Developing effective subtitle guidelines for Extended Reality (XR) is essential to ensure accessibility for Deaf and Hard-of-Hearing (DHH) users. This study synthesizes existing 2D guidelines and addresses 3D-specific challenges, focusing on subtitle placement in virtual space and guiding methods for off- and on-screen sound localization. We developed a Virtual Reality (VR) prototype in Unity simulating a business meeting with ten use cases, each combining different subtitle placements (world-spaced vs. screen-spaced) with a range of guiding methods. A mixed-method evaluation with ten able-bodied participants assessed factors such as distraction, speaker identification, comprehensibility, and cognitive load. Quantitative results revealed a statistically significant preference for world-spaced subtitles, which were rated as less distracting ( \(p<0.05\) ) and more effective for speaker identification ( \(p<0.05\) ). Guiding methods such as icon bubbles were consistently preferred for their unobtrusiveness and spatial clarity, while others like speech bubbles were often perceived as disruptive. Preferences varied by context, with screen-spaced subtitles favored in exploratory scenarios. These findings highlight the need for scenario-specific, adaptable, and user-centered subtitle systems to improve accessibility in immersive XR environments.