MLR makes the MP test for one $\theta_1>\theta_0$ UMP for all of them

Let $X_1,\dots,X_n$ be a random sample from one-parameter exponential family of distributions, which has monotone likelihood ratio in statistic $T(\underline x)$. Suppose $H_0:\theta=\theta_0$ against $H_1:\theta=\theta_1$ ($\theta_1>\theta_0$) is to be tested. Then MP test based on $T(\underline x)$ for testing $H_0$ against $H_1$ is: