Lipid analogs reveal features critical for hemolysis and diminish granadaene mediated Group B Streptococcus infection

Abstract

Although certain microbial lipids are toxins, the structural features important for cytotoxicity remain unknown. Increased functional understanding is essential for developing therapeutics against toxic microbial lipids. Group B Streptococci (GBS) are bacteria associated with preterm births, stillbirths, and severe infections in neonates and adults. GBS produce a pigmented, cytotoxic lipid, known as granadaene. Despite its importance to all manifestations of GBS disease, studies towards understanding granadaene's toxic activity are hindered by its instability and insolubility in purified form. Here, we report the synthesis and screening of lipid derivatives inspired by granadaene, which reveal features central to toxin function, namely the polyene chain length. Furthermore, we show that vaccination with a non-toxic synthetic analog confers the production of antibodies that inhibit granadaene-mediated hemolysis ex vivo and diminish GBS infection in vivo. This work provides unique structural and functional insight into granadaene and a strategy to mitigate GBS infection, which will be relevant to other toxic lipids encoded by human pathogens.

Fig. 1. The polyene chain length is important for hemolytic activity.

a Granadaene contains a terminal rhamnose, a polyene chain consisting of 12 double bonds, and a terminal ornithine. b–e Synthetic analogs of granadaene (structures shown) were resuspended in DTS (DMSO + trifluoroacetic acid (TFA, 0.1%) + starch (20%)) or DMSO. The analogs were co-incubated with human red blood cells at 250, 125, and 62.5 µM for 1 h at 37 °C, and hemoglobin release in cell supernatants was quantified to determine percent hemolysis relative to Triton X-100 (0.1%)-treated positive controls and PBS-treated negative controls. Mean and standard error from three independent experiments are shown. Differences in hemolysis between DTS resuspension and DMSO resuspension for a given concentration of each compound were analyzed using a two-way ANOVA with Tukey’s post test. pP7X: 250 μM, p = 0.9959, 125 μM, p = 0.2079, 62.5 μM, p = 0.9007; pP7: 250 μM, p = 0.9524, 125 μM, p = 0.8823, 62.5 μM, p = 0.8375; pP9: 250 μM, p = 0.9999, 125 μM, p = 0.5592, 62.5 μM, p > 0.9999; P9: 250 μM, p < 0.0001, 125 μM, p = 0.6957, 62.5 μM, p > 0.9999. Additionally, 5 µL of each synthetic analog at 0.02 M concentration was spotted onto red blood agar plates and allowed to incubate at 37 °C overnight. The hemolytic effect of each compound on red blood agar is shown. ****, p < 0.0001. n.s., Not significant.

Fig. 2. Proposed role of polyenes and polar head groups for granadaene-mediated hemolysis and cytolysis.

The polyene chain of granadaene spans the length of the target cell’s lipid bilayer and the two polar head groups, rhamnose and ornithine, allow for stable insertion of the GBS ornithine rhamnopolyene into the cell membrane, leading to membrane disruption and hemolysis or cytolysis. Shorter polyenes such as those containing seven double bonds are hemolytic only when one polar end is replaced with a hydrophobic group (e.g., TBS), as in pP7, in which the molecule stably inserts in the membrane due to the compatibility between the hydrophobic lipid bilayer and the TBS group. However, if both polar ends are maintained in compounds with shorter polyenes, as in P7 or R-P4, no hemolytic activity is observed because stable insertion into the cell membrane is not favored. Once the polyene chain reaches a sufficient length, as in P9, some hemolytic activity is observed even when both polar head groups are maintained. Together, this suggests that the length of the polyene chain and polar head groups found in granadaene are key to stable membrane disruption of host cells.

Fig. 3. Granadaene is cytolytic to T and B cells.

a Primary human CD4⁺ T and B cells were incubated with either hyper-hemolytic GBS (HH GBS) or non-hemolytic GBS (NH GBS) at an MOI of 10 for 1 h at 37 °C, and cytotoxicity was measured by the release of lactate dehydrogenase (LDH) into the cell supernatant relative to 100% lysis (Triton X-100 (0.1%)) and 0% lysis (PBS) controls. Mean and standard error from three independent experiments are shown. A two-tailed unpaired t test was used to compare groups. CD4⁺ T cells: p = 0.0075, B cells: p = 0.0005. b PI uptake and Annexin V staining were measured using flow cytometry following incubation with HH GBS at an MOI of 10 at the indicated time points. c CD4⁺ T cells were imaged using scanning electron microscopy (SEM) following incubation with PBS, NH GBS, or HH GBS (MOI = 10) for 1 h. Images are representative of three experiments. Scale bars are 1 μm. d Primary human CD4⁺ T and B cells were incubated with purified granadaene (0.5 µM), equivalent amount of extract from NH GBS (GBSΔcylE), or solvent (DTS), and cell death was measured by LDH release in supernatants, as above. Mean and standard error from three independent experiments are shown. Groups were compared with one-way ANOVA with Tukey’s post test. CD4⁺ T cells: granadaene vs. ΔcylE extract: p = 0.0002, granadaene vs. DTS: 0.0002, ΔcylE extract vs. DTS: p = 0.9928. B cells: granadaene vs. ΔcylE extract: p = 0.0020, granadaene vs. DTS: 0.0032, ΔcylE extract vs. DTS: p = 0.8513. e PI uptake and Annexin V staining were measured using flow cytometry in each cell type following incubation with purified granadaene (0.5 µM) at the indicated time points. f CD4⁺ T cells were imaged using SEM following incubation with DTS, GBSΔcylE extract, or granadaene (0.5 µM) for 1 h. Images are representative of three experiments. Scale bars are 1 μm. **, p < 0.01. ***, p < 0.001. ****, p < 0.0001. n.s., Not significant.

Fig. 4. R-P4 is non-toxic to T and B cells.

a Primary human CD4⁺ T cells (left) and B cells (right) were incubated with R-P4 (20 μM) or granadaene (0.5 μM) for 1 h at 37 °C, and cytotoxicity was measured by LDH release into cell supernatant relative to 100% lysis (Triton X-100 (0.1%)) and 0% lysis (PBS) controls. Mean and standard error from three independent experiments are shown. Groups were compared with a two-tailed unpaired t test. CD4⁺ T cells: p = 0.0024, B cells: p = 0.0084. b Primary human CD4⁺ T cells were treated with PMA (10 ng/mL) and anti-CD3ε (100 μg/mL) (stimulated) or media alone (unstimulated). Then, cells were treated with either PBS or R-P4 (20 μM) and incubated at 37 °C for 48 h. Cells were stained for CD69, resuspended in DAPI (0.5 μM), and analyzed by flow cytometry. Percent CD69⁺ of DAPI− cells from three independent experiments are represented with mean and standard error. Treatment groups were compared using one-way ANOVA with Tukey’s post test. PBS (unstimulated) vs. PBS (stimulated): p < 0.0001, PBS (unstimulated) vs. R-P4 (stimulated): p < 0.0001, PBS (stimulated) vs. R-P4 (stimulated): p = 0.8151. c Primary human B cells were treated with IL-4 (20 ng/mL) and anti-CD40 (5 μg/mL) (stimulated) or media alone (unstimulated). Then, cells were treated with either PBS or R-P4 (20 μM) and incubated at 37 °C for 48 h. Cells were stained for CD69, resuspended in DAPI (0.5 μM), and analyzed by flow cytometry. Percent CD69⁺ of DAPI− cells from three independent experiments are represented with mean and standard error. Treatment groups were compared using one-way ANOVA with Tukey’s post test. PBS (unstimulated) vs. PBS (stimulated): p = 0.0091, PBS (unstimulated) vs. R-P4 (stimulated): p = 0.0453, PBS (stimulated) vs. R-P4 (stimulated): p = 0.3893. *, p < 0.05. **, p < 0.01. ****, p < 0.0001. n.s., Not significant.

Fig. 5. Vaccination with a non-toxic synthetic analog diminished GBS infection.

a Mice were vaccinated with an emulsion of R-P4 (10 μM in PBS) in complete Freund’s Adjuvant and boosted 14 days after initial vaccination using R-P4 in incomplete Freund’s Adjuvant. At 21 days, vaccinated mice were euthanized for blood/plasma collection or were challenged (IP) with ∼1 × 10⁸ CFU of hyper-hemolytic GBS strain NCTC10/84. As controls, mice receiving adjuvant only were tested in parallel with the same schedule. b Approximately 4 μL solvent or purified granadaene (25 μM) was spotted on PVDF membranes, which were then blocked and probed with plasma from analog-vaccinated or adjuvant-only mice (n = 10/group). Subsequently, the membranes were probed with secondary goat anti-mouse IgG (Alexa Fluor 680) and analyzed on the Odyssey LI-COR Imaging System. Four representative spots from each group are shown. Signal intensity of each spot was determined using the Image J software, and a two-tailed unpaired t test was used to compare signal intensity between groups (p = 0.0020). Mean and standard error are shown. c Plasma (diluted 1:1000) from analog-vaccinated mice or adjuvant-only only mice (n = 10/group) was pre-incubated with purified granadaene (0.3 μM) for 1 h and then incubated with human red blood cells for 1 h. Percent inhibition of granadaene hemolysis was calculated (see Methods), and a two-tailed unpaired t test was used to compare hemolysis inhibition by R-P4 plasma vs. control plasma (p = 0.0004). Mean and standard error are shown. d GBS CFU recovered from the blood, spleen, lung, and brain of mice (n = 24) at 24 h post infection were compared using a two-tailed Mann–Whitney test. Blood: p = 0.0025, spleen: p = 0.0001, lung: p = 0.0010, brain: p = 0.0002. *, p < 0.05. **, p < 0.01. ***, p < 0.001.