Optimized expression and specific activity of IL-12 by directed molecular evolution

Abstract

DNA delivery of IL-12 has shown promise in reducing the toxic side effects associated with administration of recombinant human (h)IL-12 protein while maintaining the ability to inhibit tumor growth and abolish tumor metastases in animal models. We have developed a more potent version of IL-12 by using DNA shuffling and screening to improve its expression in human cells and specific activity on human T cells. The most improved evolved IL-12 (EvIL-12) derived from seven mammalian genes encoding both the p35 and p40 subunits of IL-12 showed a 128-fold improvement in human T cell proliferation compared with native hIL-12 during the initial screening of supernatants from transected cells. When purified hIL-12 and EvIL-12 proteins were compared in vitro in human T cell proliferation and Th1 differentiation assays, it was demonstrated that EvIL-12 exhibited a concomitant 10-fold increase in the specific activity of the protein compared with hIL-12. Furthermore, DNA shuffling improved the level of expression and homogeneity of the heterodimer synthesized by 293 human embryonic kidney cells transfected with EvIL-12 by at least 10-fold. Molecular analysis of the variant revealed strategic placement of amino acid substitutions that potentially may facilitate heterodimer formation and product expression. The enhanced expression and biological activity of EvIL-12 may improve the effectiveness of IL-12 gene-based vaccines and therapeutics without the toxic side effects sometimes associated with hIL-12 protein administration.

Figure 1

Library and screening. sh-p35 and sh-p40 libraries for the first round of shuffling (Lib1-p35 and Lib1-p40, respectively) were constructed by using the corresponding cDNA sequences from seven mammalian species. COS-7 cells were cotransfected with a single plasmid from Lib1-p40 and a plasmid encoding h-p35. After 48 h, individual culture supernatants containing either human or chimeric IL-12 were screened for activity in a human T cell proliferation assay. The best six clones were reshuffled to create shuffled Lib2-p40 and the screening repeated with h-p35. The best second round sh-p40 clones were isolated and large-scale DNA preps made for screening with round II sh-p35 clones. Round I shuffled Lib1-p35 was screened in the same manner except COS-7 cells were cotransfected with a plasmid encoding h-p40. Three clones were identified with improved activity after the first round and used to create Lib2-p35. One thousand individual clones from Lib2-p35 were cotransfected along with plasmid DNA encoding sh-p40 and screened in the proliferation assay to yield a representative EvIL-12 molecule.

Figure 2

Human T cell proliferation induced by sh-p40 after two rounds of shuffling and screening. Human PHA-activated T cells cultured with serial dilutions of COS-7 supernatants containing hIL-12 (black bar), IL-12 chimera h-p35/sh-p40 (gray bar), Evolved (Ev)IL-12 (sh-p35/sh-p40, white bar), or pCDNA-GFP (hatched bar). ³H-thymidine incorporation (cpm) was measured during the last 16 h of a 72-h culture. Data were derived from two separate transfections per construct and analyzed by using T cells derived from two separate donors to generate a total of 24 data points per dilution.

Figure 3

Homogeneity of p70 heterodimers. Purified EvIL-12 and hIL-12 derived from transfected 293 cells were fractionated on a 4–12% NuPAGE gel, blotted, and detected with an anti-e-tag mAb. (A) Serial dilutions of the EvIL-12 were conducted to demonstrate the homogeneity of the p70 product compared with hIL-12. (B) Silver-stained gel comparing 293 HEK-derived EvIL-12 (indicated with a down arrow) with CHO-, Sfi-, and 293 HEK-derived hIL-12.

Figure 4

Functional analysis of purified EvIL-12 and hIL-12. (A) T cell proliferation. Shown is a representative experiment of a series of three assays comparing proliferation of human PHA-activated T cells by affinity purified e-tag EvIL-12 (gray bars) and hIL-12 (black bars), derived from transiently transfected human 293 cells. Samples were assayed in triplicate for each dilution and mean ± SD is shown. (B) Induction of human Th1 cell differentiation. Purified human T cells were cultured in the presence of varying concentrations of purified EvIL-12 (gray bars) or hIL-12 (black bars) followed by activation with anti-CD3 and anti-CD28 mAbs. A representative of three experiments is shown.

Figure 5

Alignment of representative round 2 sh-p35 and sh-p40 amino acid sequences with parental protein sequences. Sequence alignment of the most improved sh-p40 (A) and sh-p35 (B) with hIL-12. Numbering starts at the first amino acid residue of the mature form of the human subunit proteins. Colored bars identify predicted parental relationships to shuffled parents. Accession nos. of p35/p40: yellow, human AF180526/AF180563; gray, rhesus U19842/U19841; blue, bovine U14416/U11815; green, porcine L35765/U08317; orange, capra AF003542/AF007576; rose, canine U49085/U49100; and red, feline U83185/U83184. Regions showing homology to multiple species have been indicated with colored dots as follows: bovine–capra–canine–feline (blue dots); canine–feline (red dots); porcine–rhesus–human–feline (green dots); human–rhesus–feline (black dots); no homology (white). Amino acids highlighted in red indicate amino acids different from hIL-12; predicted contact residues for heterodimer formation (44) are highlighted in bold black.

Figure 6

Influence on heterodimer formation and receptor binding by amino acid substitutions. The location of selected amino acid substitutions derived from DNA shuffling is shown by using a molecular model of hIL-12 (PDB1F45). (A) Positioning of glycine residues at the end of β strands. In domain 2 (D2) of p40, two glycine substitutions (G160 and D174G) are shown on β strand E connecting a heterodimer contact residue E181 on loop 3. The exact location of glycine, which corresponds to position 160 in hIL-12, is uncertain due to low resolution of atoms between residues 156 and 164 in the crystal structure and the deletion of four amino acids in strand E of the sh-p40 subunit. Another substitution, E235G, at the end of β strand B in domain 3 (D3) of p40 is shown that may influence the hydrophobic interaction of loop5 with p35. (B) Replacement of bulky side-chain R groups in domain 3 of p40. Substitutions S283K, S285R, R287Q, and Q289R are highlighted in yellow on β strand F along with the dominant contact residue, D290, binding to p35.