|Home | About | Journals | Submit | Contact Us | Français|
Ureaplasma species are the bacteria most frequently isolated from human amniotic fluid in asymptomatic pregnancies and placental infections. Ureaplasma parvum serovars 3 and 6 are the most prevalent serovars isolated from men and women. We hypothesized that the effects on the fetus and chorioamnion of chronic ureaplasma infection in amniotic fluid are dependent on the serovar, dose, and variation of the ureaplasma multiple-banded antigen (MBA) and mba gene. We injected high- or low-dose U. parvum serovar 3, serovar 6, or vehicle intra-amniotically into pregnant ewes at 55 days of gestation (term = 150 days) and examined the chorioamnion, amniotic fluid, and fetal lung tissue of animals delivered by cesarean section at 125 days of gestation. Variation of the multiple banded antigen/mba generated by serovar 3 and serovar 6 ureaplasmas in vivo were compared by PCR assay and Western blot. Ureaplasma inoculums demonstrated only one (serovar 3) or two (serovar 6) MBA variants in vitro, but numerous antigenic variants were generated in vivo: serovar 6 passage 1 amniotic fluid cultures contained more MBA size variants than serovar 3 (P = 0.005), and ureaplasma titers were inversely related to the number of variants (P = 0.025). The severity of chorioamnionitis varied between animals. Low numbers of mba size variants (five or fewer) within amniotic fluid were associated with severe inflammation, whereas the chorioamnion from animals with nine or more mba variants showed little or no inflammation. These differences in chorioamnion inflammation may explain why not all women with in utero Ureaplasma spp. experience adverse pregnancy outcomes.
Ureaplasma species are among the smallest free-living, self-replicating bacteria and are bounded only by a cell membrane. The Ureaplasma species, U. parvum and U. urealyticum, are the microorganisms most frequently isolated from human amniotic fluid  and the placenta [2–4] and are the bacteria most frequently associated with preterm birth . The asymptomatic ureaplasma colonization of amniotic fluid collected at the time of amniocentesis for genetic testing has been associated with fetal loss , premature rupture of membranes , preterm labor, and preterm delivery, but also with apparently normal pregnancy outcome [6–8].
The original 14 serovars (and 2 biovars) of U. urealyticum are now classified into two separate species: U. parvum (serovars 1, 3, 6, and 14) and U. urealyticum (serovars 2, 4, 5, and 7–13) . Ureaplasma parvum serovar 3 is the most prevalent serovar detected in nonpregnant women [10–12], pregnant women , and infertile women and men , and serovar 6 is the second most prevalent ureaplasma also found in both men and women. Ureaplasma parvum serovar 6 was the most frequently isolated serovar from women who delivered preterm , and this same serovar was also the most adherent to spermatozoa and was detected most frequently in washed semen samples, after standard assisted-reproductive-technology semen-washing procedures .
Ureaplasmas and mycoplasmas can bind to host cell-surface glycolipids such as sulfogalactogylcerolipid, found on the surface of spermatozoa, and sulfogalactosyl ceramide, a component of the glycolipid fraction of the human endometrium; ureaplasma adherence to these receptors may interfere with sperm/egg recognition and implantation, respectively . However, spermatozoa infected with ureaplasmas in vivo  and in vitro  have higher motility levels, and ureaplasmas adherent to the surface of spermatozoa could therefore gain access to the female upper genital tract and colonize the endometrium prior to implantation and the amniotic fluid persistently throughout pregnancy.
The multiple-banded antigen (MBA) is a major virulence factor of Ureaplasma species [17, 18], is a ureaplasma-specific, surface-expressed lipoprotein. The MBA is the predominant antigen recognized by antibodies from ureaplasma-infected humans tested by Western blotting . The MBA contains a signal peptide and acylation site in the N-terminal, whereas the C-terminal consists of multiple repeat units and serovar-specific and cross-reactive epitopes . The multiple banded antigen gene (mba) consists of a 5′ region that is conserved in size in all 14 serovars and a 3′ region that consists of multiple repeat units (18 bp for U. parvum serovar 3, 12 bp for U. parvum serovar 6), which vary in number both in vivo and in vitro . Size variation of the MBA may be a mechanism by which ureaplasmas avoid recognition by the host immune system . Previously, we demonstrated that intra-amniotic injection of U. parvum serovar 3 and U. parvum serovar 6 in sheep caused histologic chorioamnionitis, fetal pulmonary colonization and inflammation, and induced lung maturation [23–25]. For this study, we hypothesized that the effects on the fetus and the chorioamnion of ureaplasma colonization of amniotic fluid are dependent on the serovar, the dose, and the variation of the MBA. We investigated the effects, at 125 days of gestation, of 70-day colonization of the amniotic fluid beginning on day 55 of gestation (term is 150 days) and compared separately the effects of U. parvum serovar 3 or U. parvum serovar 6 inoculations given at both high and low doses. We also asked if there was a relationship between size variation of the MBA/mba and the severity of chorioamnionitis in pregnant sheep.
All experimental procedures were approved by the Animal Ethics Committees of The University of Western Australia, Queensland University of Technology, and Cincinnati Children's Hospital. Ureaplasma parvum serovars 3 and 6 (serovar identities confirmed by PCR assays)  used in these experiments were originally isolated from semen samples collected from men attending the Wesley IVF Service (Brisbane, QLD, Australia); patients gave informed consent for the use of the samples for research. Low-passage ureaplasmas were prepared for injection using first passage (P1) and P2 ureaplasmas and stored at −80°C . Before injection, the U. parvum was thawed and diluted in sterile, cold PBS to concentrations of 2 × 104 colony-forming units (CFU; low dose) or 2 × 107 CFU (high dose) in injection volumes of 2 ml. Inoculates were mixed, kept on ice, and warmed immediately prior to injection.
Merino ewes (JRL Hall & Co., Darkan, WA, Australia) bearing single fetuses were randomly allocated to one of five treatment groups (n = 8 per group) in order to receive a single ultrasound-guided intra-amniotic injection at 55 days of gestation: U. parvum serovar 3 low dose (S3LD), U. parvum serovar 3 high dose (S3HD), U. parvum serovar 6 low dose (S6LD), U. parvum serovar 6 high dose (S6HD), or media control. All ultrasound imaging and intra-amniotic injections were performed  in an agricultural environment, and ewes had minimal human contact until they were transported to paddocks at a research facility at 90 days of gestation. Preterm operative deliveries were performed in an adjacent facility at 125 days of gestation.
Anesthesia for the ewes was with i.v. ketamine (12 mg/kg) and metatomidine (0.12 mg/kg), followed by spinal anesthesia with lignocaine (60 mg). Investigators involved in the deliveries and postmortem studies were blinded to the treatment allocation. Deliveries were performed using strict aseptic techniques. The uterus was incised and amniotic fluid was aspirated through intact fetal membranes for subsequent culture and quantitation of U. parvum. The fetus was delivered, and a lethal dose of pentobarbitone (100 mg/kg) was injected into an umbilical vein. Umbilical arterial blood was collected for blood gas and pH analysis (Rapidlab 865; Bayer Diagnostics, Pymble, NSW, Australia) and for total and differential white blood cell counts. Samples of chorioamnion, placenta and umbilical cord, and maternal liver and kidney were frozen in liquid N2 and stored at −80°C for subsequent culture to determine the number of colony-forming units of ureaplasmas per gram of tissue. Samples of chorioamnion were also immersion fixed in 10% buffered formalin for histological analysis.
Each lamb was weighed and exsanguinated. The lungs were weighed, the left lung was lavaged three times with physiological saline solution, and the bronchoalveolar lavages were pooled and used to assess ureaplasma colonization and inflammation. Pieces of the lower lobe of the right lung were frozen in liquid N2 and stored (−80°C) for subsequent analyses. The upper lobe of the right lung was fixed by airway instillation of 10% buffered formalin at 30 cm water pressure. Samples of fetal cerebrospinal fluid (CSF) and lung, liver, spleen, and gut were also collected and frozen for subsequent ureaplasma culture.
The fluid and homogenized tissue samples were cultured in 10B broth  (in 10 × 10-fold dilutions) as previously described  to determine the number of colony-forming units of ureaplasmas per gram of tissue or per milliliter of fluid. To quantify the ureaplasmas, six 5-μl drops of the 10−3, 10−4, and 10−5 inoculated broth dilutions were subcultured onto A8 agar , and the plates were incubated in CO2 at 37°C for 48–72 h; subsequently, colonies were counted at 40× magnification using a stereo microscope (Leica Microsystems, North Ryde, NSW, Australia).
Homogenized chorioamnion was also cultured onto or into a broad range of nonselective, selective, and selective-and-differential culture media to detect other bacterial species. Plates inoculated included horse blood agar, chocolate I agar, MacConkey agar without crystal violet, Sabouraud dextrose agar, anaerobic blood agar, Schaedler anaerobic agar, De Man Rogosa Sharpe agar, and thioglycollate broth (Oxoid, Adelaide, SA, Australia). Aerobes, facultative anaerobes, and microaerophiles were identified at 24–48 h after incubation by traditional microbiological rapid tests (Gram stain and rapid catalase and oxidase tests); confirmatory commercial diagnostic tests API 20E, API Staph, API Strep (Biomerieux, Baulkam Hills, NSW, Australia); Staphytect plus and Streptococcus grouping kits, Bacitracin and Optochin discs (Oxoid); and rabbit plasma coagulase tests (Becton Dickinson, North Ryde, NSW, Australia). Anaerobes were subcultured onto anaerobic blood agar plates after 48 h and incubated for up to 7 days. Anaerobes were identified by API 20A (Biomerieux) strips.
DNA was extracted from each ureaplasma culture as previously described . PCR primers were designed using the U. parvum serovar 3 and serovar 6 genome sequences (Genbank accession numbers NC_002162 and AAZQ01000001, respectively) to amplify the upstream and downstream regions of the mba. The upstream region of the mba of U. parvum serovars 3 and 6 was amplified with the forward primer MPUF 5′TGCAATCTTTATAT GTTTTCG3′ and the reverse primer MPUR2 5′TTAACAAACCTGAAGTCT TG3′. The reaction mixture (50 μl) contained final concentrations of 2.5 U of Platinum Taq (Invitrogen, Mt. Waverly, VIC, Australia), 1× PCR buffer (pH 8.7; Tris HCl, KCl, (NH4)2SO4; Invitrogen), 1.5 mM MgCl2 (Invitrogen), 100 μM of each deoxynucleotide triphosphate (Roche, Castle Hill, NSW, Australia), distilled water, 0.5 μM of each primer, and DNA template (8 μl of prepared sample). The DNA thermal cycler PTC-200 (Global Medical Instrumentation, Ramsey, MN) was programmed for 1 cycle of denaturation at 94°C for 15 min; 35 cycles consisting of denaturation at 94°C for 1 min, annealing for 1 min at 54°C, and extension at 72°C for 1 min; and a final extension cycle of 72°C for 10 min. Positive controls included the original ureaplasma inoculums (serovar 3 and serovar 6) and U. parvum serovar 3 or 6 reference serovars (courtesy of H. Watson, University of Alabama, Birmingham, AL). Two negative controls, master-mix only and dH2O instead of template, were included in each assay.
The downstream region of the mba was amplified with PCR primer sets MPDF3 5′TAATCAAGACTTCAGGTTTG3′ and 3DR3 5′TCGCTTTTTTCATTACGAGTC3′ for U. parvum serovar 3 isolates and MPDF3 and 6DR4 5′TAATGTAAATAAAGCACTTATTC3′ for serovar 6. The assay was performed in a reaction volume of 50 μl, as above; however, the PCR cycling conditions were adapted from Monecke et al.  with an initial denaturation at 94°C for 9 min; 40 cycles of denaturation at 94°C for 45 sec, primer annealing at 54°C for 75 sec, and extension at 72°C for 2 min; and a final extension step at 72°C for 15 min. The PCR products were separated by electrophoresis in a 2% agarose gel, visualized by ethidium bromide staining, and digitized using Grab-IT (Ultraviolet Products Ltd., Cambridge, U.K.).
Because of the complexity of this work, only six amniotic fluid cultures from animals inoculated with S6LD (animals E22 and E36), S6HD (animals E18, E24, E31, and E35), and the original inoculum of U. parvum serovar 6 were selected for cloning and filtration (to obtain ureaplasma cultures expanded from a single ureaplasma colony-forming unit). The cloning and filtration was performed as previously described  and was repeated twice for each specimen. Briefly, the P1 culture (containing a mixture of mba variants) in 10B broth was filtered and then serially diluted in 10B broth (10-fold dilutions), and selected dilutions were subcultured onto A8 agar. After 48 h incubation (in CO2), individual ureaplasma colonies were excised aseptically, placed into fresh 10B broth, and cultured (C1 cultures). The filtration, subculture, and aseptic excision of individual colony-forming units then were repeated (C2 cultures). This process was performed for six selected amniotic fluid specimens and the two controls (the initial inoculum and reference serovar), and for each specimen this resulted in one P1 cultured specimen, six C1 cultures, and 36 C2 cultures (a total of 43 cultures per specimen/control). Each of these cultures was analyzed by upstream and downstream mba PCR assays and Western blot to detect variation of the gene and of the surface-exposed antigen, respectively.
For Western blot analysis P1, C1, and C2 cultures were subcultured in 10 ml of 10B broth, grown to late log stage, centrifuged at 3000 rpm for 30 min, and then the resultant pellet was resuspended in 100 μl of PBS. This prepared antigen (12 μl) with 4 μl of SDS-PAGE loading buffer (Tris HCl [pH 6.8], 50% glycerol, 8% w/v SDS, bromophenol blue, and 1M dithiothreitol) was boiled for 5 min and then electrophoresed by 10% SDS-PAGE  at 120 V for 60 min. Proteins were transferred onto a polyvinylidene fluoride or nitrocellulose membrane (Pall Corporation, Cheltenham, VIC, Australia)  and incubated overnight at 4°C with primary antibody (1:2000 [serovar 6] or 1:8000 [serovar 3]; serovar-specific polyclonal rabbit ureaplasma antisera, courtesy of Dr. Patricia Quinn, The Hospital for Sick Children, Toronto, ON). Then the membrane was washed and incubated with the secondary antibody (1:5000; goat anti-rabbit IgG conjugated with horseradish peroxidise; Sigma-Aldrich Pty Ltd., Castle Hill, NSW, Australia) for 1 h. Proteins were visualized by 3′, 3′-diaminobenzidine tetrahydrochloride with cobalt chloride enhancer (Sigma-Aldrich). Positive controls for the Western blots were the U. parvum serovar 3 and serovar 6 cultures of the initial inoculum and cultures of the reference serovars. A 10B media negative control was incorporated to detect cross-reactivity with media components.
Formalin-fixed chorioamnion samples were paraffin embedded, and 5-μm sections were heated overnight at 60°C before staining with hematoxylin and eosin (H&E). All stained sections were examined blindly to count total numbers of white blood cells and the numbers of monocytes, lymphocytes, and neutrophils present in 20 fields per slide at 1000× total magnification. The H&E tissue sections were also scored by a perinatal pathologist (D.P.) according to the diagnostic criteria of Redline et al. , which defines the maternal inflammatory responses as stage 1 (early; acute subchorionitis or chorionitis), stage 2 (intermediate; acute chorioamnionitis), and stage 3 (advanced; necrotizing chorioamnionitis), and the fetal inflammatory responses as stage 1 (early; chorionic vasculitis or umbilical phlebitis), stage 2 (intermediate; umbilical vasculitis, one or two arteries, and/or vein or umbilical panvasculitis, all vessels), and stage 3 (advanced; [subacute] necrotizing funisitis or concentric umbilical perivasculitis). Each of these histological observations was also graded as mild, moderate, or severe.
Data are presented as mean ± SEM. Data were compared between groups by one-way ANOVA. If normality could not be obtained by data transformation, ANOVA on ranks was used. Two-way ANOVA was used to compare ureaplasma colonization titers and MBA variant numbers between ureaplasma groups, as well as differential chorioamnion inflammatory cell counts. Relationships between numbers of MBA variants and indices of the extent of ureaplasma colonization were examined using linear regression.
Pregnancy losses were two from the serovar 3 low-dose ewes, one from the serovar 3 high-dose ewes, and two from the media control group. The U. parvum did not increase pregnancy loss. Fetal weight, cord blood PCO2 and pH values, and cord blood total and differential white cell counts were not different between control and the U. parvum groups (Table 1). Values of umbilical arterial PO2 and neutrophil counts were variable between ureaplasma groups, but all tended to be higher than control. When data from all ureaplasma groups were combined and compared to control by t-test, PO2 (P = 0.027) and neutrophil counts (P = 0.002) were higher in the ureaplasma-exposed animals than in controls.
Ureaplasmas were cultured from the amniotic fluid, chorioamnion, fetal lung tissue, and bronchoalveolar lavage fluid of all animals (100%) that received intra-amniotic injection of ureaplasmas and variably from other tissues (Table 2). The gut and the lung are contiguous with the amniotic fluid and were consistently colonized with ureaplasmas. The pattern of colonization in other fetal organs indicated that a few of the fetuses had systemic colonization: the CSF of two of six fetuses from the serovar 3 low-dose group and of one of eight fetuses in the serovar 6 low-dose group tested ureaplasma positive. Ureaplasmas were isolated from the kidneys and livers of animals that were exposed to serovar 3 high dose (1/7 and 2/7, respectively) and serovar 6 high dose (1/8 and 1/8, respectively) ewes.
High ureaplasma titers were detected in the amniotic fluid (>105 CFU/ml) and chorioamniotic membranes (>108 CFU/g) of ureaplasma-infected animals (Fig. 1). In the fetal lung fluid there were higher ureaplasma CFU counts in the high-dose serovar 3 compared to the low-dose serovar 6 fetuses (P < 0.05). There were no differences in ureaplasma colony-forming unit counts in the amniotic fluid, chorioamnion, or fetal lung tissue between low- or high-dose U. parvum groups for either serovar.
No bacterial species other than ureaplasmas were detected after primary subculture. However, after enrichment in thioglycollate broth, four bacterial genera, including Staphylococcus, Streptococcus, Propionibacterium, and Actinomyces, were detected in eight animals from the following groups: 1/6 controls, 1/7 serovar 3 low-dose, 2/7 serovar 3 high-dose, and 2/8 for both the serovar 6 low-dose and high-dose groups. Culture of these known skin microorganisms only after prior enrichment is consistent with skin contamination during the surgical delivery or during specimen collection and handling.
Abnormalities associated with infection/inflammation of the fetus and membranes were evident at delivery for many of the fetuses exposed to U. parvum (Table 3). Interestingly, the three fetuses with positive ureaplasma CSF cultures and >50% of all fetuses exposed to serovar 3 (either low or high dose) had a normal appearance at the time of delivery.
The chorioamniotic membranes from 18/29 (62%) animals exposed to ureaplasmas (serovars 3 and 6, high and low dose) showed evidence of maternal intermediate stage (M2) and fetal early stage (F1) inflammatory responses of a mild-to-moderate grade (Fig. 2, A–C, and Table 3). Surprisingly, the chorioamnion from 3/29 (10%) animals injected with ureaplasmas showed little or no maternal or fetal inflammatory responses, despite recovery of high titers of ureaplasmas (serovar 3 low dose, n = 1; serovar 6 low dose, n = 2; Fig. 2, D and E, and Table 3). The chorioamnion of one control animal demonstrated an M2 and F1 inflammatory response even though no ureaplasmas or other bacterial species could be cultured from these membranes. Chorioamnion inflammatory cell counts were higher in all ureaplasma groups than in controls, principally as a result of increases in monocyte numbers (Fig. 3). Monocyte counts were higher (P < 0.05) than lymphocyte or neutrophil counts in all ureaplasma groups but not in controls. Monocyte counts for all ureaplasma groups were higher than control (P < 0.05); counts in the low-dose serovar 3 group were higher than in all other groups (P < 0.05). Lymphocyte and neutrophil counts were not different between groups. Total and differential inflammatory cell counts were not different between serovar 3- and serovar 6-inoculated groups or for groups exposed to low- or high-dose ureaplasmas. Ureaplasmas caused visible abnormalities of the membrane and cord, and histological chorioamnionitis that ranged from no abnormalities to severe involvement, but these changes were not related to the ureaplasma titer, the serovar, or the initial dose of U. parvum.
Upstream mba PCR of all serovar 3 and serovar 6 isolates (P1, C1, and C2) produced an amplicon of 534 bp (results not shown), indicating that the size of this region of the mba is conserved. By contrast, PCR assays of the downstream mba produced amplicons of 890 bp and 900 bp for the original inoculums of serovar 6 and serovar 3, respectively. The size of the downstream amplicons for the P1 amniotic fluid isolates had large size variations ranging from 300 to 1120 bp (Fig. 4, A and B). P1 amplicons demonstrated either a single PCR band (E17, E21, E23 E25, E35, and E15); numerous bands, indicating that more than one mba size variant was produced in vivo within the amniotic fluid (E22, E29, E31, E36, E11, E12, and E13); or no downstream amplicon was detected (remainder of samples), which may indicate that the MBA surface antigen was not produced or that no amplicon could be detected due to multiple priming within the mba gene, despite optimization of this downstream assay (Fig. 4, A and B).
Amplicons from C1 cultures (from the six amniotic fluid isolates; 36 cultures in total) varied in size from 550 to 1115 bp (Fig. 4C); a total of 10 C1 size variants were detected (550, 680, 690, 750, 800, 890, 900, 920, 1110, and 1115 bp products). Mba PCR assays of the serovar 6 amniotic fluid C2 cultures demonstrated even further variation in the size of the mba generated in vivo: a total of 16 different downstream amplicons and size variants (320, 450, 500, 550, 600, 670, 690, 700, 750, 800, 890, 900, 950, 1000, 1115, and 1120 bp) were detected (Fig. 5, Table 4). Figure 5, A and B, demonstrates the mba size variation of two selected C2 isolates: E22 (nine C2 size variants detected) and E24 (four C2 size variants detected), respectively. The number of mba size variants detected were summarized for each specimen (C1 and C2) and are shown in Table 4. In some animals, greater total numbers of mba variants were elaborated in vivo (E22, 14 variants; E31, nine variants). By contrast, in other animals with fewer mba variants (E18, E24, E35, and E36: ≤5 total variants; Table 4), only two to three variants in addition to the antigens present in the original inoculum were generated in vivo. Of the C2 isolates, 9.3% produced no PCR amplicons for the downstream repeat region.
The number of mba variants within the inoculum culture (306S) remained constant during subculture and cloning: C2 isolates of 306S still only produced two size variants (890 and 900 bp; Fig. 5C and Table 4). However, after long-term infection/colonization (in amniotic fluid), the U. parvum serovar 6 inoculum elaborated 16 size variants (in cultures tested from six animals). The number of mba variants generated within these six animals was not related to the ureaplasma dose administered to the animals.
The P1 amniotic fluid ureaplasma cultures from each animal demonstrated size variation in the MBA proteins. The inoculum cultures of U. parvum serovar 3 (442S) and serovar 6 (306S) had one antigenic band (50 kDa) and two antigenic bands (50 and 60 kDa), respectively, which correlated directly with the 890-bp (442S) and 890- and 900-bp (306S) PCR amplicon variants (Fig. 6, A and B, serovar 6; Fig. 6, C and D, serovar 3). When Western blots of the P1 serovar 6 amniotic fluid cultures were compared, more MBA size variants (more individual antigenic bands, Fig. 6, A and B) were observed for serovar 6 than for amniotic fluid P1 cultures from serovar 3 animals (Fig. 6, C and D; P = 0.005). Amniotic fluid cultured from low-dose serovar 3 group animals contained only one or two MBA size variants per animal (Fig. 6C). However, the number of MBA size variants was not different between low- and high-dose groups (P = 0.7).
Western blots of ureaplasmas (C1 and C2) from two animals are illustrated in Figure 7. From amniotic fluid sample E24, the ureaplasma isolates produced mba amplicons of 890 and 1115 bp (Fig. 5B) at the C1 stage, which corresponded to the 60-kDa and 100-kDa bands on the Western blot (Fig. 7A). From another animal (E22), ureaplasma isolates produced a number of antigens that could be clearly delineated on the C2 Western blot (Fig. 7E): antigens of approximately 25, 32, 40, 45, 50, and 60 kDa, which corresponded to the 500, 550, 750, 890, 900, and 1114 bp C1 and C2 PCR amplicons, respectively.
Unique multiple-banded antigenic variation was observed for P1 isolates cultured from each animal (except for two animals exposed to serovar 3 low-dose E12 and E13, which produced identical antigenic profiles; Fig. 6C), and the variation of the antigen correlated directly with the size variation in the downstream mba PCR amplicons. The extent of MBA/mba variation was serovar dependent.
Ureaplasma titers in amniotic fluid were inversely related to the number of MBA variants identified in ureaplasma P1 cultures (Fig. 6E; P = 0.025). There were not significant associations between ureaplasma titers in chorioamnion, fetal lung tissue, or lung fluid samples and the number of MBA variants detected in amniotic fluid.
The variation in chorioamnion histology was compared with the extent of variation of the downstream mba of the six serovar 6 ureaplasma isolates from amniotic fluid (those that were cloned and filtered). Low numbers of mba size variants (≤5) within amniotic fluid were associated with severe inflammation, loss of tissue architecture, fibrosis, and scarring (Fig. 2A, E18; Fig. 2B, E24; Fig. 2C, E36), whereas the chorioamnion from animals with ≥9 mba variants was histologically indistinguishable from uninfected controls (Fig. 2D, E22) or showed very mild inflammation (Fig. 2E, E31).
In this study we established chronic, 70-day intra-amniotic U. parvum serovar 3 or serovar 6 infections in sheep. Five pregnancy losses occurred, but the majority of the pregnancies continued until the pregnancy was terminated by preterm surgical delivery at 125 days. The majority of human pregnancies with amniotic fluid positive for ureaplasmas also continue and deliver at >37 wk gestation [7, 8]. Ureaplasma species are the most prevalent potentially pathogenic colonizing bacteria within the male and female genital tracts and are also isolated from asymptomatic infections of amniotic fluid in the second trimester of pregnancy . The ureaplasmas can colonize/infect the amniotic fluid in the absence of labor or membrane rupture and can elicit an inflammatory response that persists in the amniotic fluid of women for as long as 2 mo without clinical signs or symptoms of amnionitis [7, 8, 36]. Ureaplasmas are also isolated from the amniotic fluid and chorioamnion of asymptomatic women who deliver at term [4, 6–8, 37, 38].
In this study, pregnant ewes were inoculated intra-amniotically with U. parvum serovar 3 or serovar 6, the serovars that are the most prevalent serovars detected in men, women, pregnant women and infertile couples [10–14], or with media control. Serovar 3 has previously been detected more frequently in women with spontaneous abortion (55%)  and adverse pregnancy outcomes (55.5%)  when compared to serovar 6 (25% and 19%, respectively). It is also interesting that U. parvum serovar 6 was isolated more frequently from women with normal pregnancies (35.5%) compared to those women who experienced adverse pregnancy outcomes (19%) . Novy et al.  recently proposed that the U. parvum serovar 1 that was used in a rhesus monkey animal model was more virulent than U. parvum serovars 3 and 6. However, in our current study, regardless of the U. parvum dose or serovar, the ureaplasmas persistently colonized the amniotic fluid, the chorioamniotic membranes, and the fetus (fetal lung fluid and lung tissue) of all animals. Both serovar 3 and serovar 6 produced histological changes consistent in severity to those reported by serovar 1 in vivo .
We now have also observed the systemic spread of ureaplasmas to fetal tissues, including the placenta, umbilical cord, and fetal liver, spleen, and gut, independent of dose or serovar. However, infection/colonization of maternal tissues (the liver and the kidney) only occurred in ewes that were injected with high doses of either U. parvum serovar 3 or serovar 6. Although reported rarely, systemic ureaplasma infections in preterm human neonates can be serious, with postmortem detection reported in the lungs [40–43], rectum, spleen, kidney, peritoneal fluid, gall bladder , liver [40, 43], and in the CSF and brain in association with intra-ventricular hemorrhage . The extent of these systemic infections and the reports of the timing of death of the neonates were consistent with infection acquired in utero, and this could be a sequelae of chronic intrauterine colonization. There have been other reports of the isolation of ureaplasmas from the CSF or brain tissue of preterm infants, and many of these preterm infants also presented with intraventricluar hemorrhage, hydrocephalus, or moderate ventriculomegaly [45–50]. However, in these studies it is not clear if the ureaplasmas were acquired in utero or by vertical transmission at birth.
With monoclonal antibodies, Zheng et al.  serotyped clinically significant neonatal isolates of Ureaplasma spp. (10 from CSF and three from blood cultures). Of these isolates, eight were identified (with the antisera used) as U. parvum serovars 1, 3, or 6 and U. urealyticum serovars 8 or 10. Since this represents five of the 14 Ureaplasma spp. serovars, it was concluded that no single serovar was associated with invasive infections. Our results further support that conclusion because we have detected both U. parvum serovar 3 and serovar 6 in the CSF of sheep fetuses after intra-amniotic ureaplasma injection, and U. parvum serovar 1 has now also been detected in the CSF of rhesus monkey fetuses . Since U. parvum is more prevalent in both men and women  than U. urealyticum, it is likely that this species will be isolated more frequently from invasive infections.
Even though there were no significant differences in the numbers of ureaplasmas isolated per gram of chorioamnion for the experimental groups (all >108 CFU/g chorioamnion), there were large differences in the histopathology. The majority of chorioamnion samples (18/29 [62%]) demonstrated M2 or F1 chorioamnionitis of mild-to-moderate grade. Four of these 18 samples also showed evidence of chronic infection, and seven demonstrated fibrosis, loss of the normal tissue architecture, and subsequent scarring.
Perhaps our most surprising observation is that three of the chorioamnion samples (one exposed to serovar 3 low-dose inoculum and two exposed to serovar 6 low-dose inoculates) showed little or no evidence of pathology despite all being heavily infected with ureaplasmas. A further 8/29 (28%) tissues demonstrated only early chorioamnionitis of a mild grade despite exposure to ureaplasmas for 10 wk. These differences in chorioamnion inflammation may explain why not all women infected in utero with Ureaplasma species during pregnancy experience adverse pregnancy outcomes [6–8].
Reyes et al.  recently investigated a single strain of U. parvum that was inoculated into the bladder of inbred F344 female rats, resulting in asymptomatic urinary tract colonization or urinary tract infection (UTI) with complications. Similar to our findings, they found that the development of UTI with complications was not dose dependent, but rather the severity of infection was dependent on host-specific factors: two distinct innate immune profiles were identified, and these correlated with the outcomes. However, they did not correlate outcomes with the MBA/mba variation of the ureaplasmas.
Shimizu et al.  demonstrated that U. parvum lipoproteins, predominantly the MBA, were responsible for the activation of the proinflammatory transcription factor nuclear factor-κB (NFKB1) through toll-like receptor (TLR) 1, TLR2, and TLR6-dependent pathways, and they concluded that the MBA is therefore likely to be the major virulence factor. The adaptive host immune responses are activated by host cell receptors, such as TLRs, which initiate cellular signaling after recognizing microbial structures also known as pathogen-associated molecular patterns (PAMPs) . Ureaplasma infection of amniotic fluid is associated with increased levels of interleukin 6 (IL6), tumor necrosis factor (TNF), interleukin 1beta (IL1B), and interleukin 8 (IL8) in women and in experimental animals, demonstrating the capacity of these microorganisms to elicit an inflammatory response [25, 39, 53]. Ureaplasmas also stimulate TNF and IL6 in vitro in human monocytic cells and rat alveolar macrophages  and nitric oxide, inducible nitric oxide synthase, and NFKB1 in rat alveolar macrophages, which may contribute to the inflammation in chronic lung disease .
One effective microbial strategy for avoiding host recognition is the modification/variation of PAMPs . Zheng et al.  investigated four ureaplasma isolates (each from a single colony-forming unit) obtained from a single clinical specimen and demonstrated size variation of the mba after PCR amplification of the entire gene. They hypothesized that size variation of the MBA, the surface-exposed lipoprotein, may be associated with immune evasion and may play a significant role in the in vivo survival strategy of ureaplasmas . Monecke et al.  also demonstrated ureaplasma antigenic variation but only after in vitro selection. Our current study extends this work by demonstrating both genetic and protein variation of the surface-exposed lipoprotein, the mba/MBA of ureaplasma isolates generated in vivo. Our assessments of mba variation using PCR and Western blots of all P1 U. parvum serovar 3 and serovar 6 isolates from sheep inoculated intra-amniotically confirmed that the size variability of the mba and its expressed protein (MBA) can be attributed to the size variation of the downstream repeat region of the gene. The number of mba variants detected by PCR assay corresponded directly with the number of antigenic bands detected by Western blot. By comparison, the upstream region of the mba was conserved in size for all serovar 3 and serovar 6 isolates. We further investigated variation of the initial serovar 6 inoculum and isolates obtained from six animals injected with high- or low-dose U. parvum serovar 6. In the six animals investigated, 22 different downstream mba size variants (Table 4) were elaborated in vivo from the initial ureaplasma inoculum that contained only two size variants (even after subculture and cloning in vitro). Since the MBA is a major virulence factor and stimulates an immune response , we propose that this same antigen is a PAMP and that the variation we have observed in vivo is a mechanism of immune evasion.
Our study has shown that MBA antigenic/size variation is associated with differences in host responses and the severity of amniotic fluid infection. Ureaplasmas were cultured from amniotic fluid from ureaplasma-exposed animals with chorioamnion that displayed no visible histological inflammation, and these isolates demonstrated highly size-variable downstream mba regions and numerous mba size variants (≥9). In contrast, chorioamnion specimens demonstrating severe inflammation, gross morphological changes, and profound loss of tissue architecture had low numbers (≤5) of amniotic fluid mba variants (Fig. 2). These observations support the hypothesis of immune evasion by variation of PAMPs: ureaplasmas may be able to chronically colonize amniotic fluid without inducing histologic inflammation, avoiding recognition by host immune cells by varying the downstream mba repeat region and generating many MBA variants. It is also interesting that for fetuses that had positive CSF ureaplasma cultures (indicative of the greatest systemic spread of infection), all were injected with low-dose ureaplasmas (serovar 6 low dose, E29; serovar 3 low dose, E13 and E16), and these isolates demonstrated only one or two amniotic fluid MBA variants (Fig. 6). Similarly, within the chorioamnion the highest monocytic cell counts were also observed in animals exposed to low-dose serovar 3 (P < 0.05). Other studies have also demonstrated that antigenic variation as a result of changes in surface antigen genes and expressed antigens is an important mechanism for organisms that cause long-term or repeated infections. This mechanism has been shown to contribute to avoidance of adaptive immune responses, to tissue tropism, and/or to the pathogenesis processes for Borrelia burgdorferi  and for Mycoplasma pulmonis .
A technical limitation of our work involves the potential for selection bias when selecting microscopic ureaplasma colonies to be excised for subsequent cloning and filtration. We cannot be sure that the colonies excised were representative of the entire amniotic fluid population of ureaplasmas. However, due to the difficulty in selecting isolated colonies and the large amount of work required to clone and filter ureaplasmas, it is likely that this limitation will always exist. No other study has analyzed this number of ureaplasma isolates.
In two prior studies, ureaplasmas were detected by PCR in the amniotic fluid of 12% of 433 asymptomatic women undergoing screening amniocentesis at 15–19 wk of gestation [7, 8]. Though detection of ureaplasmas in these women was significantly associated with preterm premature rupture of membranes  and preterm labor , the majority of these women delivered at term, and the presence of ureaplasmas would not have been suspected in these women but for the PCR testing of the amniotic fluid. Our results have shown that high-titer ureaplasma infections of the amniotic fluid can produce no obvious macroscopic or histological signs of infection. Furthermore, we have identified the complexity of interactions between Ureaplasma species and the host, demonstrating that the number of MBA/mba variants is associated with differences in the extent of systemic infection within the fetuses and in the histopathology of infected tissues.
The authors would like to acknowledge JRL Hall & Co., in particular Sara Ritchie and Fiona Hall, who have been responsible for breeding and supplying us with the high-quality research animals necessary for this project. Their input continues to be invaluable. We also wish to thank Drs. John and Janet Allan at Wesley Monash IVF for the research that has provided low passage clinical ureaplasma isolates and Emeritus Dr. Patricia Quinn for her generous provision of anti-ureaplasma rabbit antisera.
1Supported by the National Institutes of Health (HL-65397) and the National Health and Medical Research Council of Australia (303261 & 458577).
3These authors contributed equally to this work.