Multilingual learners (MLs) often struggle with science conceptual learning partly due to the abstractness of the concepts and the complexity of scientific texts. This study presents a case of a Virtual Reality (VR) enhanced science learning unit to support middle-school students' science conceptual learning. Using a transformative mixed methods design, this study examines ML's science learning and participation in VR educational programing from the multimodal perspective. Combining the strengths of quantitative and qualitative data analysis, the results of this study show that engaging in multimodal meaning-making in multiple ways, including the VR game, can enable multilingual learners to gain conceptual understanding of science content equally as well as their English-speaking counterparts. Based on the findings, a framework for multimodal instructional design to support MLs is also discussed. Overall, this study contributes to the growing body of research on the uses of VR-enhanced multimodal learning in K-12 science education, especially for multilingual learners.