Bacterial lipoproteins can disseminate from the periphery to inflame the brain

Abstract

The current view is that bacteria need to enter the brain to cause inflammation. However, in mice infected with the spirochete Borrelia turicatae, we observed widespread cerebral inflammation despite a paucity of spirochetes in the brain parenchyma at times of high bacteremia. Here we studied the possibility that bacterial lipoproteins may be capable of disseminating from the periphery across the blood-brain barrier to inflame the brain. For this we injected normal and infected mice intraperitoneally with lanthanide-labeled variable outer membrane lipoproteins of B. turicatae and measured their localization in blood, various peripheral organs, and whole and capillary-depleted brain protein extracts at various times. Lanthanide-labeled nonlipidated lipoproteins of B. turicatae and mouse albumin were used as controls. Brain inflammation was measured by TaqMan RT-PCR amplification of genes known to be up-regulated in response to borrelial infection. The results showed that the two lipoproteins we studied, LVsp1 and LVsp2, were capable of inflaming the brain after intraperitoneal injection to different degrees: LVsp1 was better than LVsp2 and Bt1 spirochetes at moving from blood to brain. The dissemination of LVsp1 from the periphery to the brain occurred under normal conditions and significantly increased with infection. In contrast, LVsp2 disseminated better to peripheral organs. We conclude that some bacterial lipoproteins can disseminate from the periphery to inflame the brain.

Figure 1

Dissemination of Eu-labeled proteins. A and B: Dissemination kinetics from peritoneum to plasma and urine 2, 4, 12, and 18 hours after i.p. injection of 10⁷ counts per second of each Eu-labeled protein (n = 5 mice each). C–F: Dissemination kinetics from peritoneum to tissues 2, 4, 12, and 18 hours after i.p. injection with 100 μg/kg of each Eu-labeled protein (n = 5 mice each). The bars represent mean (SD) of two separate experiments. P values were calculated by two-way analysis of variance and show the tissue distribution for each Eu-labeled protein. Notice that LVsp1 shows preferential dissemination from the peritoneum to the brain and lung.

Figure 2

Protein dissemination from blood to tissues: brain (A), lung (B), kidney (C), and liver (D). Groups of five mice each were necropsied 2, 4, 12, or 18 hours after i.p. injection with 100 μg/kg of each Eu-labeled protein. The bars represent mean (SD) of two separate experiments. P values were calculated by two-way analysis of variance and show the distribution from plasma to tissue over time for each Eu-labeled protein. Notice that LVsp1 shows preferential distribution from plasma to the brain.

Figure 3

Effect of infection on the dissemination of LVsp1 from the periphery to the brain. Dissemination ratios of peritoneum to the brain (A) and plasma to the brain (B) 18 hours after i.p. injection with 100 μg/kg of each Eu-labeled protein into uninfected C3H/HeJ mice or relapsing (C3H/HeJ) or persistently (C57BL/6-RAG1 deficient) Bt1-infected mice (n = 5 each).The results are presented as box plots for two to three separate experiments for each mouse model. P values were determined by using the Kruskal-Wallis test. Notice that Eu-LVsp1 shows the greatest dissemination to the brain of all of the proteins tested. Also notice that either relapsing or persistent infection increased the dissemination of Eu-LVsp1 to the brain, whereas this occurred to a lesser extent with Eu-rVsp1 and Eu-albumin.

Figure 4

Dissemination of LVsp1 to the brain represents crossing of the blood-brain barrier. Dissemination from plasma to brain fractions 18 hours after i.p. injection with 100 μg/kg of each Eu-labeled protein in uninfected mice (A) or mice with relapsing (B) or persistent (C) Bt1-infection. Mice with relapsing infection were C3H/HeJ necropsied on day five after inoculation of Bt1, whereas mice with persistent infection were C57BL/6-RAG1 deficient necropsied 12 days after inoculation with Bt1. The results are presented as box plots for two to three separate experiments for each model. P values were calculated by using the Kruskal-Wallis test. Notice that LVsp1 is significantly increased in the brain parenchymal fraction compared with the brain capillary fraction in uninfected (A) and relapsing (B) or persistently (C) infected mice.

Figure 5

Entry of LVsp1 into the brain parenchyma causes up-regulation of brain pro-inflammatory and anti-inflammatory mediators. Correlation between brain gene expression with ratios of dissemination of Eu-labeled proteins from plasma to brain 18 hours after i.p. injection with 100 μg/kg of either Eu-LVsp1 (n = 7) or Eu-albumin (n = 5) into uninfected C3H/HeJ mice. Correlation and P values were determinate by Spearman’s test. Notice a positive correlation between the entry of Eu-LVsp1 into the brain parenchyma and up-regulation of F4/80 (A), CXCL13 (B), and IL-10 (C). This was not observed with the Eu-albumin used as control.

Figure 6

A: LVsp1 disseminates into the brain parenchyma better than Bt1. Ratios of distribution of Eu-LVsp1 and Bt1 spirochetes from blood to brain four hours after i.p. injection with 100 μg/kg of Eu-LVsp1 to C57BL/6-RAG1 deficient mice that had been infected for two weeks (n = 7). Bt1 was measured by TaqMan PCR amplification of the borrelial gene 16S rRNA. Results show box plots for two separate experiments each. P values were calculated by using Mann-Whitney tests. Notice preferential distribution of Eu-LVsp1 into the brain parenchymal fraction, whereas Bt1 spirochetes preferentially distributed into the brain capillary fraction. B: LVsp1 disseminates from the periphery into the brain better than LVsp2. Dissemination ratios of plasma to whole brain or brain fraction four hours after i.p. injection with 100 μg/kg of Sm-LVsp1 or Eu-LVsp2 into groups of uninfected mice or mice persistently infected with Bt1 (n = 6 each). Results are presented as box plots for two separate experiments for each model. Notice that LVsp1 disseminated to whole brain and brain parenchymal fraction better than LVsp2. C: LVsp2 reduces the dissemination of LVsp1 into the brain parenchyma. Dissemination ratios from plasma to whole brain or brain fraction four hours after i.p. injection with Sm-LVsp1 alone (100 μg/kg) or in combination with Eu-LVsp2 or Eu-albumin (100 μg/kg of each) into groups of uninfected mice (n = 6 each). Results are presented as box plots for two separate experiments for each model. P values were determined by using the Kruskal-Wallis test. Notice that simultaneous injection with Eu-LVsp2, but not with Eu-albumin, reduced the dissemination of Sm-LVsp1 from plasma to brain parenchyma.

Figure 7

Differences in dissemination of LVsp1 and LVsp2 to blood, urine, and tissues other than the brain. Dissemination ratios from peritoneum to plasma and urine (A) or tissues (B) and from plasma to tissues (C) 4 hours after i.p. injection with 100 μg/kg of Sm-LVsp1 or Eu-LVsp2 (n = 5 each). The results are presented as box plots of two separate experiments, and P values were calculated by using the Kruskal-Wallis test. Note the higher dissemination for LVsp2 to liver and kidney compared with LVsp1.