I think 'the dark wanderer' and MC-Hambone have really good answers, but I wanted to add a few more suggestions. I read up on this specific scenario to get an idea of what the players may have missed, so I'll give my suggestions for how to handle it in this specific situation, and possibly how to use these suggestions more generally. Hopefully at least a little of this will be helpful.
What are the symptoms of the problem the players are trying to solve?
Think of the plot conflict as an illness that the players are trying to diagnose. If they keep thinking that it's Lupus (it's never Lupus), give them another symptom. First things first, when you read through the scenario prior to the session, keep in mind how the main conflict could be effecting the world. The official book should give you a lot of this, but try to think of any "unintended consequences" to serve as a fallback to give clues to the players.
In this situation: the plants along the riverbank could've displayed signs of being poisoned downriver of the poison source, but no signs of poisoning upriver. This way, if the players move past the source, you can just mention that they move into an area where the plants are lush and healthy. Don't specifically say "the vegetation here is not poisoned"...just say that it's beautiful glade, or has green grass, etc.
If they continue to move in the wrong direction, then mention that the most perceptive character notices that the plants there are not poisoned. This should give them a pretty strong clue that they have moved beyond the source of the poison and need to double back. If their samples keep showing up poisoned, then it should tell them that they have a flaw in their methodology.
More generally: regardless of how the conflict began, consider: were there witnesses? Could those witnesses have mentioned it to anyone? Would they seek out help? Would they seek out any of the player characters specifically?
If there were no witnesses, what sort of "footprint" would've been left in the immediate and surrounding areas? ie, did the responsible party have a campfire in the area? Did they leave in a hurry and forget to extinguish it? Maybe this caused a small forest fire that the players could spot, or even smell days after it was extinguished.
What's special about the player's characters?
Within the game world, why is it that these specific people are the ones who can solve this problem? Why can't the town guard, or random civilians, figure it out on their own? What skills/background/items/traits do the players' characters possess which make them uniquely suited to this task?
In this situation: If any of the players' characters hail from the region, they could have a childhood friend who might work in the place where the poisoning began, and would be willing to confide in the player's character. Maybe one of the players has a skill which might help them notice that the poison levels are too regular (if they aren't cleaning out the pot/vial they're using for the samples), such as magience, or a nature skill. This gives you an opportunity to simply say "You notice _____" if the players are going completely off-base.
More generally: At least glance through the players' character sheets beforehand (though ideally you may want to have a copy of each one handy during the game) so you have an idea of what skills/feats they have. If none of the characters has a high enough passive perception (or equivalent) to notice a clue, and the clue itself is vital for game completion, then consider how their unrelated skills might color their perceptions.
For instance, if the players are trying to track down some bandits, and they notice footprints heading north away from the site of an attack, they might just start heading north. If they don't think to actively track the trail (ie, they just say "we go north"), then they won't notice that the bandits turned east after a few miles. You can push them in the right direction by rolling 1d4 and stating that after that many hours, the character with the highest tracking/nature skill (or equivalent) notices that there is no sign of anyone traversing the area. This tells the players that they need to double back and try to find the bandits' trail, but also gives them a realistic little penalty for going off half-cocked.
Start small — give your players one specific problem to solve
It seems the main issue is:
they don't take initiative AT ALL no matter how I try to make them. I have to ask "What do you do ?" every 5 minutes and most of the time they say they don't know or very vague stuff like they just wanna get the story going
Naturally, when you don't contribute to the story, you have no reasons for role-playing.
In other words, your players lack agency. Perhaps they treat D&D as a computer game, where the story is pre-written, and all you need to do is to make choices (select one from a pre-written list) and participate in combat scenes. So they expect the DM will lead the story, giving them minor choices sometimes.
Instead, let them choose the whole approach. Don't make the adventure too open-ended — for new players it might be paralyzing — but propose a decent task (say, to save hostages from a castle) and let your players decide all the details. How do they do that? Will they choose a sneaky infiltration? Will they try to deceive the guards? Will they make a forceful approach? It is up to them.
Best Answer
Focus Fire: Don't spread out damage any more than you have to. It's far more effective for most of the monsters to concentrate on a single target. If you can get a player down, that's that much less damage per round Team Monster is taking, and the party is suddenly on the defensive to try to save the downed player (only particularly nasty monsters should take the opportunity to coup de grace an unconscious player). Which player should you target first? That leads to the next question...
Target the Squishies: Monsters should always focus fire the available target with the least current HP & lowest defenses. That's who they're most likely to successfully take down. Brutes in particular benefit from attacking low-defense targets, since they have high damage but low attack. Note that I said available target, though. If a sticky defender is guarding a doorway, don't try to rush past him to get at the wizard. But if several monsters can get at the wizard...
Weigh Cost/Benefit for Marks: If a monster is marked, you shouldn't have it automatically attack the fighter or always ignore the mark; only ignore the mark when it's worth it to do so. How much damage as a percentage of the max HP are you likely to do to the defender, versus how much are you likely to do to whoever else you could be hitting? If you're probably going to be more effective against someone else, how much HP do you have left and how unpleasant is the defender's mark punishment? Remember that any target that has defenses that are lower than the defender's by more than 2 is easier to hit than the defender even when the monster is marked.
Flank Them: Have a couple lurkers or skirmishers come in behind the party (they were hidden, used secret doors, or whatever). It's very hard for a single defender to stop attackers coming from 2 different directions. Getting a couple melee units in among the party's ranged backline makes it much harder for the party to focus fire, which means monsters last that much longer.
Control, Control, Control: Make good use of controllers and leaders. Unlike damage, control should be targeted at the characters it will be most debilitating on. Immobilize/slow are for melee attackers, blindness & daze should be dropped on the biggest threats, daze shuts down most defender mark punishment (except paladins), and so on. Control has a huge effect on the difficulty of a fight.
Protect Your Squishies: Artillery and ranged controllers, and when possible leaders, should be stationed towards the back. Use soldiers and minion packs to tie down melee players so that they can't get to your ranged units and to keep defenders busy so lurkers, skirmishers, and brutes can slip past to the party's weaker ranged units.
Use the Terrain: Have your ranged attackers hide behind pillars (remember, if you're directly behind a one-square pillar, you have cover but enemy targets don't). Remember that monsters can use each other as cover if necessary. Position your frontline and backline to maximize the amount of difficult or dangerous terrain melee players have to go through to get past the frontline to the backline.
Alpha Strike: Winning initiative helps; if your group simplifies initiative by having all the players go in order then all of Team Monster go, or vice versa depending on which team won initiative, then winning initiative really helps. Monsters should use any encounter or recharge powers they have early & often. It's better to use it at the start of the fight in a slightly less-than-ideal targeting scenario than to save it for later and end up dying without using it. Even if you can't alpha strike (and it's admittedly much harder for Team Monster than it is for the party), it's hugely beneficial to prevent the players from alpha striking. To give an example, with a proper alpha strike the group I play with have often killed or incapacitated (prone + 1 square away for melee enemies, prone & around a corner for ranged enemies, blinded, etc.) over half the monsters in an encounter before any monsters got to act; the record is 7 monsters (no minions) getting a total of 3 attacks before the fight was over (not 3 per monster; Team Monster as a whole got in 3 attacks before the fight was over, including misses).
Retreat & Reinforce: If a group of monsters that are part of a larger faction in the area are getting their butts kicked, have a few of the remaining ones retreat to try to join an allied group a room or two over. This not only makes the next fight harder without technically altering XP budgets, it has the added bonus of reducing the boring mop-up phase.
Encounter Building: Picking the right monsters to use can make a huge difference in an encounter's difficulty, even when your options all have the same total XP budget. Here are some things to remember:
Frontline & Backline: Try to have a mix of frontline monsters that are designed to keep melee players busy and backline monsters that are designed to hit hard. You will especially need a good frontline if most of your "backline" monsters have to stay in melee range (brutes, some skirmishers and lurkers).
Monster Synergy: Pick monsters whose abilities complement each other. If you have a skirmisher/lurker that does bonus damage to slowed foes, find a controller with a nice AoE or save-ends slowing attack. Daze & prone is another nice combo, especially against melee players; immobilize & forced movement is also strong against melee players. If you have several monsters that give a bonus to adjacent allies, go heavy on melee monsters so more of them can benefit.
Terrain Synergy: Are there a lot of hazards players will have to avoid? Bring lots of forced movement to try to shove them into it. Lots of terrain sources of elemental damage? Pick monsters that resist or are immune to that damage, so they can move through it with impunity. Lots of difficult terrain or features that block movement but not line of sight? Go heavy on ranged monsters so you can pincushion the players as they move to engage.
Control Is Good: Control effects, used properly, will generally be more effective than damage if you have a couple high-damage monsters to take advantage of them and you have some blockers to keep your control monsters from dying the instant the players notice them. Generally speaking, a good ratio is 30% blockers, 20% damage, 50% control. This doesn't mean use nothing but controllers, though: remember that lots of non-controller monsters offer decent control effects, especially as you start getting into paragon tier (by epic, almost every non-minion has some sort of control).
Don't Use Pre-MM3 Solos: This one is admittedly a little specific, but it's good advice nonetheless. Before MM3 solos depended entirely on having lots of HP to survive, but this meant a typical solo fight entailed the party stun-locking the solo for a round or two while the strikers went nova on him, typically ending the fight with little or no damage inflicted on the party. I've actually seen a mid-heroic solo killed in the first round by a single player (admittedly he critted with a daily power, action pointed, and then critted with a second daily power, but still). MM3 and later solos tend to have more ways to negate or ignore status effects (especially stun/daze) and more ways to interrupt multi-attack chains, thus giving them considerably more survivability.
Use Higher Level Monsters When Possible: All else being equal, 4 level+1 foes will generally be a harder fight than 5 level-1 foes. Monster and terrain synergy definitely trump this one, though.
That said, not every monster is a tactical genius. Animals in particular will tend to just concentrate on whoever is closest or doing the most damage to them, and will almost always ignore defender marks (though most predators will attempt to start the fight by going for whoever looks the weakest). Dumber and more cowardly enemies (goblins, kobolds) will usually attack in a big pack, and will almost always respect defender marks.