Statistical Inference — sufficient statistic combining two normal samples of unequal precision

Let X and Y be independent $N(\theta,\sigma_1^2)$ and $N(\theta,\sigma_2^2)$, where $\sigma_1^2$ and $\sigma_2^2$ are known. Then the sufficient statistic for $\theta$ is