How to nest parameter optimization loop and feature selection loop

I have created the following workflow where i try to find the best subset of features to analyze using the feature selection loop. I use a K-NN node as my classifier and wanted to add a phase of hyperparameter tuning by trying out different possible values for k. I know how to do this 2 steps separately but when i try to connect all the nodes the second loop end doesn’t seem to see the Accuracy metric so i can’t maximize it