Immunological functions of the liver

This chapter is relevant to Section N1(i) from the 2017 CICM Primary Syllabus, which expects the trainee to "describe the storage, synthetic, metabolic, immunological and excretory functions of the liver."  Though it might feel unusual to think of the liver as an essential organ of the immune system, one must be reminded that it secretes basically all of the complement proteins, is 15% Kupffer cell macrophages by weight, and contains huge amounts of dendritic antigen-presenting cells NK cells and T-cells, which officially makes it a huge lymph node.  The fact that none of this has ever come up in the CICM part one exam makes this reading an optional side quest for the CICM First Part exam candidate, unless they really enjoy reading about complement proteins.

In summary, the immune functions of the liver are:

There is no shortage of papers on the topic of liver immunology. Some of the best would have to be Gao (2016), Gao et al (2008), Zhou, Xu & Gao (2016) and Bogdanos et al (2013) where one of the et al is also Gao. In fact an entire issue of Cellular & Molecular Immunology (May 2016) is dedicated to the various immune things the liver does, and Bin Gao was unsurprisingly asked to write the editorial to introduce it. 

Immune secretory functions of the liver

Broadly, these could be summarised with the statement that the liver is responsible for producing things necessary for immune system function, and for stopping the production of things necessary for bacterial proliferation. Both are good.

Production of complement

No discussion of the immune function of the liver would be complete without a mention of complement, the soluble sci-fi nanoweapon of the immune system.  It would be unfair to say that the liver is doing all of the work here: immune cells of all sorts contribute to the synthesis of key complement components (Lubbers et al, 2017). Still, the liver contributes C1-C9 (except C1q and C7) and fB, which are the most numerous elements in terms of their plasma concentration. To mangle some data from Morgan (2000), one could compile the total complement protein levels into a table, and easily illustrate that the liver produces most of them:

Component	Origin	Plasma concentration (g/L)
C1	Mostly liver	0.18
C4	Liver	0.6
C2	Liver	0.02
fB	Liver	0.21
fD	Monocytes	0.002
Properdin	Endothelial cells	0.005
C3	Liver	1.3
C5	Liver	0.07
C6	Liver	0.065
C7	Neutrophils	0.055
C8	Liver	0.055
C9	Liver	0.06
Total		2.622
Total from liver		2.56 (97.6%)

This reliance on the liver for complement production suggests that liver disease should be considered an immunocompromised state. Certainly, patients with liver disease seem to experience complement deficiency in proportion to the degree of "synthetic failure". i.e. where albumin is low or prothrombin time is prolonged one should also expect complement levels to be proportionally decreased from normal values (Ellison et al, 1990). Does this translate to immunosuppression in the same sense as neutropenia? Reader, it does, but similarly to neutropenia, the levels have to be frightfully low before a clinically significant propensity to death from sepsis can be observed. It appears that even the very last few healthy hepatocytes will continue to secrete complement proteins until the end. Only when the levels of key proteins drop to below 10% of their normal values do you start seeing an increased risk of bacterial infection, and even then the situation is not exactly the fungus apocalypse you tend to see in (for example) bone marrow transplantation. Altorjay et al (2010) found that the mannose-binding lectin (MBL) had to fall to below 100 ng/ml (normal is 800-1000) before their Hungarian cirrhosis patients had an increased propensity for infection, and even then the propensity was not particularly impressive - it manifested as an association with a shorter time to develop the first infectious complication of liver disease (579 vs 944 days from diagnosis).

Soluble pattern recognition receptor secretion

"Pattern recognition receptors" are a class of leukocyte receptor that bristle from the surface of each macrophage and neutrophil, acting as the direct mechanism of recognising something foreign which deserves be be immediately engulfed and destroyed.  For some of these molecules, particularly for the soluble ones, the term could be synonymous with "opsonin", i.e. molecules which recognise undesirables in the bloodstream and bind to them, thereby labelling them as targets for phagocytosis. Without going into too much detail, there is a broad range of these pattern recognition receptors and the majority are localised to immune cells, i.e upon their surface they are expressed, and there they shall remain. Here we are mainly concerned with soluble pattern recognition receptors which are secreted and released by the liver. Of these, we have already dealt with the most important one (complement), which leaves us with the following:

• Penthraxins, an ancient arm of the innate immune system which are often described as the functional precursors to antibodies. These include serum amyloid protein P, C-reactive protein, penthraxin 3, and numerous others. 
• Collectins, which confusingly collect nothing but are collagen-containing C-type lectins,  are a system that overlaps somewhat with complement, as many of these proteins can activate complement via the lectin pathway. MBL is often described as both a collectin and a member of complement. 
• Ficolins are also lectin pathway activators for complement 
• Peptidoglycan recognition proteins, as their unimaginative name might suggest, bind to bacterial peptidoglycan and hydrolysed it, i.e. they have a direct antimicrobial effect
• Lipid binding transferase, an opsonin that binds to lipopolysaccharide

Thus, as one can clearly see, some of these are "opsonin-like" in that they promote phagocytosis, some are complement activators, and some are just fundamentally cytotoxic to bacteria. That's probably the sentence most CICM trainees need to carry with them to the exam, where they can safely drop and leave it. The reader interested in more detail and a lengthy bibliography is redirected to Zhou et al (2016), with the cautionary reminder that this will only enrich their practice and help them grow as critical care specialists, without affecting their exam performance. It could even help address one shameful sign of these decadent times, where one cannot help but note that 90% of the people who order a CRP test do so without questioning what it is, or what it does, or where it comes from, which betrays a depressing lack of curiosity.

Immune function of bile

Apart from being alkaline and extremely corrosive to anything made of lipid, bile also contains antibacterial attack vectors in the form of immunoglobulins. According to Reynoso-Paz et al (1999), the concentrations of the major species are:

Immunoglobilin species	Concentration in bile	Concentration in blood
IgA	0.032 - 0.48 g/L	0.8-3.0 g/L
IgG	0.12 - 0.07 g/L	6.0-16.0 g/L
Polymeric IgA	0.025 - 0.146 g/L	-

It is though that the main reason for the presence of immunoglobulins in the bile is the possibility that the integrity of the lower biliary sphincters may be imperfect, and that the peristalsis in the intestine, rasing intraluminal pressure in the gut, may produce a pressure gradient that causes eneteric organisms to reflux up into the biliary tree. 

Limiting bacterial access to nutrients

Hepatocytes produce numerous proteins which are mainly tasked with the carriage of nutrients to and fro, which makes these nutrients available in the bloodstream. Logically, if some kind of undesirable parasite species were also in the bloodstream, one would expect these nutrients would also be available to them. It would therefore be desirable to pause these mechanisms of nutrient delivery while suffering an incursion of such parasites, so as to minimise their comfort and prosperity. Unfortunately you can't really do that with most of the macronutrients, as they are fairly essential - the brain will not compromise on its need for glucose, for example. Thankfully there are some micronutrients which bacteria need desperately, and humans need only sparingly. Of these, the most important one is iron. 

Iron is required for bacterial cell function. They don't need a lot- trace amounts are enough, such that the importance of iron for bacterial metabolism was not recognised for years because "procedures for purification and detection were not sensitive enough to cope with the minute amount of metal ions ordinarily required by bacteria" (Knight, via Coughlan, 1971). For the majority of clinically interesting bugs, is essential for their growth, reproduction, spore formation, metabolism of energy substrates such as pyruvate, toxin synthesis (eg. for Corynebacterium diphtherae) and all manner of other complex functions. In the presence of abundant iron, infection thrives, and there are numerous examples of this. Therefore, clever mechanisms have developed to restrict microbial access to host iron, and counter-mechanisms evolved to steal it, in a remarkable evolutionary arms race. Before the cynical CICM exam candidate clicks out of this tab to look this up on Part One, in summary of the liver's role in this battle is to produce more chaperone proteins to thereby sequester iron into an inaccessible form:

• Ferritin, which chelates iron with a very high affinity, making it chemically inaccessible, and transport it to sites of even more inaccessible storage
• Hepcidin, a regulatory protein that decreases the availability of free iron from cells which would normally export it, such as macrophages and enterocytes
• Hemopexin, a chaperone for heme (haem? In which case haemopexin?)
• Lipocalin-2, which binds to and disables bacterial siderophores (high affinity iron-binding proteins that have evolved to compete with mammalian iron chaperones like transferrin)

All of these proteins are among the "acute phase reactants" which increase in concentration as a part of the non-specific response to infection. And obviously in this economy it would be impossible to just keep belting out acute phase proteins while also producing the same amount of normal infrastructure proteins (that would be an  unsustainable development model). Thus:

Redistributing priorities in protein synthesis

Though this might be described as a reaction to nonspecific stress rather than an immune-antimicrobial function per se, it is probably still worth mentioning, as it falls into the category of broadly defensive postures adopted by the liver to help the organism survive critical illness, which on an evolutionary timescale has probably often been infectious. If one wanted to take from this paragraph one punchy phrase to throw at a viva examiner, it would be that the liver "considerably changes the plasma proteome and metabolome in conditions of sepsis" (Strnad et al, 2017). A lot of this would fall under the category of a "negative acute phase response", i.e. the liver, which had previously produced copious amounts of albumin, stops doing this during periods of sepsis in order to focus on more important things. Other such proteins include transferrin, retinol-binding protein, antithrombin and adiponectin. These are not just getting sequestered in the tissues because of leaky endothelia - their rate of transcription is genuinely affected by the same cytokines (eg. IL-6) that stimulate the production of stuff like CRP. 

So, one might look at transferrin and agree that in the case of sepsis it would be better to have less of it around, in case some Enterobacter in the blood had a siderophore strong enough to wrestle the iron off of it. But how does it benefit you to have less albumin during sepsis? "The functional significance of the reduced plasma concentrations of the negative acute phase proteins during the acute phase response is not clear", shrug the sort of people who would be asked to write a textbook chapter about this subject. All they could come up with is some sort of amino acid accounting: where the balance must remain the same, some proteins need to be produced less when others need to be produce more, and albumin is a low-hanging fruit for the hepatic budget auditor, as it is produced in copious quantities, has a long half life, and performs no mission-critical task.

Immune filtration functions of the liver

Being a centrally placed organ that receives a huge proportion of the systemic arterial and intestinal blood flow places the liver into a unique position of advantage, from where it can sample and filter the bloodstream. This makes it an effective sensor organ for antigen detection, and an effective strainer for blood purification.  

Elimination of soluble macromolecules by sinusoidal endothelial cells

Those "Pattern recognition receptors" mentioned above are not only confined to the plasma, or to the membrane surfaces of phagocytes. Endothelial sinusoidal cells themselves are able to identify offensive substance in the bloodstream, internalise them, and degrade them. Because many of these are in the form of peptides polysaccharides and lipids, you could arguably describe this process as digestion, as ultimately their breakdown products are going to be exportable macronutrients. Whereas the Kupffer cells are mainly involved in the removal of large insoluble things (eg. red cell fragments), sinusoidal endothelial cells are only capable of moving large soluble macromolecules, with a size limit of apparently around 100 nm. That is still pretty large. For the adventurous reader in need of a long random digression, the article about protein size measurements by Erikson (2009) could give a better idea of what falls under this size threshold. For the rest of us, the spoiler is that this covers virtually all plasma proteins, as an albumin molecule is only 3.8nm in diameter and 15nm long, and 100nm covers a range of size which includes the COVID19 virus particle. Functionally, this means the sinusoidal endothelium clears things like collagen chunks, fibronectin, IgG-complement complexes, as well as smaller pieces such as peptides and cytokines. 

Endocytosis and phagocytosis by Kupffer cells

The liver is about 15% Kupffer cells by weight. This huge concentration of phagocytes is a necessary bulwark against the angry sewer of the portal circulation. Even normal portal venous blood is rich in bacterial cell wall components like lipopolysaccharide, opsonised microorganisms and occasionally actual whole live bacteria. This is not to say that the immune defences of the gut are somehow defective- quite the opposite, we should find it rather startling that more organisms from the gut aren't making it into the blood stream, considering the sheer volume of "luxuriant microflora" in the colon (by mass, over 50% of stool is bacteria). That only a manageable amount of bacteria and bacterial fragments end up in the portal blood is absolutely miraculous.

But one should not develop the impression that the portal circulation is teeming with parasites. Proper portal vein bacteraemia is clearly a very individual and transient phenomenon, as not every sample of portal venous blood grows anything when cultured. Taylor (1956) canullated the portal veins of six cholecystectomy patients and found that, of 167 portal venous blood cultures incubated, only 10 grew anything at all, and out of those only one was a clearly colonic organism (the others all being S.epidermidis). Most interestingly, he specifically cultured some of the patients before and after meals, demonstrating that gut activity does not inevitably shower the liver with gram negatives.

On the other hand, for bacterial fragments such as lipopolysaccharide, there does appear to be a constant baseline leak which is considered normal. Guerville & Boudry (2016) estimated that about 1g of raw untreated LPS is present in the gut of normal healthy people at any given time. Some of this inevitably makes its way into the portal blood, where it encounters a gauntlet of barriers both humoral and cellular (eg. lipopolysachharide-binding proteins, dendritic cells, neutrophils, circulating macrophages). Some LPS molecules make it through this process, and the concentration of LPS in portal blood is variably reported as being around 70-80 pg/ml.  This ends up in the liver, which mops them up so completely that LPS is barely ever seen in the systemic circulation, outside of the scenario of cirrhosis (Lumsden et al in 1988 reported central venous LPS levels of around 30 pg/ml in liver failure patients sick enough to be considered for TIPS).

So, in summary, portal venous blood has plenty of dangerous material in it, and some of it comes in the form of large chunks. Aside from the whole unchewed pieces of bacteria, LPS, ordinarily a 10-20 kDa polysaccharide, can spontaneously aggregate into complexes up to 1000 kDa in size. Also the portal circulation is not the only source of toxic microbial material: occasionally, bacteria and LPS can also appear in the systemic circulation. More broadly, antigens of all sorts end up in the bloodstream, often glued into a tangled mass of immunoglobulin molecules or mobbed by a pile of writhing complement. It is then the job of reticuloendothelial macrophages to identify, engulf and digest these particles, and the Kupffer cells are the most numerous of these, representing 80-90% of the total tissue macrophage population. 

Kupffer cells are exceptionally efficient at their job. When presented with a swarm of S.aureus, mouse Kupffer cells gobbled up 80% of them within 2 minutes, as Zeng et al watched through live in-vivo confocal microscopy. This level of efficiency is largely due to the opsonisation of antigens: Kupffer cells possess high affinity complement and immunoglobulin receptors (eg. the CRIg receptor) which streamlines the recognition and phagocytosis of recognisably foreign materials. Though some authors mention that the size range of potentially phagocytosable particles goes up to 1μm, this is probably an underestimate, as Kupffer cells can also "efferocytose" whole RBCs and tumour cells which can be up to 8-10μm in diameter. The size of a normal Kupffer cell being about 15-20 μm, the size range of objects they should be able engulf is left to the reader's imagination.

Antigen presentation by liver sinusoid endothelium and Kupffer cells

Being the destination for all this garbage makes the sinusoidal cells and Kupffer cells  a very effective system for antigen analysis, as they get to inspect each piece of flotsam in the bloodstream. They are then capable of presenting this material to CD4 T-cells with the appropriate accompanying signals. Even more importantly, they are also capable of not presenting aforementioned antigens, if it is in the interest of the organism. This system acts as a barrier against the inappropriate activation of systemic immunity against common enteric microbial elements and food antigens. This "liver tolerance effect" mechanism necessarily incorporates various forms of anti-inflammatory counter-regulation for branches of the specific immune system, where  suppressive cytokines and negative co-stimulators are used to sedate and mollify the resident hepatic T cell population. This can be a positive thing, preventing you from collapsing with distributive shock each time you eat Vegemite, or it can be a negative thing, dampening the immune response to genuine hepatic pathogens like the hepatitis B and C viruses. 

NK cells

Natural killer cells in the liver are another form of innate immunity, this time specifically directed against tumour cells. It remains to be established whether these are meant to concern themselves with controlling liver neoplasia specifically, or whether they perform some sort of anti-metastatic role for the whole of the body. The liver is certainly a uniquely dense concentration of them, with about 65% of the total liver lymphocyte population being made up of NK cells, whereas the normal proportion in the circulation is about 5-20% (Peng et al, 2016). Their functions, apart from directly killing hepatocellular carcinoma cells, include numerous immunomodulatory activities of a high complexity the understanding of which could not possibly be expected from a fully differentiated intensivist, let alone a mid-lineage trainee.