I'm surprised that almost all of the other answers, while generally providing good advice on "how to say no", seem to be ignoring the bigger picture of "what is the problem?".
Your PM has a problem to solve. Somebody - perhaps his manager, or another team, or a customer or group of customers - is relying on him for a solution, and he is in turn relying on you for a solution. Of course a "no" answer is going to put him on the defensive, because instead of solving his problem, it's simply creating an additional problem (how to explain this to the boss/customer/etc. without getting yelled at, losing his job, or just looking like an idiot).
Maybe he has already made some kind of commitment around this, which, yes, is a poor decision for a manager to make without first asking people in the know, but it happens, sometimes unintentionally.
It doesn't matter how you dress up the "no". It doesn't matter if you phrase it in the form of a question, and it doesn't matter how calm and diplomatic you are about it. Those things help grease the wheels of your ongoing relationship, but they don't help solve his problem!
As the "maker" it is your job, if you must object, to suggest alternatives. Find a way to actually solve the problem, even if it's not in the way that he originally expected. In my experience, only the most stubborn and incompetent managers will insist on a specific solution without justification even when they have been presented with reasonable alternatives.
By far the most common example of this that we probably all experience almost every day is time management. We are told that we have X days to do Y. It will invariably get you nowhere to simply say "Sorry, but I can't do Y in X days, that's just not enough time." You will likely get told to "make it happen" or "find a way to get it done."
So you don't say that. You say:
I am not sure if we can do all of Y in X days, but I have a suggestion: If we assume A, B, and C, then we might be able to take a few shortcuts in Q, and defer part Z until a little later, and I think we could manage that in X days. Would that work?
If your problem isn't time-based, but rather based on some technical constraints, you only need to change a few words. For example, in your case, there seems to be a dependency on some other work or component being completed, so change "in X days" to "without component/feature X being completed first", and instead of phrasing it as an inability, phrase it as a risk.
If you just had this sprung on you and haven't had time to think of any alternatives, then just say so:
I'd like a little bit of time to investigate this request/requirement. I think there might be some complications, but if you can explain to me a little about why you need this right now, I'm confident that I can have a solution mapped out for you by the end of the day.
It's still possible you won't get your way and be "forced" to go ahead despite your objections, but if you've discussed alternatives then presumably he has already explained to you why they won't work, which means that he is actually right, or at least that his rationale is as valid as yours. Also, it's important to time-box this investigation and actually make a commitment to get back to him, otherwise it will just appear (and rightly so) that you're stonewalling.
Incidentally, this works in pretty much any negotiation, and you really should look at this as a kind of negotiation. It's a lot harder to negotiate if you're only willing to put one issue on the table (like, say, an amount of money); that makes the negotiation adversarial rather than constructive, two people fighting over who gets a bigger piece of the pie.
You want to make the pie bigger, by offering other options or concessions that are preferably cheap for you but valuable to them. In a financial negotiation it might be payment plans, interest rates, free gifts, testimonials, that sort of thing. In the workplace, the major currency is usually effort, but you can also negotiate scope, time, quality, and a handful of other domain-specific modifiers (e.g. performance and availability in IT work).
Maybe there is something really simple you could do in a short amount of time that will produce almost the same result or at least eliminate most of the risks that you're concerned about. Maybe not, but at least give yourself the opportunity to think about it before just saying "no" (in any form).