See here: https://www.researchgate.net/post/What_does_negative_factor_loading_imply_in_PCA

or here: https://www.researchgate.net/post/What_does_a_negative_value_for_factor_loading_mean

You must take into account the wording of the items, can be negative by having a reverse content, so in spss must reverse the options of the items to be all the positive factors, according to its theoretical model.

I agree with Cristian Ramos-Vera that all your items should be coded in the same "direction" (usually toward positive wordings). Beyond that, I would recommend using Principal Factors as your extraction technique with oblique rotation (correlated factors).