Clearly tell your players that you intend to run the game in a low-combat mode and that skills and out-of-combat abilities play a much more important role than in a "standard" game. This allows the players to create and advance their characters accordingly and not become stuck in a dead end (this is still possible even with the generous retraining rules).
But this is not a problem, you can still use combat pretty much as written. In contrast to other systems in 4e a player has to go a long way to create a really useless or unplayable character because a certain minimum of combat efficiency is hard-wired into all classes and powers - and even "bad" power choices (e.g. those marked red or worse in the class's handbook) can be useful in certain circumstances.
Just don't tell your players to make characters with a good background without telling them that combat will be a second-class citizen in your campaign. That's just mean because it definitively deviates from the usually expected style of play.
2. Combat is important
Don't use combat as a placeholder or filler. There's nothing that reduces the dramatic importance of combat like random encounters or senseless minion massacres. Make every single combat important and relevant to the plot. Make each combat have consequences and let the PCs face these consequences. For example, if they encounter a suspect or hostile entity while being inside a city they may have to put up with the guard if they simply slaughter the creature.
The latter thing happened to my group last week. I'm currently running the Scales of War adventure path and my group has started the Temple Between part of the AP. While investigating the Erathis shrine in Tradetown the party's dwarf shaman attacked Grovald (both with his axe and a magical effect) in front of several witnesses and was shortly after the incident convinced by the city guard to take a time out in prison. The adventure explicitly mentions (both to the DM and to the players via an NPC) that the PCs have no legislative or executive powers and should not "hero" their way into a temple or something. I'm still contemplating on a credible way to get the dwarf out of prison considering the background of the adventure.
When the players know beforehand what they can expect, they can create and plan their characters accordingly. With little combat in the campaign there's no point in optimizing your character for combat (e.g. spending feats to properly abuse cold or radiant damage). Instead, I'd encourage the players to select powers that have additional, "interesting" effects instead of the highest damage.
Also, remind them to select their utility powers not for combat usage. There is a lot of nice stuff hidden in utility powers that often doesn't get selected because it's not useful in combat.
Another thing to keep in mind are the skill powers from PHB3. Usually you can only swap your class utility powers for skill utility powers of the same or lower level, and you can take Skill Power[DDI] only once to gain an additional skill power. But you could change that limit: allowing to also swap attack at-will/encounter/daily powers for equivalent skill powers or allowing the feat to be taken multiple times. However, the first suggestion may drastically change the base-line combat abilities of the PCs so you should be careful with that option.
Skill challenges are a great way to provide challenges and dramatic events without breaking out into combat. Also, you can tinker around with the mechanics of skill challenges quite a lot as long as you follow the general guidelines of skill check difficulties and complexity without messing the system up too much.
For example, in the campaign I actively play this happened: we were following the trail of a hostile agent of an organization that steals eggs of dragons and dragonborn and corrupts them. We finally made it into the lair of an ancient silver dragon where we unveiled the agent's true intent. The dragon's reaction was not very kind - she bit off his legs and spat him at our feet to do as we wished. My sorcerer and the other player's wizard needed to run a skill challenge to keep the agent alive for long enough to interrogate him. The DM changed the mechanics so that we had a total limit of how much we could miss. Each time we failed a skill check, the difference of
DC - total skill check result was deducted from a pool. Once the pool reached 0 the agent would die. This made it more dramatic since it wasn't just "hit or miss" (as with the normal rules), but instead every single roll was dramatic since two exceptionally bad rolls in a row would have killed the agent in a single round. The number of failed checks didn't matter in that case because the agent was meant to die in the end anyway, it was just a matter of how long we could keep him alive (and thus, ask questions).
In the same campaign in a different situation we had to achieve the opposite. The difference of
total skill check result - DC was accumulated into a pool and the pool had to reach a certain level before hitting 3 failures in the skill challenge.
Rituals and [martial] practices are a good way to support skills, feats and utility powers. They also open up another way to diversify your character or make it more unique. If asked, I would certainly skip over some restrictions (e.g. allowing a character who gets Ritual Caster[DDI] as a bonus feat to swap it for Practiced Study[DDI] if it fits his character/background instead of having him spend a separate feat for it).