The problem is the part of "depends on each other" - if they are separate works, they should stand on their own, as they are going to be reviewed (and potentially read/published) separately. If one cannot stand without the other that's a pretty easy reject for each paper.
Now, it's perfectly acceptable to have two papers that build upon each other. I think I've personally seen a maximum of 3 papers, which end-to-end build upon each other, and were all presented at the same conference.
The key is that each paper must be worthy in its own right, and you must frame each acceptably. I'll use an example from what I witnessed working:
Paper 1: Establish a method fully described in Paper 1 to decide if a person is thinking of Object A or Object B. Mostly theoretical, describing machine learning algorithms and with only a small sample of data to establish if it worked.
Paper 2: Extend method described in 1 to an experiment with human participants, in two experiments which used different approaches of applying the method and tested on a good sized sample of participants. Starts to show potential practical applications of this method, and also highlights issues of accuracy, confusion, how to design objects that are more easily discriminated, etc.
Paper 3: Use methods from Paper 1 and lessons learned in Paper 2 and apply them to a live video game controller scheme, and report the findings on how it worked and what participants thought of the system, which was interesting because the system wasn't all that accurate and it made the games harder.
Each paper had their own unique contributions, and other than a similar intro, listening to all three talks was fascinating, and each paper was it's own accomplishment. If any of the three papers were rejected, the others would still have been appropriate for publication, and you didn't really need to read them in order or all of them to appreciate each one.
So, in the end it is fine if you submit papers that are improved by each other, and which are intimately linked, and even which are better together but which have a different focus and are better written up separately. But if both papers are inextricably linked and truly, strongly depend on each other, that is grounds for rejection of both of them (at least in fields I'm familiar with). This could vary by field, of course.
Ideally, you could make each one sufficiently self-contained as to stand on it's own merits, and still reference each other, making it clear how the papers are distinct and the focus and/or experiment is different. If that isn't possible, I don't know how approaching the editor would help as I haven't seen such a situation, so I'll have to leave it to someone else to advise about that.