Tough question, and to answer it, let’s first figure out what we’re talking about. In the broadest possible terms, metaplot encompasses all elements of a setting which are important to the setting, yet not known (or knowable) to people who buy the products, but which may or may not be revealed with the purchase of additional products.
Even as I write that, I realize it’s not the definition I thought it would be. I was thinking about secret history and events unfolding, and while those are definitely types of metaplots, I really came to realize that the unifying factor is much simpler. There are three important elements that make that up:
First, it’s knowledge we don’t have but the developers (hopefully) do. Now, by the nature of fiction, this is something of a necessity. The author almost always knows more about the setting than the reader, and that’s entirely normal. It’s not unfair for game developers to be protected by the same umbrella. Hell, in many ways, this is desirable. Ed Greenwood may know what brand of tobacco Elminster likes, but I will be entirely comfortable never finding out. In large part, this lack of knowledge protects us from trivia, and that’s a good thing (except to the most enthusiastic of completists), so that seems a promising start, were it not for the second point.
Second, the knowledge is important and interesting. Now, both of these are a little subjective, so I’ll concede some fuzz around the edges, but by and large they’re easy to spot and agree on. Important information is information which changes the setting or game in an impactful way. If, for example, the metaplot is going to depower all wizards when the god of magic dies, that’s kind of important, especially if you have wizards in your game. Interesting is trickier. Interesting things are what capture our mind and make us want to know more. We’re given a set of events which are intriguing or exciting, but the last act is blacked out. This seems like an ungrateful complaint – interesting material is interesting because the writer has done his job well; does that really create an obligation to complete the story? But the reality is this interesting stuff is what makes sales. Money is changing hands, and I’ll say that yes, that does create an obligation. And that leads to the third point.
Third, you have to pay for the knowledge. Most often this means you need to buy more books, but it’s possible the knowledge may require jumping through different hoops. Whatever the structure, the lack of knowledge is being intentionally exploited to entice you, the reader, to get more into it. That enticement is what separates most game splats from, say, novels where the interest in further information may be unanticipated, but admittedly most authors make use of the teaser effect to hook you into the next book. 
Looking at these three things, I don’’ think you can make a good metaplot. The tactics behind it are just too gamer unfriendly.
BUT, I suspect you can make something like a metaplot by knocking out elements that make it problematic.
The easiest and most obvious thing you can change is to stop making people dance for the reveal. Put the answers somewhere and let people see them, and be prepared. The reality is an explained metaplot is always lamer than an imagined one. People get excited about metaplots because filling in the gaps inspires them. The good news is that by offering the reveal up front, you don't have time for the reader to get invested in their version and thus more angry and disappointed. Rip off the band aid and just get it out there.
If you must have a release cycle, then make sure it's something inessential. A subplot or a story that might keep the reader's interest from book to book is ideal, especially if it deals with matters that are peripheral to the game, if it turns out people get very invested in the sidebar, then fine, roll with that, but treat it as the stroke of good luck it is.
Now, all that said, if you feel you Absolutely must have a metaplot, consider using the following guidelines:
- Remember Tone - the metaplot should have the same general tone and flavor of the rest of your game. That is, don't go adding Chthulhu in places he's not needed. 
- Wrap it up - stick to short arcs. A single metaplot puts all your eggs in one basket, and the longer it takes, the more likely it is to go horribly wrong.
- Don't Undercut the Reasons to Play - Look, if magic has been lost in your setting, but a prophecy says it will be found again, do not have it found again through your metaplot. That's a job for your players. Too much metaplot gets written as the campaign the designer _wishes_ he could play, thus depriving anyone else of ever getting the chance to do so.
- Clever Sucks - Very often, the need to introduce an idea because it is clever overrides the voice that suggests that it's not actually any fun at all. This includes hidden wordplay, shaggy dog jokes, and obscure cameo appearances.
1 – At this point, “metaplot” may make the jump to becoming “transmedia” if the answers are out there and freely available, but scattered across multiple sources that you must engage with. This is less bad than charging money, but it’s still a tricky line to walk.
2- And I’m ok with that, which lead to me questioning my own perception. I’m opposed to the practice in RPGs but forgiving in novels and movies: why? I think it comes back to #2, importance. When an author teases us, it’s part of the implicit agreement. In contrast, when I buy an RPG, the implicit agreement is that I have what I need to play the game. By excluding things that are key to my playing (important) or which are part of the reason I’m excited about the game (interesting), it feels like the agreement has been broken.
3 - Like swashbuckling. But, really, anywhere. Chthulhu is pretty much the least interesting ingredient you can add to any game at this point. It's done to death. Exception made if, of course, you're actually playing Chthulhu.