Immunity induced by a broad class of inorganic crystalline materials is directly controlled by their chemistry

Abstract

There is currently no paradigm in immunology that enables an accurate prediction of how the immune system will respond to any given agent. Here we show that the immunological responses induced by members of a broad class of inorganic crystalline materials are controlled purely by their physicochemical properties in a highly predictable manner. We show that structurally and chemically homogeneous layered double hydroxides (LDHs) can elicit diverse human dendritic cell responses in vitro. Using a systems vaccinology approach, we find that every measured response can be modeled using a subset of just three physical and chemical properties for all compounds tested. This correlation can be reduced to a simple linear equation that enables the immunological responses stimulated by newly synthesized LDHs to be predicted in advance from these three parameters alone. We also show that mouse antigen-specific antibody responses in vivo and human macrophage responses in vitro are controlled by the same properties, suggesting they may control diverse responses at both individual component and global levels of immunity. This study demonstrates that immunity can be determined purely by chemistry and opens the possibility of rational manipulation of immunity for therapeutic purposes.

Figure 1.

Illustrations of typical LDHs and the systems vaccinology approach used in this study. (a–e) Transmission electron micrographs of LiAl₂-CO₃ (a), Mg₂Al-NO₃ (b), Mg₂Fe-Cl (c), Imject alum (d), and Alhydrogel (e). Size data on the LDHs are in Table S1. (f) A schematic showing the systems vaccinology approach. To the left, the general LDH structure is depicted, showing the positively charged layers (yellow/blue/red circles) and interlayer anions (green circles), with a surrounding layer of water (top and bottom). The in vitro DCs and in vivo antibody responses stimulated by a series of LDHs were evaluated, and the datasets were then independently subjected to multivariate analysis, with the physicochemical properties of LDHs detailed in Table S4. All observed responses were highly correlated with the three key physicochemical properties indicated on the left side and conformed to the equation (Eq. 1) illustrated on the right side. This equation was then used to predict, de novo, the immunological (DC) responses stimulated by newly synthesized LDHs from their respective physicochemical properties.

Figure 2.

Chemically different LDHs drive diverse DC responses in vitro. Human monocyte-derived DCs were cultured without or with the indicated LDH or commercial adjuvant preparation for a period of 24 h. The concentration of cytokines and chemokines in the supernatant was then determined using ELISA (IL-6 and TNF) or Luminex (others). Cell surface expression of co-stimulatory and co-inhibitory molecules was assessed by flow cytometry. MFI, mean fluorescence intensity. Error bars show one standard error. *, P < 0.05; and **, P < 0.01 versus cells alone. The number of individual experiments performed for each response ranges from n = 6 to n = 22 and is given in Table S3. Each experiment contained at least three biological replicates.

Figure 3.

Multiple DC responses induced by newly synthesized LDHs can be predicted with a high degree of accuracy. (a) DC responses to LiAl₂-NO₃ and Mg₂Al-Cl were assessed as in Fig. 2. Error bars show one standard error. **, P < 0.01 versus cells alone. Four independent experiments were performed, each with three or four biological replicates. (b and c) DC responses to LiAl₂-NO₃ and Mg₂Al-Cl were predicted with Eq. 1 following calibration of the model using data from Fig. 2. In b, the mean and 95% CIs for the measured responses are indicated (diamonds and short horizontal lines) with the predicted value (triangles) immediately below. In c, observed ln responses are shown along a straight line of gradient 1, and the predicted ln responses as squares on the same plot.

Figure 4.

Chemically different LDHs have distinct adjuvant activities in vivo. Mice were immunized i.p. with 10 µg OVA admixed with 1 mg LDH, Alhydrogel, or Imject alum and boosted i.p. 10 d later with 10 µg OVA alone. Control mice were given the booster treatment only. (a) 7 d after boost, mice were bled via the tail vein, and serum levels of OVA-specific antibody isotypes were measured by ELISA. (b) Starting 1 wk later, the prime-boosted mice were challenged on three consecutive days with a 1% OVA aerosol. 1 d after the final challenge, ELISA was used to measure serum levels of OVA-specific antibody isotypes. Antibody titers are shown as OD450, as assessed by ELISA. Error bars depict one standard error. **, P < 0.01 via pairwise comparisons versus OVA alone. Data are from two independent experiments, each with at least five mice per group.