This is a simple, if uncomfortable, situation to be in. Unfortunately, it does happen from time to time and you need to be ready for it.
The fact that the defect has been found close to the deadline is, in the short term, irrelevant. Your team has found a high severity defect, so you report it. Given the short timescales, you ensure that everyone who needs to know about it knows about it, so they have the information they need to determine -their- best course of action as soon as possible.
You must -absolutely- not ever hold off from reporting an issue, at least to your local management structure. That would, at the very least, ruin the reputation of your team and could potentially have much more serious consequences.
The next thing to do is to determine the answer to the obvious question: "Why was this found so late?". There are many reasons why this situation could arise - your test preparation could've been too light, you could've mis-prioritised some work, there may simply have been too much to do. As a member of the test team, you need to know what caused the issue and therefore how you can reduce the risk of it happening again.
Obviously, we don't live in a perfect world and it's possible that no action may be taken to resolve the defect before release. It's entirely possible that it makes more sense to go live with a bug and then release a quick fix, than to decide not to release at all. That's a different discussion, though.
For what it's worth, this has happened to me a couple of times over the past dozen or so years. It's a horrible feeling but as long as you've done your best, there's nothing more you can do.