The role of the multiple banded antigen of Ureaplasma parvum in intra-amniotic infection: major virulence factor or decoy?

Abstract

The multiple banded antigen (MBA) is a predicted virulence factor of Ureaplasma species. Antigenic variation of the MBA is a potential mechanism by which ureaplasmas avoid immune recognition and cause chronic infections of the upper genital tract of pregnant women. We tested whether the MBA is involved in the pathogenesis of intra-amniotic infection and chorioamnionitis by injecting virulent or avirulent-derived ureaplasma clones (expressing single MBA variants) into the amniotic fluid of pregnant sheep. At 55 days of gestation pregnant ewes (n = 20) received intra-amniotic injections of virulent-derived or avirulent-derived U. parvum serovar 6 strains (2×10⁴ CFU), or 10B medium (n = 5). Amniotic fluid was collected every two weeks post-infection and fetal tissues were collected at the time of surgical delivery of the fetus (140 days of gestation). Whilst chronic colonisation was established in the amniotic fluid of animals infected with avirulent-derived and virulent-derived ureaplasmas, the severity of chorioamnionitis and fetal inflammation was not different between these groups (p>0.05). MBA size variants (32-170 kDa) were generated in vivo in amniotic fluid samples from both the avirulent and virulent groups, whereas in vitro antibody selection experiments led to the emergence of MBA-negative escape variants in both strains. Anti-ureaplasma IgG antibodies were detected in the maternal serum of animals from the avirulent (40%) and virulent (55%) groups, and these antibodies correlated with increased IL-1β, IL-6 and IL-8 expression in chorioamnion tissue (p<0.05). We demonstrate that ureaplasmas are capable of MBA phase variation in vitro; however, ureaplasmas undergo MBA size variation in vivo, to potentially prevent eradication by the immune response. Size variation of the MBA did not correlate with the severity of chorioamnionitis. Nonetheless, the correlation between a maternal humoral response and the expression of chorioamnion cytokines is a novel finding. This host response may be important in the pathogenesis of inflammation-mediated adverse pregnancy outcomes.

Figure 1. Ureaplasma colonization of amniotic fluid and fetal tissues.

(A) Chronic infection of the amniotic fluid was observed in all ewes experimentally infected with ureaplasmas from the time of inoculation (55 d) until fetuses were delivered at 140 d. Amniotic fluid ureaplasma colonization was significantly increased in the avirulent group when compared to the virulent group at 87 d (p<0.05, denoted by #). Statistically significant differences in amniotic fluid ureaplasma colonization within animal groups occurred between 87 d and 101 d; and 87 d and 126 d (p<0.05). (B) Ureaplasmas were isolated from the chorioamnion, cord and fetal lung; however, recovered ureaplasma CFU/g was not different between animal groups for the tested tissue types. * = statistically significant difference between time points in the virulent group only; ** = statistically significant difference between time points in both groups. AF = amniotic fluid; CFU = colony forming units; d = days of gestation. Data are presented as mean ± SEM.

Figure 2. Fetal inflammation induced by intra-amniotic ureaplasma infection.

Inflammatory cell infiltrates within chorioamnion tissue (A) and the severity of histological chorioamnionitis (B) were increased in animals from the virulent and avirulent groups when compared to the control group. Representative chorioamnion sections ((C) stained with haematoxylin and eosin (top row) or Masson's trichrome stain (bottom row), photographed at ×200 total magnification) demonstrate the 4 stages of histological chorioamnionitis. From left to right: Grade 1 (uninfected control), minimal inflammatory cell infiltrate and no tissue fibrosis, necrosis or abscesses; Grade 2, mild inflammatory cell infiltrate and mild tissue fibrosis, necrosis or abscesses; Grade 3, heavy inflammatory cell infiltrate and moderate tissue fibrosis, necrosis or abscesses; Grade 4, heavy inflammatory cell infiltrate and sever fibrosis, necrosis or abscesses. Stars on haematoxylin and eosin stained sections indicate localized inflammatory cell influx. Arrows on Masson's trichrome stained sections represent tissue fibrosis and disruption of normal tissue morphology. Size bars represent 50 µm. Inflammatory cell infiltrates within cord tissue (D) and fetal lung tissue (E) were not statistically different between treatment groups. Data are presented as mean + SEM. * p<0.05 when compared to the control group.

Figure 3. Size variation of the MBA was observed in amniotic fluid samples.

PCR of the repeat region of the mba from ureaplasma isolates E22 5.8.1 and E24 3.2.1 produced single amplicons prior to injection into pregnant sheep (A), indicating that these clonal isolates contained only one mba size variant; however, western blots of the same isolates (B) demonstrated that the MBA appears as a double band. MBA size variants were detected in vivo within the amniotic fluid of animals from the avirulent group (C) and the virulent group (D) by western blot. Each western blot demonstrates the ureaplasma MBA size variants generated at 73, 87, 101, 115, 126 and 140 days of gestation in each animal. Note: samples were not collected at some time points for certain animals due to oligohydramnios or other complicating factors. Protein preparations from the avirulent-derived (E22 5.8.1) and virulent-derived (E24 3.2.1) ureaplasma clones used for intra-amniotic injection were included in each western blot (in the last lane) as a positive control and for size comparison. M = DNA molecular weight marker VIII (Roche, Castle Hill, New South Wales) or protein marker (BioRad, Gladesville, New South Wales). The average number of MBA size variants generated over time (E) was not different between animal groups.

Figure 4. Demonstration of maternal and fetal anti-ureaplasma humoral responses.

Anti-ureaplasma IgG antibodies were detected in the maternal serum (A) and the fetal serum (B) of animals in both the avirulent and virulent groups. Antibodies in serum samples reacted with ≥1 ureaplasmal protein, over a wide molecular weight range (approximately 45 kDa to 80 kDa). Numbers above western blots indicate the animal number from which the serum was obtained. M = protein marker (BioRad).

Figure 5. Pro-inflammatory cytokines were up-regulated in animals which produced ureaplasma-specific ureaplasma antibodies.

The relative expression of IL-1β, IL-6 and IL-8 within the chorioamnion was significantly increased in ewes that tested positive for anti-ureaplasma IgG antibodies (IgG positive) within serum samples, when compared to animals in which these antibodies were not generated (IgG negative) (A). Conversely, the relative expression of TNF-α and IL-10 (B); and TLR1, TLR2 and TLR6 (C) within the chorioamnion were decreased in IgG negative animals when compared to IgG positive animals. Data are presented as mean fold change ± SEM. * p<0.05. Expression of genes is determined relative to the expression of GAPDH after normalisation against 10B medium control animals.

Figure 6. Ureaplasmas are phase variable in vitro.

Western blot analysis demonstrated that MBA expression was not affected in avirulent-derived (E22 5.8.1) and virulent-derived (E24 3.2.1) ureaplasmas, which were serially transferred in 10B medium without the presence of polyclonal antibodies (top panel). MBA negative escape variants were generated for both avirulent-derived and virulent-derived ureaplasma strains after serial transfer in 10B medium containing anti-ureaplasma polyclonal antibodies (α-Up, bottom panel). P = passage number.