Targeted disruption of pi-pi stacking in Malaysian banana lectin reduces mitogenicity while preserving antiviral activity

Abstract

Lectins, carbohydrate-binding proteins, have been regarded as potential antiviral agents, as some can bind glycans on viral surface glycoproteins and inactivate their functions. However, clinical development of lectins has been stalled by the mitogenicity of many of these proteins, which is the ability to stimulate deleterious proliferation, especially of immune cells. We previously demonstrated that the mitogenic and antiviral activities of a lectin (banana lectin, BanLec) can be separated via a single amino acid mutation, histidine to threonine at position 84 (H84T), within the third Greek key. The resulting lectin, H84T BanLec, is virtually non-mitogenic but retains antiviral activity. Decreased mitogenicity was associated with disruption of pi-pi stacking between two aromatic amino acids. To examine whether we could provide further proof-of-principle of the ability to separate these two distinct lectin functions, we identified another lectin, Malaysian banana lectin (Malay BanLec), with similar structural features as BanLec, including pi-pi stacking, but with only 63% amino acid identity, and showed that it is both mitogenic and potently antiviral. We then engineered an F84T mutation expected to disrupt pi-pi stacking, analogous to H84T. As predicted, F84T Malay BanLec (F84T) was less mitogenic than wild type. However, F84T maintained strong antiviral activity and inhibited replication of HIV, Ebola, and other viruses. The F84T mutation disrupted pi-pi stacking without disrupting the overall lectin structure. These findings show that pi-pi stacking in the third Greek key is a conserved mitogenic motif in these two jacalin-related lectins BanLec and Malay BanLec, and further highlight the potential to rationally engineer antiviral lectins for therapeutic purposes.

Figure 1

Malay BanLec is structurally similar to BanLec. Amino acid sequence alignment of BanLec and Malay BanLec (middle rows) from amino acids 1 (top left) to 141 (bottom right). The 6 × His-tag (LEHHHHHH) is indicated at the end of the Malay BanLec sequence. The amino acids predicted to undergo pi–pi stacking (“pi–pi”) and the predicted ligand binding loops of the carbohydrate binding sites (“ligand binding”) in Malay BanLec are outlined in black and red, respectively, aligned with those known sites in BanLec. Sites of amino acid identity (“consensus”) are indicated in the top row in the colored boxes; “x” indicates that identity is not shared at a particular site. The bottom row (“sequence logo”) depicts the sequence of both lectins, with amino acids from both sequences represented if there is not identity at a particular site. Malay BanLec is 63% identical to BanLec at the amino acid level. The figure was generated using Microsoft PowerPoint 2016 (https://www.microsoft.com/en-us/microsoft-365/powerpoint) and Adobe Illustrator, version 22.1.0 (https://www.adobe.com/products/illustrator.html).

Figure 2

Malay BanLec has both mitogenic and antiviral activity. (a) BrdU incorporation into peripheral blood lymphocytes (PBLs). PBLs from a healthy human donor were treated with a range of concentrations of WT, H84T, or D133G BanLec, or Malay BanLec for 3 d and then BrdU incorporation over 18 h was measured by chemiluminescent ELISA. RLU relative light units. (b) TZM-bl cells were pre-treated with the same lectins as in A and exposed to HIV BaL for 2 days. To assess levels of infection, luciferase activity was measured. Data in A and B are representative of one independent experiment each. Error bars represent the SEM of triplicate values. The figure was generated using GraphPad Prism 7 (https://www.graphpad.com/scientific-software/prism/) and Adobe Illustrator, version 22.1.0 (https://www.adobe.com/products/illustrator.html).

Figure 3

F84T Malay BanLec is less mitogenic than is WT Malay BanLec. (a) BrdU incorporation into PBLs. PBLs from healthy human donors were treated with a range of concentrations of WT or F84T Malay BanLec, or WT or H84T BanLec for 3 d and then BrdU incorporation over 18 h was measured by chemiluminescent ELISA. Data are representative of two independent experiments using cells from two donors, with data from one donor shown. Error bars represent the SEM of triplicate values. (b,c) Activation markers in lectin-treated peripheral blood mononuclear cells (PBMCs). PBMCs from three healthy human donors were treated with a range of concentrations of WT Malay BanLec (WT) or F84T Malay BanLec, griffithsin (GRFT), or PHA and the percentage of CD4 + CD25 + (B) or CD4 + CD69 + (C) cells assessed by flow cytometry. GRFT, known to be virtually non-mitogenic, serves as the negative control, whereas PHA serves as the positive control. Bars represent the average values from the three donors and error bars the SEM. Black asterisks indicate statistically significant differences between the means in the GRFT group and the corresponding means in the WT group. Blue asterisks indicate statistically significant differences between the mean in the PHA group and the mean in the control group. *P < 0.05, **P < 0.01, ****P < 0.0001. (d–f) Cytokine/chemokine secretion in lectin-treated PBMCs. PBMCs from multiple healthy donors were treated with GRFT (d), WT (e), or F84T Malay BanLec (f) and supernatants collected for analysis of expression of various cytokines and chemokines by BioPlex assay. Results are expressed as the percentage of PBMC donors whose stimulated PBMCs express the indicated fold increase values of cytokine/chemokine concentrations with respect to untreated PBMCs. The figure was generated using GraphPad Prism 7 (https://www.graphpad.com/scientific-software/prism/) and Adobe Illustrator, version 22.1.0 (https://www.adobe.com/products/illustrator.html).

Figure 4

F84T Malay BanLec binds to HIV gp140 as determined by surface plasmon resonance. Single cycle kinetics of WT Malay BanLec (a), F84T Malay BanLec (b) and GRFT (c) binding to gp140. The concentrations of the injected lectins ranged from 0.63 to 10 nM for WT and F84T Malay BanLec and from 0.12 to 10 nM for GRFT. The red lines indicate the actual response measured during the experiment; the black lines indicate the 1:1 Langmuir model fitting. WT and F84T Malay BanLec bind 4 and 8 times slower, respectively, to the gp140 trimer than does GRFT and dissociate 5 and 3 times slower, respectively, than does GRFT. The binding avidities, K_D, of WT Malay BanLec, F84T, and GRFT are 1.22 ± 0.62, 3.68 ± 0.41, 1.37 ± 0.35 (10^–10) (M), respectively. The graphs depict one of the three replicate experiments performed. The figure was generated using Biacore T200 Evaluation Software 3.1 (https://www.cytivalifesciences.com/en/us/shop/protein-analysis/spr-label-free-analysis/systems/biacore-t200-p-05644) and Adobe Illustrator, version 22.1.0 (https://www.adobe.com/products/illustrator.html).

Figure 5

F84T retains antiviral activity against HIV. TZM-bl cells were pre-treated with WT or F84T Malay BanLec, or WT or H84T BanLec and exposed to HIV BaL for 2 d. To assess levels of infection, luciferase activity was measured. Data are representative of two independent experiments. Error bars represent the SEM. The figure was generated using GraphPad Prism 7 (https://www.graphpad.com/scientific-software/prism/) and Adobe Illustrator, version 22.1.0 (https://www.adobe.com/products/illustrator.html).

Figure 6

The F84T mutation interrupts pi–pi stacking between Y83 and F84 without altering the overall structure of Malay BanLec. (a,b) Crystal structures comparing WT BanLec (cyan) to WT Malay BanLec (navy). H84 in BanLec and F84 in Malay BanLec are indicated. Y83 exists in both structures. In (a) the alignment of the monomers is depicted, and in (b) a close-up of the ligand binding loops showing the pi–pi stacking interactions between H84 and Y83 and F84 and Y83 in WT BanLec and Malay BanLec, respectively. (c,d) Crystal structures comparing WT Malay BanLec (navy) to F84T Malay BanLec (yellow). In (c) two views of the alignment of the monomers are depicted. F84 in WT Malay BanLec and T84 in F84T Malay BanLec are indicated, along with the Y83 in common. CBS I and II are also shown, with the ligand binding loop containing the residues of interest in between them. NT N-terminus, CT C-terminus. In (d) the alignment of the dimers is depicted. The figure was generated using PyMol v2.3 (https://pymol.org/2/) and Adobe Illustrator, version 22.1.0 (https://www.adobe.com/products/illustrator.html).