IRON AND THE KIDNEY: THE LIPOCALIN STORY |
Jonathan Barasch, New York, USA
|
|
Chair: Kai-Uwe Eckardt, Erlangen, Germany |
David Goldsmith, London, United Kingdom |
Prof J. Barasch
Department of Medicine
Columbia University College of Physicians and Surgeons
New York, NY, USA
|
Slide
1
It’s a good opportunity for me to share with you a story about a new kidney protein called Ngal. My lab is devoted to finding growth factors in the kidney that make epithelial cells form and grow and make nephrons form from mesenchyme. In this process we ran into an iron mechanism that I’d like to talk to you about today.
A fundamental question is how cells obtain iron, and how organs obtain iron? The problem seems to be trivial on the surface but when you begin to look at the chemistry of the species of iron, it’s a non-trivial issue. This type of molecule called ferrous iron is soluble at neutral pH but it reacts with oxygen, so the amount of ferrous iron available in circulation is quite low. In fact, it does exist in cells and it is generated at the microenvironments for the purpose of crossing membranes as Professor Hoerl told us a few minutes ago. But the major form of iron in circulation and in food is this molecule and the problem with ferric iron is that it’s completely insoluble at neutral pH. In pure water its Ksp is 10 to the minus 10 Molar which is insufficient to provide cells with necessary iron and in the presence of phosphate this number is probably even lower.
Slide 2
So how do organs obtain iron? The movement of ferric iron from site to site in the body is a non-trivial issue. It requires carriers and of course, the most famous carrier and the one that’s known to everyone is transferrin. There’s been an explosion of new information about iron transport in the past 5 years. Part of the new information has to do with the fact that probably transferrin is not a global iron donor. It’s absolutely essential for red blood cells and for the cycle that our previous speaker discussed. However, if you look at developing kidneys where we feed the embryonic kidney fluorescent transferrin you can see it’s only taken up in these cells right around the forming collecting ducts and in this future distal convoluted tubule. This corresponds to what we see in the in situ for the dominant transferrin receptor number 1. Only the cells around these collecting ducts have this receptor. The middle of the kidney is devoid, the stroma is devoid of this receptor, the outside of the kidney and other parts of the forming nephron, the glomerulus simply does not take up transferrin. In fact, we can take embryonic stem cells that are knocked out for the transferrin receptor and by a genetic trick make these cells fluorescent. Then we can make a chimeric animal composed of wild type embryonic stem cells and stem cells that don’t have the transferrin receptor whatsoever.
Slide
3
These animals are born and their kidneys are normal and you can see that the knockout cells which are labelled with GFP in green are completely competent to form the collecting ducts, the distal tubule, the proximal nephron, the glomerulus, the interstitium, the stroma of the kidney. Transferrin is a molecule that is critical to move iron from many sites to the erythron and probably to other hematopoietic cells but it is not essential for normal organogenesis. Hypotransferrinemic and atransferrinemic animals have been described both in the US and in Japan and apparently there’s normal organogenesis. The knockout of the transferrin 1 receptor is embryonic lethal (created by Doctor Andrews) because of anaemia but you can see that kidney development is normal and we provide chimeric animals to demonstrate that the molecule is not essential.
Slide 4
So what does this mean? How do organs get iron? Well, it turns out that there are many solutions to the problem of iron insolubility. In addition to the transferrin cycle there’s a pool of iron which has not been well described called the non-transferrin bound pool. This pool of iron is heterogeneous and includes circulating citrate, albumin and other proteins and has been most documented initially in the 1980s by Hersko and Kaplan and Cabantchik. NTBI, in other words non-transferrin bound iron is certainly found in iron overload which includes a dialysis patients.
Slide
5
So, in our work we’ve pulled out a protein that we think may be a member of the NBTI pool in certain special circumstances that I’ll tell you about. Our work is devoted to finding growth factors in the kidney. In fact, when we pulled this protein out we did it because it induces nephrons to form from mesenchyme. Here’s our purification of the protein, it’s called Ngal. We produced it from litres of conditioned media and through column chromatography purified this molecule. After we published it and contacted a number of individuals working in allied fields, we found that NGAL is an iron chelator acting in antimicrobial defense. This was found by chance and by chance we discovered that NGAL is one of the most overexpressed proteins in diseases of the kidney.
Slide
6
Ngal is a lipocalin, there are about 50 of these proteins, they’re binding proteins. Some of them are famous like retinol binding protein. Most of these names would be obscure to most biologists and most nephrologists. You can see that we know of their ligands in many of these circumstances such as Heme and in some cases pheromones and retinoids.
Slide
7
Here’s nitrophorin carrying Heme. This is how the vector of Chagas gives you the disease, Rhodnius prolixus bites you and this Heme group carries NO to cause vasodilation at the site of the bite. But when it came to Ngal, information was missing as to the identity of its ligand. So what we pulled out Ngal from a condition media as an inducer, we didn’t know its ligand, we didn’t have a specific function for it.
Slide
8
Then we found out that NGAL is one of the most overexpressed proteins in the kidney after acute and chronic stresses.
Slide 9
This is one lambda of urine from patients who have acute renal failure. These are patients in the intensive care unit. This is the normal amount of Ngal that is in the urine of normals. This is a calibration curve here. This is Ngal in the urine of patients with chronic renal failure. In patients with acute renal failure the level of urinary Ngal rises between 1000 and 10,000 fold. This is reproducible in a mouse model. A ligation of the renal artery for a few minutes produces extraordinary amounts of this protein in the urine. Just take a lambda of urine and run a Western Blot and you’ll see it.
Slide 10
The amount of Ngal produced by the kidney that’s found in the urine is disease dose-dependent. In other words, the more ischemia that is applied to this mouse model from 5 minutes to 20 minutes, the more Ngal is produced and the earlier it’s found in the urine.
Slide 11
Ngal appears in the urine within hours of an ischemic event. It precedes more common markers, urinary markers such as beta 2 microglobulin and NAG by either hours or days. So in this low dose model of cisplatin toxicity Ngal is up within a day, beta 2 many days later. The kinetics of this are quite interesting.
Slide 12
Here is a prospective study of children undergoing cardiac surgery. Kids who raise their serum creatinine by greater than 50% somewhere after 48 hours to 96 hours after the operation those same kids had a bump in their urinary Ngal level at 4-6 hours after the operation. Children who did not raise serum creatinine after the operation never expressed Ngal.
Slide 13
There are rather impressive numbers coming from this study. Additional collaborators at Columbia University working with an adult population after cardiac surgery have found that those patients who double their serum creatinine over a period of days will be those patients that have a rise in urinary Ngal within 3-5 hours after the operation, those are the same patients who have progressive azotemia.
Slide 14
This is adult data now. It’s not only ischemia but as I showed you, cisplatin can activate Ngal expression in the urine, urinary obstruction and delayed graft function in cadaveric renal transplants. You can distinguish between a cadaveric and a living related on the basis of Ngal levels.
Slide 15
So, as a urinary protein, NGAL is a marker of kidney disease. Its level is intensity dependent. I’m told from the literature that it’s reversible. For instance, David Stec, and M.T. Watkins have papers where they use their favorite inhibitors of renal ischemia and the Ngal levels fall in their models, so it’s reversible. Conversely, it can be persistent if the disease is very severe. It can be expressed before there’s an appreciable rise in plasma creatinine. This can take hours or days whereas this Ngal appears within hours. It’s sensitive to low levels of stimuli and many stimuli induce this molecule.
Slide 16
So our next question is, if there’s so much Ngal in the urine of kidneys that have undergone a damaging stimulus the question is where does it come from? The answer here is simple: if we take a kidney after an ischemic event and do real time PCR, the RNA for Ngal rises close to a thousand fold. If you compare that to other markers of ischemia and changes that are known to occur after ischemia, you can see that these are modulated as the literature suggests, but Ngal is more intensive than these other markers.
Slide 17
This RNA we think comes from the kidney tubule. Here’s a surprise. In ischemia it turns out it’s the thick ascending limb. We believe this is the thick limb that produces enormous quantities of this protein. However, if you change the type of stressor and use obstruction, it’s different epithelia that expresses NGAL and it turns out it’s the collecting ducts shown by in situ hybridizations.
Slide 18
Perhaps some of the urinary Ngal derives from the neutrophil, after all Ngal is neutrophil gelatinase-associated lipocalin and neutrophils produce a lot of it. However, the isoelectric points of the protein in the urine, and acute renal failure and in chronic renal failure differ from what we found at least from peripheral neutrophils. We’re going to pursue this data by destroying the neutrophils in circulation. There’s a wonderful neutralizing antibody that completely depletes neutrophils and I’ve made a conditional knockout of Ngal in order to remove it from neutrophils and definitively say that urinary Ngal comes from these tubules as a result of a disease stimulus
Slide 19
There’s a second pool of Ngal besides for the urine. The second pool of Ngal is found in circulation and it’s made by the liver. In these patients with acute renal failure and sepsis and shock you can see it in the serum. Unlike the urinary form of Ngal, the plasma form of Ngal does not make it to the urine, it is filtered but it simply doesn’t make it past the proximal tubule. We know this because when we inject huge quantities of Ngal that’s fluorescently labeled, it all accumulates in the proximal tubule. If you use a knockout of megalin that receptor in the proximal tubule that picks up many ligands, then Ngal appears in the urine but in the wild type not. So there’s a second pool of Ngal outside of the urine, it’s in the circulation. It’s probably made by spleen and liver and it’s filtered by the kidney and ends up in the proximal tubule.
Slide 20
This is ongoing in humans and diseases as demonstrated by this biopsy where in a number of disease models you can see the proximal tubules accumulating Ngal here in Bowman’s and the take off of S1 in the proximal tubules.
Slide 21
So we have two pools of Ngal, a urinary pool that we think could be a clinically important marker of kidney disease, it’s expressed by renal tubules and a plasma pool that is filtered by the glomerulus and reaches the proximal tubule. Two parts of the kidney, 2 different pools of the protein. Are these pools of protein important for the kidney? We think so but I’d like to stress these are preliminary data, but the knockout of Ngal is an animal that has a number of issues, including when subjected to ischemia, these animals all die, after the creatinine rises, whereas in our low dose ischemia model this one, the normals peak creatinine at 1.5 and then come back to normal. We think that one of these pools of NGAL, perhaps the peripheral pool that perfuses the proximal tubule or perhaps the urinary pool could play a critical role.
Are these pools of protein important for the kidney? We think so and I’d like to stress this word preliminary, please preliminary but the knockout of Ngal is an animal that has a number of issues, I’ll show you in a minute but when subjected to ischemia, these animals completely lose it, they all die, the creatinine rises, they all die. Whereas in our low dose ischemia model this one, the normals peak, their creatinine at 1.5 and then come back to normal. The knockout kidney and the normal kidney. We think that one of these pools, either the peripheral pool, the circulating that goes to the proximal tubule or the urinary pool could play a critical role.
Slide 22
So the question then is so what does this protein do? It is made in abundance. What does it do? When you ask that question of a lipocalin the first consideration is the ligand, because these are carrier proteins. So what are the ligands in the urine? What are the ligands in the plasma? Is there one ligand, is there more than one ligand? Here again the story depended on serendipity. When we cloned this protein, the protein was reddish.
Slide 23
The solution to this red color came from a crystallographer in Seattle, R. Strong. He told me that what seemingly was an artifact had occurred with NGAL. When we cloned the protein in bacteria, the protein calyx bound a bacterial molecule, a tris-catecholate, 2, 3-dihydroxybenzoic acid three of them suspended by this serine-lactone bridge. This molecule is called enterochelin. It’s the highest affinity substance that’s known to bind iron, it’s made from bacteria and it’s the way that Gram negative organisms such as E.Coli obtain iron from their environment. They secrete this tris catecholate and then recover it with a receptor. Ngal interrupts that pathway by binding this iron cassette.
Slide 24
As I say when you clone a protein in a heterologous environment like bacteria it is subject to many artifacts and despite the argument that this is a high affinity interaction of this cassette with this protein it seemed rather bizarre until the knockout published in Nature by Flo et al. demonstrated that these animals die of septic shock very readily. Again completely unexpected. If you let the animals get old in the lab they develop skin lesions that are filled with bacteria. So it turns out that our carrier protein, made by the kidney is an antimicrobial. It binds to bacterial iron carrier and interrupts an iron delivery pathway for bacteria. It’s essential for that function.
Slide 25
Well, I told you that when NGAL is expressed in the plasma, it is filtered by the kidney. So when we have an infection, Toll-2 or Toll-4 is ligated and activation of the liver results in Ngal expression. By an experimental model that I have demonstrated, Ngal binds the siderophore and iron and drags the radioactive iron complex to the proximal tubule of the kidney. You can see the radioactive decompositions on this radioautograph. And this molecule is therefore circulating in plasma in the setting of many types of bacterial infection, or possibly fungal infection and it will move iron in the form of a siderophore form complex to the proximal tubule. This is not simply a clearance event because approximately 700 genes turn on in the setting of this delivery.
Slide 26
We know that the delivery of Ngal with its iron to the proximal tubule is an active process because it actually upregulates iron dependent events. These are iron probes that we invented, this one increases in fluorescence in the presence of iron. This one decreases in fluorescence in the presence of iron and when we add Ngal, we see an upshift with this probe, a down shift with this probe.
Slide 27
Hence Ngal donates its iron. In doing so it seems to create protection from ischemia. Here it’s a siderophore, siderophore antagonist or the siderophore iron plus Ngal. The iron in the complex seems to be protective because if you use a “dominant negative” metal, gallium i, you don’t get that protective effect. Hence, this complex seems to have activity in the proximal tubule. One possibility is that it activates heme-oxygenase as a downstream molecule, but as I said there are approximately 700 events downstream of Ngal delivery to the proximal tubule.
Slide 28
Does this make sense? Well, this is how we first discovered the molecule in the first place. Here is mesenchyme. When we add Ngal, we get conversion to epithelia and to tubules resembling nephrons.
Slide 29
This is a routine assay in our lab, this conversion, and it completely depends on having iron in its pocket. In the absence of iron or in the presence of that dominant negative metal the conversion doesn’t work. So iron Ngal is a bioactive molecule that stimulates epithelialization and epithelial gene expression.
Slide 30
This has been shown in cell lines by V. Sukhatme, Beth Israel Boston. Ngal with iron activates E-cadherin expression. Ngal without iron does not. E-cadherin expression is iron-dependent, take iron away there’s much less of that epithelial protein after iron chelation.
Slide 31
So what do we have? We have a new kidney molecule. It’s a chelator of small organic molecules. One of these small organic molecules is a siderophore made by bacteria and it traffics siderophore to the proximal tubule. Ngal also serves to upregulate epithelial genes both in the proximal tubule and in embryonic tissue and in cell lines. Is this molecule only active in bacterial infection? Are there other ligands? Here we’re going to leave this at speculation. The only firm evidence is for this bacterial substance so far. But perhaps there are mammalian forms of these organic molecules.
Slide 32
Take a drop of urine, mix it with radioactive iron, let it run on a chromatogram, the urine will capture the iron and drag it to the front. Here is the iron without the urinary substance. Normal human urine, mouse or dog urine all contain substances that bind iron and solublize it so that it runs on a chromatogram. These urinary molecules also bind Ngal. Perhaps there’s an underlying set of organic molecules that have functions now demonstrated for a bacterial substance.
Slide 33
So how do we move iron? How do cells obtain iron? In a number of conditions we have to look for transferrin independent mechanisms of moving iron. Transferrin is not the full explanation, it is for red blood cells but it is not essential for other cells. In the plasma there’s this new molecule called Ngal which binds siderophores. This is where the data is very strong and this transports iron to the proximal tubule of the kidney where it activates a set of epithelial genes. In the urine there’s also the same molecule produced by a different part of the nephron.
Does it serve as an antimicrobial? Does it interact with an endogenous molecule? Classical papers from Paller and Nath did demonstrate that in many forms of acute renal failure, urinary iron appears and we think that Ngal could be acting to bind this iron with endogenous organic molecules. It’s very clear from many, many studies that Ngal is massively expressed in common kidney diseases. We think it’s time to finish small studies and move on to larger multicenter clinical studies. You’re invited to join me as a collaborator, I’d be glad to help any of these studies. We think it could even come into the clinical routine as an early marker of renal tubular dysfunction.
Slide 34
Thank you for the invitation to come here and to tell you about a new molecule in the kidney. I’ll point to two members of my lab Kiyoshi Mori is now an assistant Professor at Kyoto University. He’s the one who took our basic studies with Ngal and moved them into the clinical arena. And I’ll point to Kai Schmidt-Ott, a person who came to me from Humboldt University in Berlin and he has demonstrated many hundreds of genes downstream of Ngal most of which are epithelializing genes.
Slide 35
Chairman: Thank you so much for this elegant summary of a really fascinating story. Paper is open for discussion.
Question: A very interesting story. Uropathogenic microorganisms need iron to survive in the urinary tract and to do this the siderophores have very high association constants for iron in a range of 10-30 or 10-50 the highest iron binding constant you ever will find in biology. So my question is, did you have a chance to measure the association constant between Ngal and iron because otherwise it could be difficult to prevent iron binding to iron bacteria or other siderophores in bacteria if Ngal does not have a similar high association constant?
Chairman: Right ok. So the question has a number of parts to it. The first is that bacteria are specialists at obtaining iron. They do so by secreting these organic molecules as Professor Horl said with extraordinarily high affinity for iron.
Prof Barasch: Now how do they capture this? Well they have a receptor on their cell surface for these small organic molecules with iron. The affinity of some of those receptors for this iron cassette is around 0.4 nanomolars. But the affinity of Ngal for the same complex with iron is 0.4 nanomolars. So we think an answer to the first part of the question is that Ngal can serve as a decoy to pick up a molecule that should be trafficking to bacteria. So if you try to express Ngal in bacteria, it’s hard to actually express the protein, you have to keep adding extra iron into the media. So Ngal could provide a decoy, an interruptive pathway for bacteria to obtain iron. Now the question goes on to ask, what about endogenous molecules? Would they have similar high affinities? And here I don’t know, but the catecholate family as a whole if known to interact with iron. The last part of the question I think is the most interesting idea. That is if Ngal is made in large quantities by the kidney, the question is, is it being made to inhibit a bacterial infection? I heard some data earlier today that demonstrated that during renal failure there’s a much higher rate of sepsis, but what I’m not sure about is that during renal failure (maybe some of you could tell me) there’s a higher rate of urinary infections, catheter independent and so on. I’m not sure that there are higher rates of infection. After all it’s not clear to me that when you give a bolus of cisplatin, why the kidney makes so much Ngal when the urinary tract is sterile in the first place. You can say that Ngal is preventive and therefore the knockout should be subject to urinary infections but when it comes to mouse microbial models of infection compared to wild type human infections, it’s a very difficult to know if the model validates the real phenomenon. So why is Ngal made in such abundance? Well, it could prevent bacterial infections but this hypothesis is difficult to prove. So perhaps it is there to chelate iron by picking up a urine siderophore and hence to prevent iron availability to microbes would be our current hypothesis.
Questions: Does the kidney actually make a lot of Ngal under normal conditions? I mean you showed very nicely that it’s upregulated very rapidly but is a lot of Ngal actually stored in these cells and then released upon injury?
Prof Barasch: No it isn’t, it’s not stored at all. The urinary level is about 25 nanograms/ml normally and the amount in a human in septic shock could be on the order of 2-5 micrograms/ml. The RNA level is extremely low without an injury and then the RNA goes up within 2-3 hours, even 1000 fold. It’s an early response gene and the protein is made and immediately secreted into the urine. Likewise the liver upstream is excreting Ngal in the blood stream which at the same time headed for the glomerulus.
Question: Do you know whether it’s increased in expression or there’s more urinary amounts of it in renal tumours? Because human kidney molecule 1 is also increased but only acutely but in the tumour situation.
Prof Barasch: Ok that’s a very interesting question. You know in looking for clinical applications I thought what about bladder tumours. They’re very hard to diagnose, they depend on cystoscopy, so what about a urinary marker for bladder tumours? So transitional epithelia does not make this. Ok, now in terms of kidney tumours it is a very interesting question, and the only thing I can tell you so far is that we picked up a Wolffian duct tumour in a patient, massive amounts of Ngal are made by the Wolffian duct tumour, but I don’t know if other forms express Ngal. I think a diagnostic marker for renal epithelial cancers would be fantastic.