An alternative would be to label first with the Protein A gold and then with a secondary antibody with an Alexa. Since the gold is relatively big it will leave enough binding sites for the small Alexas.
If you have the protein A Gold and an Alexa coupled secondary antibody at hand already you could just try it. At least the Protein A Gold will not be influenced by the secondary antibody.
I think people can have long discussions on how quantitative Protein A Gold really is. But if you have a big effect in your experiments it will be a good tool to quantify that.