One way to do it would be to use lipid-soluble or water-soluble chemical crosslinkers at moderate concentrations. Another way would be to use fluorescence resonance energy transfer (FRET), in which one protein is labeled with a fluorescence donor and the other is labeled with an acceptor. These could be low-molecular-weight chemical dyes, or fluorescent proteins such as GFP.
Do you want to know whether they interact or in more detail, such as molecular interaction?
Are both proteins embedded in the membrane naturally?
Is there any study suggesting their interaction?
More detail would help us to suggest a solution. With the information here, I recommend using ColabFold for a quick trial (https://github.com/sokrypton/ColabFold).
ColabFold is a project using Google notebook that runs Alphafold with solely protein sequences.
Here is a reference for making protein complex prediction if it is what you mean by "check the interaction." Article Benchmarking AlphaFold for protein complex modeling reveals ...
This paper also demonstrates the complex prediction of membrane protein is accurate with the cryoEM structure. https://www.biorxiv.org/content/10.1101/2022.11.27.518129v1.abstract
One potentially useful tool would be AlphaFold multimer. I would recommend the colab version, which I like a lot: https://colab.research.google.com/github/sokrypton/ColabFold/blob/main/AlphaFold2.ipynb
Depending on sizes of your proteins, you might need to pay for premium GPU and high RAM. AlphaFold does not know anything about membranes, but if there is a conserved interface and sufficient number of orthologous sequences, it will often predict your protein-protein complex. You can then use other programs to add lipids, etc.