Engineering biochemical reactions is a key task in synthetic biology to enable biomedical and other applications. The biochemical threshold function is a crucial component in the biosensor circuits to be deployed in living cells or synthetic vesicles for disease diagnosis. In this work, based on the zero-order ultrasensitivity, we propose an economic biochemical implementation of threshold functions with reconfigurable threshold values. We show that the so-constructed threshold function module well approximates the unit step function and allows robust composition with other function modules for complex computation tasks.