Protein binding of drugs

This chapter answers parts from Section B(iii) of the 2023 CICM Primary Syllabus, which expects the exam candidate to "describe the distribution of drugs and factors that influence this. Protein binding affects distribution of drugs in a big way. For example, as a drug, one may be bound to plasma proteins exclusively, and therefore be limited to the circulating volume (like ibuprofen); or be bound to tissue proteins extensively, and have a truly preposterous apparent volume  of distribution (like amiodarone). Question 7 from the second paper of 2018 was the only question to specifically ask about this area in pharmacokinetics, but numerous topics link back to this matter, and it deserves its own page of discussion. In summary:

  • Protein binding of drugs is the formation of reversible complexes between drugs and blood components, which may include plasma proteins and the constituents of red cells
  • The main determinants of protein binding are:
    • Number of available binding sites (i.e. protein concentration)
    • Number of drug molecules (i.e. drug concentration)
    • Association constant of the drug and binding site (K)
    • Lipophilicity and pKa of the drug (lipid-soluble drugs tend to be more protein-bound)
    • Resemblance to endogenous ligands of the transport protein
  • Environmental conditions which influence protein binding:
    • pH (affects drug lipid solubility and K value)
    • Endogenous ligands  for the same binding sites, eg bilirubin
    • Protein level obviously affects the protein binding because the number of total available binding sites is affected.
    • Temperature should be mentioned, but let's face it, under the vast majority of circumstances the relevant temperature is going to be normal body temperature or 1-2 degrees above or below.
  • Proteins and blood components that bind drugs:
    • ​​​​​​Albumin (most drugs; six binding sites)
    • α-1 acid glycoprotein ("basic" drugs and steroid molecules) 
    • Lipoproteins (lipid-soluble drugs, eg. cannabinoids and cyclosporin)
    • Globulins (eg. ceruloplasmin binds vitamins A,D, E and K)
    • Haemoglobin (pentobarbital, phenytoin and phenothiazines)
    • RBC membrane (chlorpromazine and imipramine).
    • Carbonic anhydrase "(acetazolamide, chlorthalidone)
    • Specific transport proteins, eg. thyroxin-binding globulin  
  • Clinical significance of drug protein binding:
    • VOD: smaller for drugs with are highly plasma protein bound, larger for drugs which are highly tissue protein-bound
    • Metabolism and clearance: only the free fraction is usually available for dialysis, renal elimination and hepatic metabolism (with some exceptions)
    • Measured drug levels: total drug level may underrepresented clinically relevant concentration of free biologically active drug if there is hypoalbuminaemia
    • Displacement by other drugs, where there is competition for binding sites.

Olson & Christ (1996) were actually the best resource for this, but their paper is unfortunately paywalled by Elsevier. The next best resource was unfortunately Antonio Esteve's Drug-protein binding (1986), which has a number of disadvantages for the modern reader, not the least of which being its baffling lack of legal availability through libraries or online vendors. Select chapters were fortunately available from the Esteve Foundation, and were instrumental to creating the following summary. 

Plasma protein binding 

Drugs are chemically active molecules, and proteins are chemically active molecules, and the bloodstream contains a pretty serious amount of dissolved protein, to the tune of 70-90g/L. It therefore makes sense that, if the two were to meet, they would interact in some way. That interaction usually takes the form of some sort of relatively benign and reversible binding, where the two molecules remain essentially unchanged but move through the bloodstream together in the form of a loosely associated complex, connected by nothing more than the gossamer strands of hydrogen bonds ionic bonds and van der Waal's forces. This can be reliably expected from most of the drugs you're going to be working with, because if they are meant to be administered to patients, they will probably be selected for biocompatibility, i.e. they will not be some sort of horribly reactive substance which will bond covalently to everything and denature all your proteins. 

Because this binding is usually so tenuous, the protein binding is a dynamic equilibrium:

[protein-bound fraction] ⇋ [free fraction]

where

Total drug concentration (100%) = [protein-bound fraction] + [free fraction]

This equilibrium, i.e. what the proportions of each fraction are going to be at any given time, is influenced by several factors, some which will be discussed in detail below:

  • The concentration of the drug
  • The concentration of the protein
  • drug properties, eg. pKa
  • protein properties, eg. number binding sites
  • The association constant (K) of these binding sites, a measure of affinity.
  • environmental properties, eg. temperature and pH

How protein bound do you have to be before you can be described as "highly" protein bound? 70% seems to be the cut off mentioned by authoritative sources, which is completely arbitrary. Drugs which are "highly protein bound" are defined as those in which the protein binding influences pharmacokinetics to the point where you end up having to pay attention to it.

Factors which influence protein binding

Lipophilicity: in general, the more lipophilic a drug, the more protein binding we are to expect. This is a broad rule which is not 100% accurate for all drugs.

Resemblance to endogenous ligands: Generally speaking, proteins which tend to bind drugs are also proteins which bind other ligands endogenously for the purposes of acting as their transport through the circulatory system.

Specific binding sites exist which may be selective for a specific drug family; for example, the same site on albumin binds all the benzodiazepines, whereas a separate site binds all the anionic drugs like warfarin. 

Proteins to which drugs tend to bind

Albumin is the biggest drug binder, and this is likely related to its quantitative dominance (it;s 80% of plasma proteins) but also probably related to the fact that its duties in the bloodstream are transportation and janitorial, i.e. it has high affinity binding sites for bilirubin and fatty acids, which also happen to have an affinity for lipophilic drugs. The albumin molecule has two types of binding sites: Sudlow Site I, which binds warfarin and phenylbutazone, and Sudlow Site II which binds benzodiazepines and ibuprofen (Sudlow et al, 1975). As there are three identical domains making up the human albumin molecule, that gives a total of six binding sites per unit.

α-1 acid glycoprotein is another major drug-binding protein (one occasionally finds it referred to as "oromucoid"). It only has one binding site and it tends to favour "basic" drugs and steroid molecules. 

Lipoproteins bind lipid-soluble drugs, for example cannabinoids and cyclosporin  (Lemaire et al, 1986). This system is interesting because there does not appear to be any shortage of binding sites, i.e. its is neither saturable nor subject to site competition.

Globulins can act as the binding sites for certain drugs. For example, fat soluble vitamins (A,D, E and K) bind to ceruloplasmin, which is an α2 globulin. 

Haemoglobin can bind some drugs, for example pentobarbital, phenytoin and phenothiazine antipsychotics.

Red cell components and the RBC membrane can bind drugs, according to Mats Ehrnebo (1986). For example, the RB membrane has the advantage of being massive and abundantly available. Specific drugs which seem to bind there include chlorpromazine and imipramine.

Carbonic anhydrase can bind drugs. Well, obviously acetazolamide, but also chlorthalidone, a thiazide-like diuretic.

Specific ligands bind their transport proteins;  i.e. transport systems such as transferrin, sex hormone-binding globulin and thyroxin-binding globulin are all a form of protein-binding. However, some might view this "intended ligand" sort of thing as cheating. Using the same relaxed criteria for what drug-protein-binding is supposed to be could lead you to conclude that oxygen is  a highly protein-bound drug. 

Environmental factors which influence protein-drug binding

pH changes the lipid solubility of drugs and influences their protein binding, as well as affecting the association constant for the drug and their receptor. 

Endogenous ligand excess (or, deficit for that matter) can affect the number of available binding sites through pure competition, eg. in the case of hyperbilirubinaemia

Protein level obviously affects the protein binding because the number of total available binding sites is affected.

Temperature should be mentioned, but let's face it, under the vast majority of circumstances the relevant temperature is going to be normal body temperature or 1-2 degrees above or below.

Effects of protein binding on pharmacokinetics

Effect of protein binding on distribution

As mentioned above, the volume of distribution of a drug depends considerably on how much of it is bound to plasma proteins, instead of being tissue-bound. The examples given above includes ibuprofen, which has a volume of distribution of only 0.1L/kg because it binds tightly to plasma albumin and is therefore confined to the circulating volume, as albumin generally is. Amiodarone is also highly protein-bound, but it prefers to distribute into the tissues and the fraction of plasma protein bound amiodarone is rather small. 

Effect of protein binding on metabolism and clearance

In order to get metabolised, metabolic enzymes need to get to you. In order to get eliminated, you need to get through the glomerular filtration barrier. If you are bound to a circulating protein, you are not in solution, and therefore get neither filtered nor metabolised. Thus, it is generally said that only the free fraction of a drug is susceptible to elimination. In the case of highly protein-bound drug, such as warfarin, this results in the protein-bound fraction acting as something of a circulating depot.

Though this works as a broad generalisation for exam purposes, it is not entirely accurate for every drug. For example, some drugs may be eliminated or metabolised by mechanisms that have an even higher affinity for them than do the binding sites on their plasma protein. One example might be organic anions which are eliminated by active secretion in the proximal tubule, where they are extracted from the blood at the basal side of the proximal tubule cells. There, the organic anion transport proteins have such a high affinity for some of their substrates that they can strip them off the albumin they are bound to, defeating the protein binding. This is what happens in the case of furosemide, where a decrease in albumin produces a decreased delivery of the drug to its site of active secretion in the kidney, which then decreases its pharmacological effect. Similarly, propanolol is a highly bound drug which is metabolised by such a high affinity hepatic enzyme system that its rate of clearance completely depends on the rate of its delivery to the liver.

Protein binding and measured drug levels

Protein binding introduces a level of pointlessness into the measurement of drug levels, particularly for highly protein-bound drugs. Usually, when one orders a drug level, the assay involves mixing the patient's blood sample with some sort of high affinity reagent, which then identifies each and every molecule of the drug in the sample. The affinity of such reagents is usually high enough to strip all the drug molecules from their protein binding sites, and so the reported level is the truthful total amount of the drug in the sample. This sounds nice and accurate, but unfortunately it has less clinical relevance than the free drug level. The free fraction of the drug is what you are really interested in, as this is the fraction that has the biological effect. By and large the protein-bound fraction is not clinically relevant, as it is usually unable to get to the molecular tagets.

In short, in highly protein-bound drugs, the drug level may not accurately reflect the magnitude of the clinical effect. Moreover, this will be much worse in the ICU where pretty much everybody has low protein levels. The free fraction of a highly protein bound drug will be even larger, and there will be adequate clinical effect even with lower levels (or, toxicity may develop even with apparently normal levels). A great example of this would be a highly protein-bound drug that has a very narrow therapeutic range and which requires the monitoring of levels, like phenytoin.

Protein binding as a source of pharmacokinetic drug interactions

As mentioned above, there are finite binding sites, and finite protein molecules. It therefore stands to reason that drug molecules will compete for this limited number of sites, and the drugs with the higher affinity will win. If a  highly protein-bound drug was already in the circulation, and a drug with an even higher affinity was administered, it would result in the displacement of the other drug off these binding sites, increasing its free fraction and potentially pushing it into the territory of toxicity. 

References

Olson, Richard E., and David D. Christ. "Plasma protein binding of drugs." Annual reports in medicinal chemistry 31 (1996): 327-336.

Yang, Feng, Yao Zhang, and Hong Liang. "Interactive association of drugs binding to human serum albumin." International journal of molecular sciences 15.3 (2014): 3580-3595.

Gillette, James R. "Overview of drug‐protein binding." Annals of the New York Academy of Sciences 226.1 (1973): 6-17.

Koch-Weser, Jan, and Edward M. Sellers. "Binding of drugs to serum albumin." New England Journal of Medicine 294.6 (1976): 311-316.

Sudlow, G. D. J. B., D. J. Birkett, and D. N. Wade. "The characterization of two specific drug binding sites on human serum albumin." Molecular pharmacology 11.6 (1975): 824-832.

Lemaire, Michael, et al. "Lipoprotein binding of drugs." Drug–protein binding (1986): 93-108.

Ehrnebo, M. "Drug binding to blood cells." Drug-protein binding, Praeger Pub., New York (1986): 128-137.