The rapid expansion of e-commerce and social media has led to a substantial increase in user-generated content, particularly reviews. Recommendation systems have leveraged reviews as a valuable resource, as these reviews provide rich information for extracting user preferences and item characteristics. However, review data often contains fake reviews, which include overly positive claims or unfairly negative remarks and can mislead the recommender into generating inaccurate results. Existing research often overlooks fake reviews, challenging the accuracy and reliability of these systems. To address this issue, we propose Fake Review-aware Recommendation (FRRec), a mixture-of-experts model that optimizes fake review detection and recommendation tasks through a collaborative framework. FRRec employs two task-specific experts: one expert focuses on detecting fake reviews by analyzing review text, while another expert leverages reliable data to generate accurate recommendations. First, FRRec incorporates a fake review detection expert to identify and filter out unreliable reviews, ensuring that the recommendation system is based on reliable user feedback. Second, the recommendation expert utilizes the filtered reliable reviews to construct accurate user and item representations through heterogeneous representation learning. This approach enhances the robustness of the recommendation system by leveraging the reliable data. The experiments demonstrate the effectiveness of integrating fake review detection into recommendation systems.