It depends on the effect size you are going to use in the meta-analysis. Since mean values are extracted and synthesized in your meta-analysis, they could be influenced by the scoring range of the scales. However, you could transfer them to a standard score (e.g. 0-100) if you want to include them all. Hope this would be helpful.

I agree with Zhunzhun Liu In my opinion, alternatively, you could report proportions instead of mean score. For this, you have to compile frequencies of each scale point used in different studies. In addition, you can report these proportions by sub-group analysis by scale used (e.g. 5-sacales, 7-scales or so) and report separately.

As you will ultimately use means and SDs and N, I don't see it would matter.