PLOS One slaps four papers with expressions of concern for overlapping control data