So, this is an attempt to put the 2007 paper, by Gallo & team, on some of the key elements of the cause of rosacea, in ordinary language, so that more people can read it and understand it. I am pretty confident that my interpretation is all correct, but I am not a scientist, nor do I have a science background, so if anyone who is/does finds errors, please point them out so they can be corrected.
“Cathelicidins” are part of our immune system. They can kill microbes (they are 'natural antibiotics'), but they can also set off inflammation. The fact that the kind of inflammation which cathelicidin can cause is similar to that found in rosacea, was the original motivation for the team to investigate a little further, to see if there might be some real connection between the disease and this biological component.
In this study, they first measured the levels of cathelicidin in rosaceans, as compared to normal individuals. In rosaceans, a very high level of cathelicidin was found, and also it was found to be located all through the top layer of the skin, whereas in normal individuals they found very little.
The following pictures display this. To make the pictures (concentrating on (a), i.e. the two top lines of boxes) they apply to a slice of skin something that can 'see' cathelicidin; then they stain the slice, so you can see exactly where the cathelicidin is located. On the left two pictures in the top row, it shows rosacea skin. The red staining is the cathelicidin being picked up (forget all the blobs and lines and structures and strata; that's the detail of the skin. The important thing is the red patch/layer). The middle picture is just a more 'zoomed in' version of the left-hand one. The right-hand picture shows normal skin, and the point is that there is almost no red-staining. The row below is just a 'control' picture.
(b) is a graph showing the levels of cathelicidin found in three rosacea individuals, and three normal individuals. You can see that in all three rosaceans, levels are considerably higher than in the normal individuals, who seem to have levels pretty close to zero.
The pictures of(c), at the bottom right, are based on the same basic 'staining' technique as (a). Only here, they are not measuring cathelicidin itself, but cathelicidin mRNA – which is to say, how much the gene for cathelicidin is being expressed. Again, you can see a distinct difference, of red staining in one case, versus very little in the other.
For the next part of the study, they looked at what kind of cathelicidin is present in rosacea skin. As mentioned at the beginning, cathelicidins can be antimicrobial, or inflammatory. Now, which they are depends upon their size/length, and shape. In very general terms, the longer/bigger versions of cathelicidin are inflammatory, whereas as they get shorter, they're more antimicrobial. This isn't a completely clean distinction, but a general rule – so, the longer forms can still kill microbes, and some of the shorter forms can still initiate inflammation.
What the graphs below are showing are the levels of each kind of cathelicidin. Moving from left to right is moving from smaller to bigger. The higher the spike, the more there is. The top three graphs are for rosaceans, the bottom three are for normal individuals. There's a lot of 'noise' in the graphs, but broadly speaking, you can say that where there is a reasonable spike, it is registering a distinct kind of cathelicidin.
The first, most obvious thing to notice is just how much more spiky are the graphs for rosaceans than those for normal individuals. The second thing to notice are the big spikes on the far right of each of the rosacean-graphs. This corresponds to LL-37, the 'standard' form of cathelicidin. This is the most inflammatory form, and its apparent from these graphs that we've got a lot of it – it is literally off the scale in two of the samples. The third thing to notice is that, as recorded by all those spikes, we've got high levels of lots of different kinds of cathelicidin that just aren't present in normal skin.
Cathelicidin begins life in an inactive form (called hCAP18). It is activated by kallikrein, particularly kallikrein 5 (also called “stratum corneum tryptic enzyme”/SCTE). More precisely, what happens is that kallikrein 5 'chops off' part of the inactive form; and once it is released in this way, it becomes active. This active form is LL-37.
In fact, this chopping up just keeps on going, and that is how you get the shorter/smaller versions of cathelicidin – kallikrein just keeps chopping away.
So, if rosaceans have lots of cathelicidin, and it is kallikrein 5 that activates it/generates the different smaller forms, it is natural to propose that we must have lots of kallikrein 5. So the next step in the study was to check this out. They found that, indeed, there were high levels of kallikrein in rosacea skin compared to normal skin - and that it is located in the same places as the cathelicidin (i.e. so it is able to act upon the latter).
Moreover, although there is a small amount of kallikrein in normal skin, it is located right at the outer layer of the skin. It has a purpose there, namely to help in the shedding of old, dead skin cells. In rosacea skin, by contrast, it is located in the deeper layers, where it really has no business being.
This result is what the top pictures (i.e. (a)) are showing, below. Once again, how they generate these pictures is to get a slice of skin, treat it with something that can 'see' kallikrein, and then take a picture of the slice with a method that shows up that kallikrein. What you're looking at is essentially a vertical slice of the skin (at the top, its the outer layer, and then it goes deeper as you go down). The left, green box shows cathelicidin; kallikrein 5 in the middle box (red); and the two pictures merged together on the right hand side. As you can see, the top row of boxes, showing the rosacea samples, displays a very distinct patch both for cathelicidin and kallikrein; whereas the bottom row, showing normal skin, is much more faded. (In the supplementary data, there are lots more of these pictures, showing it even more distinctly).
As for what the other two parts (i.e. (b) and (c)) of this diagram are showing, well, kallikreins are a type of “protease”. What (b) is showing is total protease activity, again using the same kind of 'staining' technique. If kallikreins are higher in rosaceans, you'd naturally expect total protease activity to be higher. And indeed, again (although these pictures are a bit rubbish) you can see a difference between the rosacea samples, and those from normal skin. The two different pictures (green and blue) are just those generated by two different methods.
What (c) is showing is the result of an experiment which is narrowing down the type of kallikrein we're talking about. Different kallikreins (like number 5, or number 7, and so on) have a slightly different chemical make-up. That means that they respond differently in the presence of other chemicals, and so you can work out the identity of some kallikrein in a sample, by how they are responding to a given chemical. More specifically, the chemicals they use are one's which 'kill' or 'put to sleep' or 'inhibit' kallikreins of a particular type. In the graph of (c), the bars represent the fact that kallikreins are still alive. So, they're not alive in three cases – with the “mix”, with the aprotinin, and with the AEBSF. Since these three inhibitors inhibit “serine” proteases, it shows that the sample contains serine proteases. And kallikrein 5 is a serine protease.
OK, so now we know that rosaceans have high cathelicidin and high kallikrein 5. That's pretty interesting, especially since it is known that cathelicidin can be inflammatory. Its suggesting that our rosacea might be caused by this extra cathelicidin (itself caused by the extra kallikrein 5).
But maybe that's not right. Cathelicidin is expressed by inflamed tissue (other experiments have shown this), so maybe the cathelicidin isn't the cause; rather, its the other way around – we've got high cathelicidin because we've got inflamed, rosacea skin.
This is where the final parts of the study come in. They basically aim to come at it from the other end – to reproduce the conditions found in rosaceans (high cathelicidin, high kallikrein) and see what it yields.
First off, they took the cathelicidin forms that were highly present in rosacea skin (like LL-37, as well as some others) and basically put them in a dish with some skin cells. They did the same with the cathelicidin forms found in normal skin. In the first dish, they found that the skin cells were caused to express inflammatory “messengers” (namely, IL-8). In the body, that would mean that you kick-start the process of inflammation – these messengers tell other cells there is a problem, and basically get the immune system into gear. Nothing happened in the dish containing 'normal' cathelicidin.
OK, next up, they took these rosacea-type cathelicidins and injected them into mice (I know, poor mice. God have mercy on us). 48 hours after injection, the mice's skin starts showing signs of something that looks very much like rosacea, as you can see from the photograph contained in the diagrams below. (b) shows what happens when you inject rosacea-skin cathelicidin on the one hand (left side), versus the kind of cathelicidin found in normal skin (right side). (The supplementary data has more good pictures, showing the consequence of different dosage-levels, and so on. The boxes below the photos, here, show more staining-type pictures of slices of the different skin samples. They're picking up inflammatory cells, but its so messy, you barely know what you're looking at).
By the way, (a) is the graph from the first experiment, showing the levels of IL-8 produced by skin cells, when in the presence of two rosacea-skin cathelicidins (LL-37 and FA-29) and other, normal ones.
OK, next up, more mouse-sacrifice. They deleted the genes in mice that code for cathelicidin, so that these mice can't make any. They then applied irritants to the skin of these mice, and compared the reaction to those of normal ones. c) shows the result, via another staining-type procedure. No doubt there's something there, but personally I can't really make out what I'm looking at. (d) is a lot more unambiguous. This is a graph showing measurements of a particular inflammatory cell, which gives you a rough measure of how much inflammation occurs. The mouse with no cathelicidin (left) shows very low levels of this inflammatory cell; the one with cathelicidin shows considerable levels. So that shows that, indeed, cathelicidin can promote inflammation.
The next experiment was similar. They mice and deleted the gene that codes for an inhibitor of serine proteases (such as kallikrein 5). Because of the lack of the inhibitor, these mice would therefore have more kallikrein – just like us rosaceans. What they looked at in these mice, was the kind of cathelicidin they had, and found it to be LL-37, the type we have in abundance. This firms up the causal link between excess kallikrein, and excess cathelicidin (of abnormal kind), i.e. it shows that too much of the former is going to cause too much of the latter. This is shown in the graph (e). The only relevant spike is the one with the big triangle over it. The top line is showing the situation for mice without the inhibitor. (Note that the graph as a whole has 'scrolled to the right', so to speak, so whereas LL-37 was on the right in the former graphs, its now on the left; and to its right are just really big/large/long forms, of no interest).
Next they injected kallikrein 5 into mice, at a level that reproduces the amount found in rosaceans' skin. Redness and inflammation result. (Figure (f) shows this, in more staining-type pictures. The top box displays a slice of skin that is just overrun with red-stained matter. The bottom box, where the kallikrein is inactivated before injection, shows a slice of basically normal skin, with all the detail and structures visible. The graph to the right is recording cathelicidin, and shows a big spike in the predictable place).
Finally, (g) and (h) are displaying comparisons of the injection of kallikrein 5 into mice without (left) or with (right) cathelicidin. A lot more inflammation occurs in the mice with cathelicidin, for reasons which will now be obvious. Once again, the pictures (g) are a bit unclear, but the graph recording the levels of a particular inflammatory cell (h) leave no room for doubt.
So, these are the experiments on which Gallo's theory are based. We end up with a whole heap of tortured and dead mice, but some pretty solid evidence about rosacea. There's room for more questions to be asked – but questions based around these results.
The rest of the paper fleshes things out a bit.
They note that, since cathelicidins are antibiotics, and rosaceans have abnormal cathelicidins, its possible that in addition to the inflammation that results, we might also have an impaired ability to fight microbes that inhabit the skin. For example, it has been observed that rosaceans have more demodex mites than normal individuals. Well, maybe (the authors speculate) that's because our abnormal cathelicidins aren't as effective at keeping them under control.
They also note that tetracycline antibiotics – which have been a mainstay of rosacea treatment – indirectly inhibit protease activity in the skin. So, their theory explains why this would be effective.
Finally, there is this. Kallikreins do not just active/chop up cathelicidins. Their main job in the skin seems to be to break down the ties holding skin cells together on the outer layer, so that they can slough off. To much kallikrein, therefore, would make this happen too quickly, and the result would be an impaired skin barrier. Rosaceans do very often have skin that is sensitive to irritants, so once again, Gallo's theory would explain this phenomenon.