Due to their potent immune stimulation, tumor necrosis factor alpha (TNF alpha) variants with tumor-homing activity are attractive as novel antitumor drugs. The promising antitumor effect of NGR-TNF alpha in clinical trials triggered extensive interest in developing novel tumor-homing TNF alpha variants in recent years. Owing to its promising antitumor effect, NGR-TNF alpha is usually used as a control for newly developed tumor-homing TNF alpha variants. In our previous works, we produced a pericyte-targeting Z-TNF alpha at high levels using the Escherichia coli (E. coli) M15-pQE30 system. To further compare Z-TNF alpha and NGR-TNF alpha, we attempted to express NGR-TNF alpha using the same system. Surprisingly, native NGR-TNF alpha was expressed at a low (similar to 0.2 mg/L) level in E. coli M15 containing the pQE30 plasmid. However, a single nucleotide mutation of C to G, resulting in a substitution of leucine (L) with valine (V) at the start of TNF alpha, increased the expression of NGR-TNF alpha by similar to 100 times through improving transcription. In addition, the amino acid substitution showed a little impact on the receptor binding, in vitro cytotoxicity, and in vivo antitumor effect of NGR-TNF alpha. As fusing NGR to the N-terminus of TNF alpha with a valine substitution did not reduce the protein yield, the TNF alpha gene with a C > G mutation might be used to prepare novel tumor-homing TNF alpha when the native TNF alpha-based variant is expressed at an extremely low level in E. coli. Notably, in addition to the mutated valine, the impact of N-terminal additional amino acids provided by pQE30 vector on the function of TNF alpha variant must be carefully evaluated.