Paired-end sequencing generally offers several advantages over single-end sequencing, making it a preferred approach for many applications. Paired-end reads can be aligned more accurately to the reference database, enhancing confidence in taxonomic assignments. The overlapping region allows for error correction and increases the overall read length, potentially improving the downstream analysis. So, the 2x 150bp sequencing strategy provides the advantage of read overlap and improved accuracy, making it a preferred option for 16S rRNA amplicon sequencing.
V4 region is quite small and in perfect conditions could be covered with 2x150 bp strategy. We run one dataset like this and managed to merge the reads afterwards. However, after demultiplexing, primers and adapters removal both reads will be shortened. Also there is a possibility that there will be bad quality bp at the ends that should be trimmed before the analysis. So I would go for 1x300 strategy to avoid any issues with merging step after all trimming and truncation steps. There should be no significant differences at taxonomy annotation step due to the relatively small targeted region (V4) but the overall number of reads may be higher with 1x300 strategy. For V3-V4 or V1-V2 regions paired sequencing (2x300 for V3-V4 and 2x250 or 2x300 for V1-V2) will be better choice.