Introduction

The Chlamydiaceae family comprise obligate intracellular replicating human and animal pathogens which are of considerable clinical and epidemiological importance worldwide. Prominent members are Chlamydia (C.) trachomatis, C. pneumoniae, C. psittaci and C. abortus. C. trachomatis causes oculogenital infection, including blinding trachoma, lymphogranuloma venereum and non-gonococcal urethritis and cervicitis. C. pneumoniae causes infections of the upper and lower respiratory tract, and C. psittaci is the causative agent of ornithosis, a severe pneumonia that can be transmitted from birds to humans. C. abortus is associated with ruminant abortion and has been broadly detected in sheep, goats and calves causing considerable economic damage in the agriculture worldwide (Longbottom and Coulter 2003). As a zoonotic pathogen, it can be transmitted to pregnant women posing a life-threatening health risk. If not treated appropriately, the patients may develop severe septicaemia with disseminated intravascular coagulation resulting in spontaneous abortion of the child (Baud et al. 2008; Walder et al. 2005).

All members of the Chlamydiaceae share a biphasic developmental cycle involving infectious, spore-like elementary bodies (EB: about 300 nm in diameter) and metabolically active reticulate bodies (RB: about 1 μm in diameter), which divide within a non-fusogenic inclusion. This distinct developmental cycle is known for over 80 years, since the first description by Bedson and Bland (1932). However, the molecular details of the host–pathogen interaction including cell entry, intracellular differentiation and release are poorly understood.

Electron microscopy observations in the 1970s and 1980s revealed new ultra morphological features: Matsumoto et al. described ‘surface projections’ and ‘B-structures’ on RBs and EBs of C. psittaci (Matsumoto and Manire 1970; Matsumoto 1982a, b). The Chlamydia type III secretion system (T3SS) has been identified as a molecular syringe-like structure behind these surface projections. It enables chlamydiae to secret effector proteins into the host cell, which is considered nowadays to be a central mechanism of chlamydial virulence being essential for the intracellular survival (Betts-Hampikian and Fields 2010; Peters et al. 2007). Until the genetic proof in 1997 for the existence of the T3SS in chlamydiae (Hsia et al. 1997), the morphological observations of these surface projections were the first and only hints that chlamydiae are equipped with a T3SS.

In general, studies systematically investigating ultrastructural features and species specific differences are rare. Chi et al. (1987) demonstrated that EBs of C. pneumoniae in contrast to C. trachomatis and C. psittaci show a ‘pear-shaped’ outer membrane with a pointed end. This distinct morphological difference has been proposed to lead to a different process of attachment to and endocytosis by the host cell (Kuo et al. 1988).

In this report, we describe ultrastructural features of C. abortus strain S26/3 along the developmental cycle in experimentally infected HeLa229 and McCoy cells in comparison with C. trachomatis and C. pneumoniae.

Materials and methods

Organisms and cultivation

The strains C. abortus S26/3 (Thomson et al. 2005), C. trachomatis (serovar D strain UW-3/Cx, American Type Culture Collection: VR-885) and C. pneumoniae (TWAR strain TW-183, ATCC: VR-2282) were used for the study and cultivated on HeLa229 cell monolayers (ATCC: CCL-2.1) or McCoy cell monolayers (ATCC: CRL-1696) in Quantum 101 (PAA Laboratories, Pasching, Austria) according to standard procedures.

Sapphire discs (3 mm diameter, thickness about 50 μm; Brügger, Minusio, Switzerland, distributed by Engineering Office M. Wohlwend GmbH, Sennwald, Switzerland) were cleaned in sulphuric acid, soapwater and absolute ethanol and subsequently coated with 20 nm of carbon by electron beam evaporation with a BAF 300 (BAL-TEC, Balzers, Principality of Liechtenstein). After UV irradiation, sapphire discs were placed into the cavities of a 12-well culture plate and HeLa229 cells or McCoy cells were seeded in the culture plate wells to obtain confluent cell monolayers on sapphire discs. Cell monolayers were experimentally infected with C. abortus, C. trachomatis and C. pneumoniae, respectively, at a multiplicity of infection of five.

High-pressure freezing (HPF)

Sample preparation for electron microscopy was carried out as described (Walther et al. 2013) with some modification. Sapphire discs containing infected cells were high-pressure frozen without any chemical pre-fixation at different points after infection (6 or 8 h intervals, depending on chlamydial species) using the Wohlwend HPF Compact 01 (Engineering Office M. Wohlwend GmbH, Sennwald, Switzerland). The sapphire discs were clamped between two aluminium planchettes (diameter 3 mm). One planchette was flat, and the other had a central cavity of 100 μm in depth and 2 mm in diameter. The cavities between the tissue and the aluminium planchettes were filled with hexadecene (Studer et al. 1989). Hexadecene is not miscible with water and does not serve as a cryo-protectant, but helps to transfer the pressure and temperature during HPF. The frozen samples were stored in liquid nitrogen until freeze-substitution was performed.

Freeze-substitution and embedding

The high-pressure frozen samples were carefully removed from the aluminium planchettes in liquid nitrogen and immersed in 1 mL freeze-substitution medium pre-cooled to −90 °C. The freeze-substitution media consisted of acetone (VWR International GmbH, Darmstadt, Germany) with 0.2 % osmium tetroxide (Merck KGaA, Darmstadt, Germany), 0.1 % uranyl acetate (Merck KGaA, Darmstadt, Germany) and 5 % of water for good membrane contrast (Buser and Walther 2008). We used a specially designed computer-controlled substitution apparatus (A. Ziegler and W. Fritz, Central Facility for Electron Microscopy, Ulm University, unpublished data) to slowly warm the samples from −90 to 0 °C over a period of 16 h in an exponential warming protocol. The samples were kept at room temperature for 30 min, washed with the respective solvent and embedded in a two-step Epon series (Fluka) of 50 % Epon in solvent for 1 h, 100 % Epon for 6 h. The Epon was then polymerized for 48 h at 60 °C, and 60-nm-thin sections were cut with a Leica Ultracut UCT ultra-microtome using a diamond knife (Diatome, Biel, Switzerland), collected on bare copper grids. Section staining with lead citrate was performed.

For immuno-electron microscopy, the freeze-substitution medium did not contain osmium tetroxide and low-temperature embedding was performed to better conserve antigenicity. So, the freeze-substitution medium consisted of acetone (VWR International GmbH, Darmstadt, Germany) with 0.1 % uranyl acetate (Merck KGaA, Darmstadt, Germany) and 5 % of water. For low-temperature embedding, LR-Gold samples were substituted from −90 to −20 °C for 16 h, washed twice with pre-cooled acetone, infiltrated for 2 h with 50 % (v/v) LR-Gold in acetone and for 1 h with 100 % LR-Gold. The polymerization was carried out at −18 °C over night under an UV-lamp. LR-Gold always contained 0.1 % (w/w) benzil and 0.1 % (w/w) benzoyl peroxide. Sections were cut as described above, collected on carbon-formvar-coated copper grids and marked with different gold-coupled antibodies as described below.

Transmission electron microscopy (TEM)

The ultra-thin sections were imaged with a JEM-1400 transmission electron microscope (Jeol GmbH, Eching, Germany) at an acceleration voltage of 120 kV. The images were recorded with a Veleta digital camera (Olympus Soft Imaging Solutions GmbH, Münster, Germany) with a resolution of 2,024 × 2,024 pixel and the iTEM software (Olympus Soft Imaging Solutions GmbH, Münster, Germany). No additional image processing apart from brightness/contrast correction and insertion of marks was performed. Multiple sections of at least two independent samples were compared for their structural quality.

Scanning transmission electron microscopical tomography (STEM tomography)

For STEM tomography, semi-thin sections of 500 nm thickness were collected on bare copper grids with parallel grid bars in one direction only. To attach the relatively thick sections on the grid bars, the copper grids were coated with poly-l-lysine (10 % in water) before the sections were attached. The grids were then warmed on a heating table to a temperature of 60 °C to flatten the sections. Afterwards, the sections were again coated with poly-l-lysine to attach on both sample sides the 15-nm colloidal gold particles (Aurion, The Netherlands), serving as fiducial markers for the calculation of the tomograms. Finally, the mounted sections were carefully coated with 5 nm carbon from both sides by electron beam evaporation to increase electrical conductivity and mechanical stability. Before imaging, the samples were plasma-cleaned for 10 s.

Tilt series were recorded with a TITAN microscope (FEI) at 300 kV in scanning transmission mode (STEM) from ±70° in 1° steps and reconstructed using the IMOD software (Kremer et al. 1996) (Boulder Laboratory for 3-D Electron Microscopy of Cells, Boulder, USA) as described by Höhn et al. (2011). For segmentation and data display, the AMIRA software (Visage Imaging GmbH, Berlin, Germany) was used.

Immuno-electron microscopy

Ultrathin sections were incubated in 4 % goat serum in PBS (phosphate-buffered saline, pH 7.6) and incubated with monoclonal mouse-antibodies against chlamydial heat shock Protein 60 (HSP60) (Dianova GmbH, Hamburg, Germany) and chlamydial lipopolysaccharide (LPS) (kindly provided by H. Brade, Forschungszentrum Borstel, Germany) (Fu et al. 1992). After thoroughly washing with PBS, incubation with the second antibody (goat antimouse coupled to 10 nm gold particles, Aurion, Wangeningen, The Netherlands) followed (Roth et al. 1996). The sections were post-fixed with glutaraldehyde and post-stained with uranylacetate. TEM was performed as described above.

Freeze-fracturing and cryo-scanning electron microscopy (cryo-SEM)

For freeze-fracturing, 170-μm-thick sapphire discs have been used as described by Walther et al. (2013). Cells were grown on these discs and infected. The samples were chemically prefixed at 64 h post-infection (p.i.) with 2 % glutaraldehyde. A 50-μm-thick gold spacer ring (diameter 3.05 mm, central bore 2 mm; Plano GmbH, Wetzlar, Germany) was mounted between two 170-μm-thick sapphire discs with the infected cells facing each other. These sandwiches were high-pressure frozen without aluminium planchettes and without the use of hexadecene.

The frozen sandwiches (either sapphire or aluminium) were mounted on a Bastacky holder (Walther 2003). After transfer to the BAF 300 freeze-etching device (Bal-Tec, Principality of Liechtenstein), the temperature of the sample stage was raised to 155 K (vacuum about 2 × 10−7 mbar). The sandwiches were cracked with a steel knife and then double-layer coated as described by Walther et al. (1995) by electron beam evaporation with 3 nm of platinum–carbon from an angle of 45° and perpendicularly with about 6 nm of carbon. Thereby, the additional carbon coat increases electrical conductivity (prevents charging) and, even more importantly, reduces the effects of beam damage.

The specimens were imaged with a S-5200 ultra-high resolution field emission scanning electron microscopy (Hitachi High Technologies America, Schaumburg, USA) at an acceleration voltage of 10 kV using the backscattered electron signal (Walther et al. 1995). No additional image processing apart from brightness/contrast correction and insertion of marks was performed.

Statistic evaluation

To determine the amount of different morphological forms of intracellular chlamydiae, the inclusion of seven cells visible in thin sections based on two independent infection experiments was analysed for each point in time after infection. RBs, dividing RBs, RBs with vesicles in the periplasmic space, intermediate body (IBs) and EBs were distinguished. The total number of counted organisms was about 300 at early time points and up to 8,000 at late time points. The relative frequency of one type of morphological form is expressed as the arithmetic average for each point in time after infection. The size of the measured objects was calculated as the average of the two largest perpendicular diameters.

Results

The systematic analysis of the ultrastructure of C. abortus during its entire developmental cycle showed that it shares the known main features of the Chlamydiaceae, involving the uptake of EBs by the host cell, the intracellular maturation and division of RBs, the condensation of RBs to EBs and finally the release of EBs into the host cell’s environment. The duration of the developmental cycle of C. abortus was set between C. trachomatis (ca. 48 h) and C. pneumoniae (ca. 72 h) and took about 56 h (Fig. 1).

Fig. 1
figure 1

Relative amount of different morphological forms of chlamydiae inside the inclusion of one infected HeLa229 cell during the second half of the developmental cycle for each point in time after infection and each chlamydial species. The duration of the developmental cycle for each species was set due to the ratio of RBs to EBs. TEM samples of HeLa229 cells infected with each chlamydial species, respectively, were evaluated. For each point in time after infection, the entirety of morphological forms of the inclusions of seven host cells visible in thin sections based on two independent infection experiments was counted (300–8,000 organisms per sample depending on the point of time after infection). The height of a bar shows the arithmetic average of the relative amount, and the error bar indicates the standard deviation

Full size image

Strikingly, three noteworthy features occurred which have not been reported yet and are presented in detail in the following.

The outer membrane of C. abortus exposed surface projections and B-structures

Cryo-SEM images of freeze-fracture samples of C. abortus infected cells were analysed for the presence of ultrastructural correlates of the T3SS. Typical structures were basically identified on the surfaces of EBs at a late stage of the developmental cycle. These morphological correlates of the T3SS could only be found on approximately 10 % of EBs per inclusion. A SEM image showing ultra-structures on the surface of an EB (350 nm in diameter) being representative for the T3SS is illustrated in Fig. 2. The true surface of an EBs outer membrane showed 10–20 projections each arising from the centre of a flower-like structure. This structure measured 36 nm in diameter and consisted of 8–10 radial arrayed leaves. The inner concave face of an EBs outer membrane revealed 10–20 button structures (B-structures). These B-structures measured 30 nm in diameter and had a central cavity of about 8 nm in diameter. Both the projections on the outer convex surface and the B-structures on the inner concave surface were hexagonally arrayed and could only be found in a limited area of an EBs outer membrane surface. The centre-to-centre spacing between the different projections or B-structures was about 52 nm, respectively.

Fig. 2
figure 2

Projections and B-structures upon the surface of EBs inside an inclusion as observed by cryo-SEM of a freeze-fracture sample of C. abortus infected HeLa229 cells 64 h p.i. a The true surface of an EB is shown. White arrowhead points towards projections on the outer membrane of an EB. White arrow indicates the concave face of an outer membrane of a chlamydial body revealing the button structures (B-structures). Bar 250 nm. b A larger scale of magnification: each projection is arising from the centre of a flower-like structure consisting of several leaves (example marked by a circle), and each B-structure shows a cavity in its centre. Bar 100 nm

Full size image

Membrane complexes associated with the C. abortus inclusion occurred in the second half of the developmental cycle

A remarkable feature of the C. abortus developmental cycle could be observed from 24 h p.i. onwards: the formation of massive membrane complexes inside the host cell’s cytoplasm in close association with the inclusion (Fig. 3). In thin sections, more than 70 % of the infected host cells exhibited these membrane complexes. They consisted of tubes bordered by a single bi-layered membrane. The inside of those tubes showed a similar electron density as the host cell’s cytoplasm. Their thickness varied from 30 to 150 nm depending on the surrounding space. In some samples, the tubes were pushed closely together in stacks so that their bordering membranes added up to a pattern as shown in Fig. 3c: long rails of pairs of membranes running around one pole of an inclusion. These closely adjacent membranes resemble to the inner and outer membrane of an active mitochondrion, the only organelle having such closely related membranes. Sometimes, a single membrane of one tube made a u-turn thereby bordering a blind end of the tube (Fig. 3b, d). In the three-dimensional reconstruction of a membrane complex, these tubes exhibited such a height that they seem to be rather membrane walls (Fig. 4). The spacial build-up of these walls to a membrane complex was intricate; at some section levels, they appeared to be encased like onion skins or simply lying upon another in stacks. The number of complex-forming membrane tubes varied between 2 and 12. The dimension of the membrane complexes inside the host cell was up to 12 μm depending on the point in time after infection.

Fig. 3
figure 3

Lamelliform inclusion membrane complex running like rails all around the inclusion as seen by TEM of a thin section of C. abortus infected HeLa229 cells 30 h p.i. a Overview: white arrowheads point towards the inclusion membrane (also in c). Bar 1 μm, details in b, c, d. b Black arrow shows the switching point of the membrane (also in d), where it runs into the opposite direction thereby bordering a blind-ending tube. Black arrowhead points to the plasma membrane of the host cell (also in c and d). Bar 200 nm. c 10 membranes run parallel to each other around the inclusion. Bar 200 nm. d Inclusion membranes run around the inclusion pole. Bar 200 nm

Full size image
Fig. 4
figure 4

Three-dimensional intricacy of the C. abortus membrane complexes inside infected HeLa229 cells 30 h p.i. as revealed by STEM tomography and 3D-reconstruction. An inner pair of membranes (orange) runs parallel to an outer, encoating pair (purple) which has an extension of 4 μm. In between those membranes, there is a further membrane vesicle (green). The compartments between the different membranes show an electron density similar to the cytoplasm and a lesser electron density in an alternating pattern. a, b Two sections (section level shown in c, d) of the calculated volume with and without reconstruction show that those two membrane pairs (orange, purple) merge at some section levels. N host cell’s nucleus, M mitochondria, G golgi, bar 2 μm. c, d The three-dimensional reconstruct of the membrane complex from different perspectives. The purple pair of membranes is cut along the blue line to reveal the view upon the orange pair of membranes and their discreet merging point

Full size image

The subcellular localization of the membrane complexes was strongly associated with the inclusion. In more than 90 % of the membrane complex exhibiting cells, it seemed that the inclusion membrane itself had been twisted around forming the complex. This inclusion membrane complex was not found around the entire inclusion’s circumference, but exhibited around the nucleus-orientated pole of the inclusion or, in case of multiple inclusions in one host cell, between multiple inclusions (Figs. 5, 7). In less than 10 % of the infected cells membrane complexes were found somewhere in the host cell’s cytoplasm without any association with the inclusion. Neither in C. trachomatis nor in C. pneumoniae infected cell cultures, membrane complexes could be found (data not shown).

Fig. 5
figure 5

SEM images of freeze-fracture samples showing the inclusion membrane complex and RBs inside the inclusion of a C. abortus infected HeLa229 cell 64 h p.i. a Overview, N host cell’s nucleus, asterisk inclusion, bar 5 μm, detail in b. b White arrows point to the inclusion membrane, which shows its C. abortus specific lamelliform structure consisting of more than just one membrane running around the inclusion. Bar 1 μm, detail in c. c Near the inclusion membrane, there is an RB with a fractured outer membrane revealing its periplasmic space, and white arrowhead points towards a vesicle inside the periplasmic space and towards a protuberance of the RB’s main body which is possibly a budding vesicle. Bar 250 nm

Full size image

Further observations on the distribution of host mitochondria inside C. abortus infected host cells revealed strong mitochondrial association with the inclusion and the inclusion membrane complex (Figs. 6, 7). From the middle-to-late stages, mitochondria were found huddling against the inclusion membrane with a minimal spacing of 4 nm. The ultrastructure of these mitochondria associated with the inclusion membrane changed during the late stages of the developmental cycle. The typical cristae structure was fading. Instead, structures consisting of electron-dense vesicles bordered by one outer membrane occurred at late stages of the developmental cycle (Fig. 7c, d). These structures gave the impression of perishing mitochondria.

Fig. 6
figure 6

Relation between mitochondria and the C. abortus inclusion inside infected HeLa229 cells 42 h p.i. as observed by TEM. a Black arrows point towards mitochondria next to the inclusion huddling against the inclusion membrane. Inside the inclusion, there are RBs, RBs with vesicles in the periplasmic space, IBs and EBs. N host cell’s nucleus, bar 2 μm, details in b, c. b, c A mitochondrion (b) next to the inclusion showing its christae which seem to fall apart forming small vesicles, just like a much less healthier mitochondrion in the neighbouring cell (c). White arrow points towards a membrane complex around the C. abortus inclusion. Bar 500 nm. d Another unhealthy looking mitochondrion next to an inclusion harbouring RBs with vesicles inside the periplasmic space. Bar 200 nm

Full size image
Fig. 7
figure 7

Chlamydia abortus inclusions at 24 and 48 h p.i. showing the differences between the proliferation and the terminal phase of the C. abortus developmental cycle inside HeLa229 cells as seen by TEM a 24 h p.i.: RBs proliferating inside the yet small (5 μm in diameter) inclusion. Their periplasmic space is broadened at one side of the RB to about 100 nm. White arrow points to lamelliform inclusion membrane, and black arrowhead points towards mitochondrion lying right next to the inclusion. Bar 1 μm. b 42 h p.i.: taking most of the space inside the host cell, the inclusion has grown to a size of 13 μm in diameter containing many different morphological forms of chlamydiae. Most of these organisms show a broadened periplasmic space (up to 400 nm) harbouring small membrane-bordered vesicles (50–100 nm). White arrow points to a smaller inclusion entirely coated by a complex of five directly opposing pairs of membranes (altogether 11 membranes). N host cell’s nucleus, bar 2 μm, detail in c, d. c, d The black arrowhead points towards mitochondria outside the inclusion which do not show their typical christae, but seem to fall apart into small vesicles bordered by the outer mitochondrial membrane. N host cell’s nucleus, bar 400 nm

Full size image

Chlamydia abortus showed periplasmic vesicles during the late stages of the developmental cycle

Chlamydia abortus inclusions are shown at 24 h p.i. and 42 h p.i. (Fig. 7). The ultrastructure of RBs inside the inclusion changed during the developmental cycle. Up to 24 h p.i., RBs’ inner and outer membrane lay to one another whereby the periplasmic space measured 20 nm at most or was hardly discernible in TEM sections. At 24 h p.i., the periplasmic space began to widen starting at one side of an RB. From this time on to the late stages of the developmental cycle, vesicles in the periplasmic space of RBs occurred in every infected host cell (Figs. 7b–d, 8, also in Figs. 5c, 6b, d). Their number varied from 1 to 15 depending on the point of time after infection. These vesicles showed a spherical to tubular shape and measured about 62 ± 23 nm (mean ± standard deviation) in diameter (average of 200 vesicles). Their inside had the same electron density as the main body. They were bordered by a single bi-layered membrane. Some RBs showed vesicular protuberances of the same size as the vesicles, possibly being the morphological correlate to the process of vesicles either fusing with or budding off the main body. Immuno gold-labelling of chlamydial LPS and HSP60 (Fig. 8c, d) indicated that these chlamydial components were found inside the vesicles suggesting that budding off was the more probable process behind the protuberances. Figure 9 gives an idea of the three-dimensional morphology of RBs with vesicles in the periplasmic space underlining the spherical to tuberal shape of vesicles and the fact of protuberances being vesicles still connected to the main body. The outer membrane encased the main body and the vesicles without any discontinuity.

Fig. 8
figure 8

RBs with vesicles in the periplasmic space of different species inside infected HeLa229 cells as seen by TEM: C. abortus (a, b 30 h p.i., c, d 42 h p.i., bars 100 nm), C. trachomatis (e 18 h p.i., bar 500 nm) and C. pneumoniae (f 40 h p.i., bar 500 nm). a White and black arrowhead point towards the inner and outer membrane of the RB; in between those membranes, there is the periplasmic space which harbours spherical membrane-bordered vesicles (black arrow) containing material showing the same electron density as the main body. These vesicles could be found in every infected host cell from 24 h p.i. to the late stages of the developmental cycle. b RB at another section level: the vesicles are still in connection with the main body, and they are possibly just budding off. c, d The immunogold-labelling of C. abortus RBs using a murine anti-chlamydial LPS-antibody (c) and a murine anti-chlamydial HSP60-antibody (d) with a goat antimouse-antibody coupled to 10 nm gold particles as secondary antibody indicates that the vesicles harbour chlamydial components. e C. trachomatis RBs show slightly smaller vesicles in the periplasmic space than C. abortus RBs on average. f C. pneumoniae RBs typically show an expansion of the periplasmic space at a pointed end of the RBs which harbours the vesicles

Full size image
Fig. 9
figure 9

Three-dimensional structure of C. abortus RB with vesicles in its periplasmic space inside an inclusion of an infected HeLa229 cell 30 h p.i. illustrated by STEM tomography and 3D reconstruction. a Overview, bar 1 μm. b–e The reconstructed RB consists of the main body (red): a homogenous, electron-dense mass with a size of about 700 nm in diameter bordered by a single membrane. The main body shows some protuberances (light red) that are possibly vesicles currently budding off (white arrow). Around the main body, there is another single membrane bordering the periplasmic space: the outer membrane of the RB (transparent turquoise). Besides the main body, there are further spherical and tubular bodies inside the periplasmic space (white) of the same electron density as the main body, but of smaller size of about 50–100 nm in diameter. Bars 200 nm

Full size image

The number of C. abortus RBs with vesicles in the periplasmic space started to increase from 24 h p.i. and reached its maximum at 36 h p.i. with a relative amount of 41 % (Fig. 1). In this time span, the number of normally shaped RBs constantly decreased. From 36 h p.i. onwards, the number of RBs with vesicles in the periplasmic space decreased, while the IBs and EBs began to increase. Due to the asynchronous maturation of chlamydial bodies (compare Figs. 6a, 7b, both 42 h p.i.), the standard deviation of the relative amount of differently shaped bodies in one inclusion (indicated as error bars in Fig. 1) was relatively big. However, the dynamic of this sequence of different chlamydial morphologies over the time suggested that RBs with vesicles in the periplasmic space might be an intermediate step during the redifferentiation process of RBs to EBs.

The formation of vesicles in the periplasmic space could also be observed in C. trachomatis and C. pneumoniae (Fig. 8e, f). C. trachomatis showed slightly smaller vesicles than C. abortus and C. pneumoniae, and they measured 55 ± 16 nm (average of 200 vesicles). Strikingly, in C. abortus and C. trachomatis, the widened periplasmic space harbouring the vesicles was located like a crescent at one side of an RB or was massively widened, whereas the periplasmic space of C. pneumoniae was predominantly widened at one pointed end giving just enough space to be tightly filled with vesicles. Due to the different duration of the developmental cycle (C. trachomatis: ~48 h, C. pneumoniae: ~72 h, C. abortus: ~56 h), the point of time of the occurrence of the vesicles varied between the species (Fig. 1). But the dynamic of the strict sequence of the occurrence of the different morphologies was the same in C. trachomatis and C. pneumoniae as it was in C. abortus.

All of these results were independent from the addition of cycloheximide and did not show any variance in McCoy cells compared to HeLa229 cells as host cells.

Discussion

In this study, we used electron microscopic methods to show several ultrastructural features of C. abortus amongst which the occurrence of vesicles in the periplasmic space of RBs in the late stage of the developmental cycle seems to be common with other chlamydial species, whereas the formation of massive membrane complexes inside the host cell is a C. abortus specific phenomenon.

Furthermore, projections and B-structures on the surface of C. abortus EBs were identified, which correspond in detail to those structures first described by Matsumoto (1982b) and later hypothetically attributed to the T3SS of the Chlamydiaceae (Bavoil and Hsia 1998; Peters et al. 2007). The equipment of C. abortus with the T3SS seems to be probable since analysis of the genome revealed the presence of homologues to all genes thought to encode the chlamydial T3SS (Thomson et al. 2005). While a morphological as well as a molecular biological proof for the equipment with a functional T3SS apparatus has been provided for C. trachomatis (Gregory et al. 1979; Fields et al. 2003), C. pneumoniae (Miyashita et al. 1993; Lugert et al. 2004) and C. psittaci (Matsumoto 1973; Beeckman et al. 2008), the illustration of projections and B-structures on C. abortus EBs has been the first morphological evidence for the existence of the T3SS in C. abortus.

A phenomenon that seems to be exclusive to the species of C. abortus is the formation of membrane complexes that seem to develop from the inclusion membrane inside infected host cells at late stages of the developmental cycle. The questions we asked ourselves about the membrane complexes were: ‘What material is used for their formation?’ ‘How are they formed?’ and ‘What function do they have?’ At our momentary state of knowledge, answers to these questions are speculative. To gather material for the formation of such an amount of membranes, C. abortus would have to exploit the membrane pools of the host cell, which are organelles such as the Golgi apparatus, the endoplasmatic reticulum and the mitochondria. Considering the close proximity to the inclusion, already described for C. psittaci by Matsumoto et al. (1991), and the seemingly perishing of mitochondria in the late stages of the developmental cycle, one might consider host mitochondria to be the origin of the membrane complexes of C. abortus. The ultrastructural similarity between the closely adjacent membranes of the membrane complexes and the inner and outer membrane of active mitochondria supports this idea (Höhn et al. 2011; Walther and Ziegler 2002).

Moreover, the question which chlamydial mechanisms are involved in the process of the membrane complex formation and material gathering is challenging. It is known that type three secreted virulence factors enable chlamydiae to hijack host cells’ organelles (Heuer et al. 2009; Derré et al. 2011; Elwell et al. 2011). The family of Inc proteins found in all chlamydial species belongs to the type three secreted virulence factors (Dehoux et al. 2011). These proteins have been shown to be crucial for the formation and maintenance of the inclusion membrane (Lutter et al. 2012). A characteristic feature shared by all Inc proteins is a bilobal hydrophobic domain which is predicted to enable the Inc proteins’ insertion into the inclusion membrane. Mital et al. (2013) recently showed that the ectopic expression of C. trachomatis Inc proteins with an intact bilobal hydrophobic domain in HeLa229 cells leads to the formation of ‘membranous vesicular compartments’. The ultrastructure of these membranous vesicular compartments strongly reminds of the described membrane complexes of C. abortus. Therefore, Inc proteins might play an important role in the formation of the membrane complexes of C. abortus. In silicio predictions show that C. abortus is equipped with 72 putative Inc proteins (Dehoux et al. 2011). Further investigations on these Inc proteins may contribute to a better understanding of the formation of the observed membrane complexes.

However, the function behind these membrane complexes remains unclear. Because of the lack of a sufficient ribonucleoside triphosphates generation in chlamydiae and a permanent shortage of nutrients due to the living in an intracellular compartment (Saka and Valdivia 2010; Tipples and McClarty 1993; Stephens et al. 1998), the acquisition of such nutrients seems to be a probable suggestion. But as such membrane complexes could not be observed in other chlamydial species, the question emerges: is it imaginable that such an important mechanism is not conserved throughout the Chlamydiaceae?

The morphological appearance of the Chlamydiaceae can be classified into two distinct forms: EBs and RBs. In between those forms, there is the IB, a poorly defined form. We showed the existence of RBs with vesicles in the periplasmic space being an unknown morphological appearance of chlamydiae.

However, Tanami and Yamada (1973) described the formation of ‘miniature cells’ in C. psittaci looking quite the same as the RBs with vesicles in the periplasmic space mentioned here. Their explanation for the formation of miniature cells was a disturbance in membrane synthesis due to penicillin effects. Chi et al. (1987) described EBs of the TWAR strains of C. pneumoniae being enclosed by a wavy outer membrane and ‘pear-shaped’ in profile. In the widened periplasmic space, so-called  ‘miniature bodies [which] seemed to attach to the cytoplasmic mass by a thin peduncle’ could be found. These ultra-structures also remind of the RBs with vesicles in the periplasmic space mentioned here. However, Kuo et al. (1988) did not find a penicillin dependent formation of these ultra-structures, but found out that the different morphology is linked to a function in the endocytosis of TWAR EBs by the host cell.

The occurrence of vesicles in the periplasmic space of RBs in the second half of the developmental cycle is independent of penicillin/cycloheximide addition, the type of host cells and type of chlamydial species. Therefore, we suggest that this ultrastructure is a common feature of Chlamydiaceae during their intracellular development. A likely function behind this distinct ultrastructure is the involvement in the redifferentiation process of RBs into EBs at the end of the developmental cycle. Little is known about the process of redifferentiation. Current conceptions propose that every component of the RB, such as the genome, cytoplasm and membrane, is condensed forming the EB (Huang et al. 2010). We suggest that the ejection of vesicles might also play a role during the redifferentiation which is hypothetically illustrated in Fig. 10. The RB discards redundant material not needed in the metabolically inert EB by ejection of vesicles into the enlarged periplasmic space. This leads to a reduction in RB size. At the same time, DNA condensation takes place inducing the formation of the IB with its centric situated nucleus equivalent. Further condensation of the whole IB finally leads to the EB distinguished by its smaller size, its excentric situated nucleus equivalent and its cytoplasm with a higher electron density compared to RBs. Thus, the described morphology of RBs with vesicles in the periplasmic space is assigned a place in the process of re-differentiation as an intermediate step between the normally shaped RB and the IB. Clearly, further investigations are needed to elucidate the molecular functions behind the newly discovered ultra-structures of the C. abortus developmental cycle.

Fig. 10
figure 10

A hypothetical composition of different C. abortus morphologies found inside HeLa229 cells 42/48 h p.i. as observed by TEM. The different morphologies may represent different stages of development. The arrow indicates a causal sequence of an RB redifferentiating into an EB at the end of the developmental cycle as a part of the condensation process. Bar 200 nm

Full size image