This is the foundation for the research article listed in the March 2013 NBC article.

Fourth paragraph of introduction starts:

The discovery by Kanai and colleagues [15] that four brain regions implicated in risk and uncertainty (the right amygdala, left insula, right entorhinal cortex, and anterior cingulate (ACC)) differed in liberals and conservatives provided further evidence that political ideology might be connected to differences in cognitive processes.

Same paragraph concludes:

The ACC is involved in conflict and error monitoring and in action selection [21], [22]. Thus, the regions implicated in risk and conflict, cognitive processes during which liberals and conservatives have been shown to differ in physiological response, are the similar regions shown by Kanai et al. to differ structurally in liberals and conservatives. If patterns of brain activity in these regions during the evaluation of risks could dependably differentiate liberals and conservatives, then we would have further evidence of the link between mental processes and political preferences.

Conclusion of the introduction:

Previous studies [26]–[28] using this risk-taking decision-making task found activity in some of the same regions identified by Kanai et al. as differentiating liberals and conservatives.

Honestly, I think their finding is more accurately conveyed with this sample:

Although genetic variation has been shown to contribute to variation in political ideology [48] and strength of partisanship [53], the portion of the variance in political affiliation explained by activity in the amygdala and insula is significantly larger (see Appendix S1), suggesting that acting as a partisan in a partisan environment may alter the brain, above and beyond the effect of the heredity.

The argument here is not so much that brain chemistry predicts political bias, but rather that political bias can influence brain chemistry.