First of all, keep in mind that graduate admissions is really trickier than matching acceptance/rejections with GPA and GRE results. That doesn't mean they're not important. With an all A transcript and a decent general GRE, your profile will almost certainly get looked at, and that's already great. Physics GRE is also important for Physics PhD application, but GRE subjects in general don't matter much if they're not extra ordinary or terribly bad. So, if you want to retake the GRE to score 820 instead of 800, then don't; it'd be just a waste of time, but if you scored really low on PGRE (I don't know how low is too low for Physics), then retake it. If you want to retake it just to score over 900 and you're confident you can, maybe it's worth it, maybe not (I can't comment on that).
On your main question: Grad school courses certainly do matter a whole lot for US grad school admissions, especially for Math and Physics. Once, a grad admissions chair at one of the top schools told me they almost never accept a person without substantial grad-level courses. BUT, you have publications, and this changes the story quite a bit. In areas like math and theoretical physics, not having publications is not a deal breaker as long as there is a strong transcript and LoR (letters of reference/recommendation). On the other hand, grad course (unofficial and unwritten) requirement for getting into top programs is not really a part of educational "prerequisite" for being qualified, and rather an indicator of potential for success and for being able to handle tough courses under pressure. Having publications while maintaining an excellent GPA also indicates such potentials, and it also shows you're already familiar with research and its challenges.
So, I can't tell you whether you'll get into a top grad program or not, and even the chair of the admissions of the program you're applying to might not be able to tell you that. What I can tell you is that based on the information you've provided here, if you've taken the courses that are expected as prereqs to physics grad courses, your excellent GPA and your publications combined with excellent LoR could tell the admission committee what they want to learn from grad courses on an applicant's transcript, and they probably won't punish you for having a slightly less common way of showing those expected potentials. But still, keep in mind that your SoP and LoR can go a long way, and at the end of the day, grad admission process is really complicated and not a nicely-behaved linear function; rather a chaotic one!