Figure 3A). However, we clearly observed the presence of inclusion vacuoles filled with P. massiliensis in the case of A. castellanii, where bacteria appear to grow in vacuoles in the cytoplasm, forming morula-like shapes (Figure 3B).

The P. massiliensis genome consists of a chromosome sequence assembled into four contigs (Figure 4). This sequence was estimated to have 2,864,073 pbs and a GC content of 42.8%. In total, 2,389 Protein Coding Sequences (CDS) were identified as well as six sets of rRNA (two genes are 5S rRNA, two genes are 16S rRNA, two genes are 23S rRNA) and 39 tRNA genes. A total of 1,204 genes (50.39%) were attributed a putative function (by cogs or by NR blast). In addition, 589 genes were identified as ORFans (24.65%), and the remaining genes were annotated as hypothetical proteins (21.05%). The fifth contig is 92,055 bps with a GC content of 40.87%, represented a circular plasmid. The average read depth of this plasmid (294x) compared to the chromosome (153x) suggests that the plasmids are present in double copy per cell. After annotation, hypothetical functions can be assigned to more than 80% of the plasmid genes.

The phylogenetic tree of the 16S rRNA, built with a representative set of chlamydia sequences available in the NR database, clusters P. massiliensis within the Protochlamydia genus (Figure S1). The bootstrap relating P. massiliensis to the other members of the Protochlamydia genus was 100%. The chlamydial taxonomy based on the 16S rRNA sequence similarity and phenotypic characteristics was used for chlamydial description. However, the Chlamydiales order exhibits few phenotypic differences. Moreover, Tamisier et al. showed that many of the current bacterial species with validly published names do not respect the 95 and 98.7% thresholds usually considered as delineating genus and species respectively (Rossi-Tamisier et al., 2015). Pillonel et al. recommended using the 16S and 23S genes combined with 9 protein sequences to precisely classify newly discovered isolates at the family, genus, and species levels (Pillonel et al., 2015). We used Pillonel's classification method, which showed that P. massiliensis is a new Protochlamydia species.

A CRISPR subtype I-E locus was identified in P. massiliensis (Bertelli et al., 2016). In a phylogenetic tree built using 43 cas1 proteins (Figure S2), P. massiliensis groups within the Proteobacteria cluster with a higher bootstrap value (100%). A total of 15 spacers were identified in the P. massiliensis genome (Figure S3), showing 40.80% GC content, and 42.75% genomic GC content. The P. massiliensis spacers have no significant identity in the NR database. However, a search against the “My CRISPRs DB” database (Grissa et al., 2007) enabled us to correlate a part of its sequence to a spacer from Spirochaeta thermophila CRISPR. In spite P. massiliensis was isolated from warm water of cooling tower, there is no evidence now of relationships between this later and S. thermophila, a strict anaerobic extracellular extremely thermophilic spirochete isolated from salt water hot springs. P. massiliensis CRISPR repeats have 28 bp length, the first fourteen repeats sequences displayed total identity and four substitutions were observed in the last two repeat sequences (C/T, A/G, and A/T).

(mas.si.li.en′sis, L. fem. adj. massiliensis, referring to Massilia, the Latin name for Marseille, where the strain was characterized). Phylogenetic position, Chlamydiales order; Gram-negative; mature infectious particles have coccus-shaped morphology 0.6 ± 0.2 mm in size; the access number for the genome at Gen Bank is NZ_CCJF01000000; this strain has been deposited in the CSUR (http://www.mediterranee-infection.com/article.php?laref=14&amp;titre=collection-de-souches Collection de Souches de l′Unite des Rickettsies) under reference CSUR (P2508). Does not grow on cell-free media; obligate intracellular pathogen of multiple amoebal hosts such as A. castellanii, A. polyphaga, and V. vermiformis. P. massiliensis has a development cycle with two morphological stages, typical of chlamydia, and multiplies through binary fission but lacks inclusion vacuoles in the V. vermiformis host.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.