Download as pdf or txt
Download as pdf or txt
You are on page 1of 23

Expert Opinion on Drug Metabolism & Toxicology

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/iemt20

Crosstalk of physiological pH and chemical pKa


under the umbrella of physiologically based
pharmacokinetic modeling of drug absorption,
distribution, metabolism, excretion, and toxicity

Lu Gaohua, Xiusheng Miao & Liu Dou

To cite this article: Lu Gaohua, Xiusheng Miao & Liu Dou (2021): Crosstalk of physiological pH
and chemical pKa under the umbrella of physiologically based pharmacokinetic modeling of drug
absorption, distribution, metabolism, excretion, and toxicity, Expert Opinion on Drug Metabolism &
Toxicology, DOI: 10.1080/17425255.2021.1951223

To link to this article: https://doi.org/10.1080/17425255.2021.1951223

© 2021 The Author(s). Published by Informa Published online: 31 Jul 2021.


UK Limited, trading as Taylor & Francis
Group.

Submit your article to this journal Article views: 176

View related articles View Crossmark data

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=iemt20
EXPERT OPINION ON DRUG METABOLISM & TOXICOLOGY
https://doi.org/10.1080/17425255.2021.1951223

REVIEW

Crosstalk of physiological pH and chemical pKa under the umbrella of


physiologically based pharmacokinetic modeling of drug absorption, distribution,
metabolism, excretion, and toxicity
a b
Lu Gaohua , Xiusheng Miao and Liu Douc
a
Research & Early Development, Princeton, New Jersey, USA; bDrug Metabolism and Pharmacokinetics, GlaxoSmithKline, Collegeville, Pennsylvania,
USA; cSchool of Pharmaceutical Sciences (Shenzhen), Sun Yat-sen University, Guangzhou, Guangdong, China

ABSTRACT ARTICLE HISTORY


Introduction: Physiological pH and chemical pKa are two sides of the same coin in defining the Received 6 May 2021
ionization of a drug in the human body. The Henderson-Hasselbalch equation and pH-partition Accepted 30 June 2021
hypothesis form the theoretical base to define the impact of pH-pKa crosstalk on drug ionization and
KEYWORDS
thence its absorption, distribution, metabolism, excretion, and toxicity (ADMET).
Absorption, distribution,
Areas covered: Human physiological pH is not constant, but a diverse, dynamic state regulated by metabolism, excretion and
various biological mechanisms, while the chemical pKa is generally a constant defining the acidic toxicity (ADMET);
dissociation of the drug at various environmental pH. Works on pH-pKa crosstalk are scattered in the Henderson-Hasselbalch
literature, despite its significant contributions to drug pharmacokinetics, pharmacodynamics, safety, equation; model-informed
and toxicity. In particular, its impacts on drug ADMET have not been effectively linked to the physio­ drug development (MIDD);
logically based pharmacokinetic (PBPK) modeling and simulation, a powerful tool increasingly used in modeling and simulation;
model-informed drug development (MIDD). physiologically based
Expert opinion: Lacking a full consideration of the interactions of physiological pH and chemical pKa in pharmacokinetic (PBPK)
a PBPK model limits scientists’ capability in mechanistically describing the drug ADMET. This mini- model; pH-partition
hypothesis
review compiled literature knowledge on pH-pKa crosstalk and its impacts on drug ADMET, from the
viewpoint of PBPK modeling, to pave the way to a systematic incorporation of pH-pKa crosstalk into
PBPK modeling and simulation.

1. Introduction species to different extents in a medium with varying pH.


pKa is the pH of the medium where a monoprotic drug is
In the context of pharmacology, physiological pH and the
50% ionized.
acid-base dissociation constant (pKa) are two sides of the
Physiologically based pharmacokinetic (PBPK) modeling
same coin when considering the effects of ionization or pro­
and simulation of drug ADMET has been increasingly applied
tonation on drug absorption, distribution, metabolism, excre­
by biopharmaceutical companies and regulatory agencies in
tion/elimination, and toxicity (ADMET). Physiological pH and
drug discovery and development. As will be discussed, it will
chemical pKa are two necessary and sufficient components to
not be possible for a PBPK model to accurately predict the
define the drug ionization in the biological fluid and/or tissue.
pharmacokinetics (PK), pharmacodynamics (PD), safety, and
In general, the pH of a fluid (and a tissue or an organ as
toxicity if it is missing key interactions between the physiolo­
well) is defined by the concentration of hydrogen ions (pro­
gical pH and the chemical pKa therefore, lacking the funda­
ton, H+) within the fluid (and/or tissue/organ) only, with a pH
mental description of drug ionization in a specific fluid/tissue/
of 7 considered as neutral. However, a neutral pH does not
organ.
always exist in the human body which consists of various
The impacts of ionization on a drug’s ADMET properties
acidic or alkalic fluids, tissues, and organs. As a typical exam­
have been a subject of interest to the scientific community
ple, the plasma is weakly basic, with its pH carefully regulated
from the dawn of modern pharmacology. A good example is
around 7.4 (7.35–7.45). Indeed, human physiological pH is not
cocaine, the first local anesthetic which was discovered by
a constant, but a diverse and dynamic state regulated by
Albert Niemann in 1860 [2]. An early study by Bignon in
various biological mechanisms, mainly by the respiratory and
1892 showed that alkalization increased the activity of cocaine
renal systems [1].
solutions [3]. Both free base and ionized forms of cocaine are
In contrast, pKa is a key physicochemical parameter defin­
necessary for its activity as local anesthetic. In fact, cocaine
ing the tendency of a drug (molecule or ion) to keep a proton
enters nerve fiber as neutral (free) base and its cationic form
(H+) at its ionization center(s). By definition, the pKa is
blocks nerve conduction by interacting at the inner surface of
a unique parameter which is specific to chemical structure.
the sodium channel – an excellent example of how
An ionizable drug can dissociate into separate, charged

CONTACT Lu Gaohua Gaohua.Lu@bms.com Quantitative Systems Pharmacology & Physiologically Based Pharmacokinetics, Clinical Pharmacology &
Pharmacometrics, Research & Early Development, Bristol Myers Squibb, Princeton, New Jersey, USA
© 2021 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group.
This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License (http://creativecommons.org/licenses/by-nc-nd/4.0/),
which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited, and is not altered, transformed, or built upon in any way.
2 L. GAOHUA ET AL.

Article highlights
challenges remain for PBPK modelers when integrating many
● pH-pKa crosstalk determines the drug ionization and further impacts different information sources into PBPK modeling and simula­
on drug ADMET.
● However, the state-of-the-art PBPK models have not incorporated the tion. The purpose of the current mini-review is to summarize
pH-pKa crosstalk systemically and consistently in modelling ADMET. the impacts of pH-pKa crosstalk on drug ADMET at the highest
● In this review, we have thoroughly compiled our knowledge from level, from the viewpoint of PBPK modeling and simulation.
literature on hitherto elucidated impacts of ionization on drug
ADMET. Figure 1 shows pH and pKa impacts on drug ADMET.
● The review with detailed information can serve as a guidance for
considering pH-pKa crosstalk in PBPK modelling and simulation.
2. Physiological pH in the body
In general, the human physiological pH for most tissues is
around 7 ~ 7.4 with some exceptions (Figure 2). For instance,
the stomach pH is 1 ~ 3, small intestine pH is 4 ~ 7, and blood
modulation of a drug’s ionization state can affect the PK and pH is 7.3 ~ 7.45. Therefore, the same drug may exist in
PD relationship or pharmacology of the drug. different ionic states in different regions of the body.
In the middle of 20th century, organic chemists and biolo­ Elsewhere in the body, changes in pH tend to be much smaller
gists found that the chemical and biological reactivities and show less deviation from the pH of blood and human life
depend upon the degree of ionization of both the drug and requires a tightly controlled pH level in the serum of about 7.4
the receptor [4]. Shore and coworkers in the United States (a slightly alkaline range of 7.3 ~ 7.45) to survive.
Food and Drug Administration (FDA) demonstrated the Knowledge of tissue and body fluid physiological pH is
dependency of gastric secretion of weakly acidic or basic important, especially when developing PBPK model for drug
organic electrolytes upon the ionization constant of the com­ transport and delivery through different tissues, such as eye,
pound, now known as the pH-partition hypothesis [5–8]. inner ear, brain, skin, lung, and tumor tissues. Typical pH
Undoubtedly, the pH-partition hypothesis has been the theo­ values of various human body fluids are listed in Table 1
retical base for understanding the passive permeation pro­ [11–14]. Although new information is continuously emerging,
cesses of drug absorption, exsorption, disposition, and the most recent examples of reported values are compiled
excretion. Most, if not all, PBPK models (including the state- here together with their original sources.
of-art commercial software platforms GastroPlusTM, Simcyp®)
utilize this fundamental pH-partition hypothesis. 2.1. Tear
The impacts of drug pKa and physiological pH on drug
Tear pH was measured in 44 normal subjects by immersing the
solubility and permeability have been extensively discussed
lip of a micro-combination glass pH probe in the tear fluid in the
[9]. Charifson and Walters recently reviewed potential advan­
inferior cul-de-sac [13]. The normal pH range was 6.5 to 7.6, with
tages and disadvantages of both acidic and basic drugs, pro­
a mean value of 7. pH on the ocular surface was 7.11. Older
vided new analyses based on available public data, and
women had a more alkaline pH than other subjects. A pH shift
discussed the effects of ionization state on drug metabolism
from acid to alkaline was observed throughout the day [15].
and PK (DMPK) properties including solubility, permeability,
oral bioavailability, metabolism and clearance, tissue distribu­
tion, volume of distribution, and protein binding [10]. 2.2. Ear
Although there are many publications on the influence of Various fluids in the ear have varying pH [16]. The cochlear
pH-pKa crosstalk on drug ADMET, to our knowledge, perilymph pH is 7.3, the same as the cerebrospinal fluid pH in

Figure 1. pH and pKa impacts on drug ADMET.


EXPERT OPINION ON DRUG METABOLISM & TOXICOLOGY 3

Figure 2. Physiological pH values in the human body.

Table 1. pH values of human body fluids. extracellular pH is about 7.4. Therefore, the pH gradient is
pH reversed in the tumor tissue.
Fluid Mean Range
Blood, arterial 7.424 7.386–7.462
Blood: red cells 7.209 7.175–7.243 2.4. Skin
Blood: plasma, arterial 7.39 7.35–7.43
Blood: plasma, venous 7.398 7.378–7.418 The skin pH is also well known to be acidic with values as low
Bile, hepatic 7.5 6.2–8.5 as 5 at the skin surface and still below 7 in the horny layer;
Bile, gall bladder 6.0 5.6–8.0 although in atopic dermatitis and contact dermatitis the mean
Breast milk 7.01 6.4–7.6
Cerebrospinal fluid 7.349 7.327–7.371 values reported are a little higher (5.3 to 5.5 instead of 5–5.1),
Faces, adult 7.15 5.85–8.45 they are still quite acidic [18].
Faces, infant 4.9 4.6–5.2
Gastric juice, male 1.92 ± 1.28
Gastric juice, female 2.59 ± 2.08 2.5. Nervous system
Gastric juice, child 3.27 0.9–7.7
Gastric juice, newborn 2.52 1.2–7.4 Differences have been reported between general cellular and
Pancreatic juice 7.5–8.8
Saliva 6.4 5.8–7.1 interstitial pH and cerebrospinal fluid (CSF). The pHs for the
Semen 7.2–8.0 neural interstitial fluid, intracellular fluid, and cerebrospinal
Sweat 4.0–6.8 fluid are 7.44, 7.05, and 7.31, respectively [19,20].
Synovial fluid 7.434 7.31–7.64
Tear 6.5–7.6
Urine 5.75 4.8–7.5
Vaginal secretions 3.5–4.0 2.6. Lungs
The pH in the lung mass is 6.69 while the arterial blood pH is
between 7.38 and 7.43. The composition of the epithelial
the brain, while the cochlear endolymph pH is 7.5. In contrast, the lining fluid is mainly water (96%), salts, phospholipids, pro­
endolymph in the endolymphatic sac varies between 6.7 and 7.1. teins, and mucins with a pH about 6.6 in healthy indivi­
duals [21].
2.3. Tumor
Tumor tissue often has an acidic extracellular pH [17]. For
2.7. Milk
instance, in human breast cancer, the intracellular pH is 7.15 A dynamic change in human milk pH has been observed [22].
and 7.37 while the extracellular pH is 6.99 to 6.8. Note in The mean pH decreased from 7.45 for colostrum to 7.04 dur­
normal tissue, the intracellular pH is about 7 and the ing the second week of lactation. Thereafter, the pH of milk
4 L. GAOHUA ET AL.

remained between 7 and 7.1 until 3 months postpartum and will define the drug ionization, which affects the drug PK and
then increased gradually to 7.4 by 10 months. PD (either the free or the ionized, or both, to drive the PD
effects).
2.8. Species differences
4. pH effects on physicochemical properties
There exist species differences in physiological pH. It was
observed, for example, that gastric pH in the beagle dog was 4.1. Crosstalk between pKa and pH
significantly higher than that in human (1.8 vs. 1.1) and the
The Henderson-Hasselbalch equation below is the backbone
fasting intestinal pH in dog was also consistently higher than
of pKa-pH crosstalk.
that in human (7.3 vs. 6) [23]. McConnell and coworkers mea­
� �
sured the mouse and rat gastrointestinal pH and reported the ½proton acceptor�
pH ¼ pKa þ log
stomach pH was 3.0 (fed) and 4.0 (fasted) for mouse, 3.2 (fed) ½proton donor�
and 3.9 (fasted) for rat [24]. In contrast to dogs, the mean
intestinal pH in rodents was lower than that in man (pH< 5.2 Based on the Henderson-Hasselbalch equation, the fraction
in the mouse; pH< 6.6 in the rat). Thence, the difference of unionized or ionized is only defined by the pH of the drug-
ionization needs to be considered when translating the PBPK dissolving medium and the pKa of the drug. In other words,
models from preclinical species to human. the degree of ionization of a drug in its solution depends
upon only two factors, namely the pH of the solution and
the pKa of the drug [4]. The pKa is a constant for any ionizable
3. Distribution of pKa of marketed drugs drugs and thus, if the pH of the solution is controlled, as in the
Lombardo et al. compiled a dataset consisting of 1,352 com­ case of the human physiological system, the degree of ioniza­
pounds including mostly marketed drugs with a wide range of tion depends only on the nature of the ionizable drug being
fundamental physicochemical characteristics [25]. Among investigated.
those compounds with calculated pKa values, there were 457 When 50% of an acidic or basic drug is ionized, its environ­
neutral, 313 anionic, 472 cationic, and 97 zwitterionic com­ mental pH equals to its pKa. For the drugs with pKa’s between
pounds. Clearly, most of the compounds are ionizable. 6 and 8, they are always in equilibrium with at least 10% of its
For a set of 661 drugs containing a single ionizable group ionic form at the physiologically important pH of 7. After
from the ChEMBL-18 database, there were 237 acids and 424 passing from a pH value which is one unit on either side of
bases spanning a pKa range from 2 to 10 [10]. It was observed the pKa to a pH value one unit on the other side, a given drug
that the acidic compounds completely ionized at pH greater is converted 10-fold from a substantially ionized to
than 7, and basic compounds exhibit the same trend of dis­ a substantially non-ionized condition. For instance, for
sociation for pH less than 7. Interestingly, the authors noticed a basic drug with pKa of 7, changing the environmental pH
the number of neutral compounds reported in the Journal of from 6 to 8 results in its unionized from 9% to 91%. Hence,
Medicinal Chemistry decreased while the numbers of both even quite small changes in the pH value of a solution can
acids and bases were consistently increasing in the past two produce a very large alteration in the degree of ionization of
decades. a drug, thus potentially resulting in a significantly modified
The overall distribution of acid and base pKa values from biological effect.
another set of 582 drugs was also reviewed [26]. Those com­ As aforementioned, most drugs are weak acids or weak
pounds were grouped into CNS and non-CNS drugs to inves­ bases (with pKa around the physiological pH) and they exist
tigate where differences existed. The distribution of pKa values in physiological fluids, tissues, and organs at an equilibrium
for single acids differed between CNS and non-CNS substances between unionized and ionized forms. In general, only neutral
with only one CNS compound having an acid pKa below 6.1. (free, unbound unionized) drug molecules can passively pene­
The basic CNS substances also showed a marked cut-off with trate the physiological membrane, and at equilibrium, the
no compounds having a pKa above 10.5. In a contemporary concentrations of the unbound unionized species are equal
set of 907 drugs, it was found that 64% of these compounds on both sides. Increased accumulation of drug on the side of
contained an ionizable group [27]. Within this group of ioniz­ a membrane where pH favors greater ionization of drug has
able compounds, 34% contained a single basic group while led to the pH-partition hypothesis. The majority of evidence
only 20% contained a single acidic functional group. Oral supporting the pH-partition hypothesis was from studies of
drugs showed that 78.6% contained an ionizable group, gastrointestinal absorption, renal excretion, and gastric secre­
while only 11.9% were neutral [27,28]. tion of drugs.
According to Manallack, the pKa distributions of drugs
seem to be determined by two main factors. The first is related 4.2. Lipophilicity
to the nature and frequency of occurrence of the functional
groups that are commonly observed in pharmaceuticals and Lipophilicity is a molecular parameter describing the partition
the typical range of pKa values they span. The other factor equilibrium of a solute between water and an immiscible
concerns the biological targets these compounds are organic solvent such as octanol. Lipophilicity is usually mea­
designed to hit. This makes sense as the environmental pH sured by the partition coefficient (P) when the compound is in
EXPERT OPINION ON DRUG METABOLISM & TOXICOLOGY 5

its unionized form. Partition coefficients are obtained as the levels around 7, and phenytoin precipitates due to its poor
logarithm (logP) of the ratio of concentrations at equilibrium water solubility. A clinical implication is that a fall in plasma
between the organic solvent and the aqueous water [29]. levels may occur when patients are switched from oral to intra­
The most commonly used measure of lipophilicity is the muscular administration. The drop is caused by slower absorp­
n-octanol/water partition coefficient (logPo:w), which when tion from the injection site, as compared to oral administration.
written in this way refers to the neutral form of the solute. As a result, it may then take several days for the dose to be fully
Frequently, the apparent distribution coefficient (logD) at absorbed after intramuscular injection [31].
a specific pH (such as 7.4) is used for ionizable drugs. Unlike Most of published data have clearly shown that weakly
logPo:w, which is valid only for neutral species, logDpH refers basic drugs, which have low solubility at high pH, could
to a pH-dependent mixture of all neutral and electrical species have impaired absorption in patients with high gastric pH,
present at that pH. Therefore, logPo:w is always larger than thus leading to reduced and variable bioavailability. As this
logDpH as the ionized species tends to stay in water rather reduction in exposure can lead to significant loss of efficacy, it
than in the organic solvent. As a rule of thumb, if logDpH is is imperative to understand the behavior of the compound as
measured at different pHs (such as 2, 4, 7, and 10), the highest a function of stomach pH to inform of any risk of bioavail­
logDpH is generally used as the surrogate of logPo:w. ability loss in clinical studies [32].
From logPo:w and logDpH profile, the drug class (acidic, One example is gefitinib as the first-line of non-small cell
basic, or neutral) and its pKa can be estimated using the lung cancer (NSCLC) treatment. It was shown that the clinical
ionization fraction defined by the Henderson-Hasselbalch outcomes (progression-free survival and overall survival) are
equation. significantly affected when gastric acid-suppressing agents
(such as Histamine-1 receptor antagonists or proton pump
4.3. Solubility inhibitors) are co-administered [33,34]. Gefitinib is a weak
base with a pH-dependent solubility, which leads to its higher
In general, the salt form of an organic compound is more solubility when the pH is less than 5. Thus, co-administration
soluble in the aqueous medium than its unionized form, of gastric acid suppressants would reduce gefitinib solubility.
thus switching from free base to salt form is one of the Recently, FDA published a draft guidance on the evaluation
effective approaches to the enhancement of the dissolution of gastric pH dependent drug interactions with acid-reducing
rate of drugs with low solubility. agents (ARA), mainly considering the pH-dependent solubility
Indeed, an analysis of the aqueous solubility data based on of weakly basic drugs [35].
37,100 compounds extracted from PubChem indicated that
the ionizable compounds (acids, bases, and zwitterions) have
better solubility than neutrals and that the basic compounds 5. pH effects on drug absorption
may even possess a slight (yet statistically significant) advan­ 5.1. Absorption process
tage in solubility over acidic molecules [10]. The latter is
possibly due to an overall greater extent of ionization for Oral drug delivery is still the preferred and most convenient
most bases at physiological pH. Solubility affects almost all route for systemic administration, due to ease of patient com­
aspects of drug behavior and characterization from in vitro pliance, cost-effectiveness, and flexibility in designing dosage
assay reproducibility to in vivo oral bioavailability and forms. However, oral drug absorption is a complex process
formulatability. that is influenced by many factors, such as the properties of
As ionization depends on the pH and the pKa, the aqueous the drug, its dosage form, and the physiological state of the GI
solubility of an ionizable drug is pH dependent. For an acidic tract of patients. Figure 3 shows the drug absorption process.
compound, the solubility increases when the pH in the solu­ After oral administration, the solid drug form (such as tablet or
tion is increased as more drug molecules are ionized in alka­ capsule) disintegrates into granules and then further deaggre­
line conditions. In contrast, solubility of a basic compound is gates as fine particles. The drug leaves the dosage formulation
increased when the pH of the solution is decreased. For (tablet, capsule, granule, or fine particle) and dissolves into the
ampholyte/zwitterion compounds, the drug molecules are aqueous digestive fluids. All ionizable drug molecules will be
normally non-charged at the physiological pH, resulting in ionized, as determined by its chemical pKa and the surrounding
a low solubility. An increase in solubility of ampholyte and physiological pH following the aforementioned Henderson-
zwitterion compounds can be observed when the solution pH Hasselbalch equation. Both the ionized and unionized molecules
is significantly shifting high or low. Hence, pH-solubility pro­ can bind to the fluid components, which include bile micelles and
files can be utilized to identify the type of ionization (acidic or food contents, a phenomena known as solubilization. All the
basic) and the intrinsic solubility of free base. bound and unbound, ionized and unionized drug molecules can
Local pH has a significant impact on drug PK through pH- diffuse and reach the GI mucosal membrane which is covered by
dependent solubility. For instance, phenytoin sodium is formu­ a thin unstirred water layer with a relatively constant pH
lated as an injection at pH 12 adjusted with sodium hydroxide (Figures 3, 4).
[30]. On injection of phenytoin sodium, the interstitial fluid Based on the classic pH-partition hypothesis, only the free
quickly reduces the pH of the injectable from 12 to normal (unbound unionized) molecules can passively penetrate across
6 L. GAOHUA ET AL.

Figure 3. Drug absorption process.

the apical membrane of enterocyte. However, the unbound gut was monitored by surface location using a directional
ionized drug molecules can also penetrate across the cell detector. The stomach pH values of all subjects were highly
membrane of enterocyte, which has −40 mV membrane acidic (range 1 ~ 2.5). The mean pH in the proximal small
potential, based on the Nernst-Planck equation [36]. intestine was 6.6 for the first hour of intestinal recording, and
Depending on pKa and pH, drug molecules may be fully the mean pH in the terminal ileum was 7.5. There was a sharp
ionized within the unstirred water layer on the apical surface fall in pH with a mean of 6.4 as the capsule passed into the
of enterocytes. As the permeation is defined by the product of cecum in all subjects. The pH then rose progressively from the
the concentration gradient and the permeability-surface area, right to the left colon with a final mean value of 7.
if the permeability of ionized species is reasonably good, A separate study had a similar observation that the intra­
although lower compared to the unionized counterpart, luminal pH was rapidly changed from highly acid in the sto­
there will still be reasonable absorption of the fully ionized mach to about pH 6 in the duodenum [38]. The pH gradually
drug molecules when there is a high gradient of unbound increased in the small intestine from pH 6 to about pH 7.4 in
ionized concentration across the enterocyte membrane. the terminal ileum. The pH dropped to 5.7 in the cecum and
Simultaneously, the absorption process itself is influenced then gradually increased to pH 6.7 in the rectum.
by the nature and surface area of the GI mucosal membrane, Avdeef compiled the GI pH from various resources for
which varies from the stomach to the rectum, and by the fasted and fed states, where the terminal ileum has the high­
physicochemical properties of the luminal content, when est pH of 8 in both fasted and fed states and the rectal pH
drug in solution or in solid form moves along the GI tract ranges from 5 to 8 (Figure 4) [39].
with the luminal content at a variable speed. It is generally
assumed that drug absorption in the stomach is minimal due 5.2.2. Gut radial pH
to its small surface area; however, there is significant absorp­ While the luminal pH could be significantly affected by the
tion from the stomach for acidic drugs as their free (unionized) luminal component, the mucus layer between the luminal
concentrations are high in the acidic stomach [5,6]. fluid and the microvilli cells has a relatively constant pH,
The following sections will discuss how physiological pH ranging 5.2 ~ 6.2 (Figure 4) [39]. The thickness of the unstirred
and chemical pKa interact with each other to define the water layer is about 30 ~ 100 μm in vivo and 300 ~ 700 μm in
absorption process. isolated tissue, significantly larger than the sizes of drug mole­
cules, micelles, and some fine particles. Previously, it was
assumed that the particles did not exist in the mucus layer,
5.2. pH in GI
as shown in Figure 3. However, nano- and/or micro-size drug
5.2.1. Gut axial pH particles could enter the mucus layer, resulting in a significant
Luminal pH changes dramatically along the whole GI. Evans decrease of the distance of drug molecule diffusion across the
et al. measured the GI pH of 66 healthy subjects using a pH unstirred water layer before getting absorbed, known as ‘par­
sensitive radio-telemetry capsule passing freely through the GI ticle drifting effect’ [40].
tract [37]. Signals were recorded with a portable solid-state Since the mucus layer has a local pH independent on the
receiver and recording system, enabling unconstrained mea­ luminal pH, ionization within the mucus will be different from
surements with normal ambulatory activities for up to that in the luminal fluid. Passive permeations of unbound
48 hours during normal GI transit. Capsule position in the unionized and/or unbound ionized drug molecules across
EXPERT OPINION ON DRUG METABOLISM & TOXICOLOGY 7

Figure 4. GI physiological pH.

the apical membrane are determined by the mucus environ­ increased to 6.5 in the rectum. In 11 healthy adults, the median
ment (binding and ionization), which is relatively stable, rather pH was 7 in duodenum and dropped to pH 6.3 in the proximal
than the luminal environment which is changing dynamically. part, but rose to 7.3 in the distal part of the small intestine [46].
In the premature infants, the pH level of gastric contents
5.2.3. Varying GI pH under fasting condition was 3.35. It increased to 7 after feed­
5.2.3.1. GI pH and gender. Gender difference in the small ing with Pre-Gallia formula or human milk and then decreased
intestinal pH was observed [41]. Gastric fluids are generally from 7 to a minimum pH value of 3.35 [47]. These pH values
more acidic in males than females (pH 1.92 in males vs. pH suggest either a high buffering capacity of the meal or that
2.59 in females), and the basal and maximal flow of gastric the gastric acid secretion levels are not very high in premature
fluid and acid secretion are both higher in men [42]. In infants and contribute very little to the overall volume of the
a sample of 365 healthy subjects, the average basal stomach gastric contents during the digestion period.
pH was 2.16 in men and 2.79 in women [43]. Such gender
difference in stomach pH is mostly due to the gender differ­ 5.2.3.3. GI pH and food. Food intake stimulates GI secretions
ence in stomach juice secretion. Healthy women secret sig­ including hormones and bile salts, which lower gastric pH, delay
nificantly less basal and pentagastrin-stimulated acid than stomach emptying and increase GI transit time. While feeding
men, with a median 24 hour integrated acidity of 485 mmol/ can introduce a dynamic change to stomach pH within minutes
h vs. 842 mmol/h [31]. to hours, diet itself can change GI pH in the long term. In fact,
subjects on a vegan or vegetarian diet showed significantly
5.2.3.2. GI pH and age. Russell et al. measured gastric and lower stool pH than those control subjects consuming ordinary
duodenal pH in 79 healthy, elderly men and women (averaged omnivorous diet (6.3 for vegan, 6.6 for vegetarian vs. 6.8 for
75 years old) under both fasted and fed conditions [44]. It was control) [48]. However, limited data on stomach pH has been
observed in nine subjects that their fasted stomach pH was reported for vegan or vegetarian diets.
greater than 5 and that in five of these subjects the stomach
pH remained greater than 5 postprandially. The gastric pH 5.2.3.4. GI pH and drug. Many weakly basic drugs exhibit
values of approximately 50% of elderly subjects decreased pH-dependent solubility and are therefore subject to drug–
more slowly than in young subjects after administration of drug interaction mediated by stomach pH that can be mod­
a large meal. Therefore, the drugs and/or formulations that ified by the co-administration of gastric acid reducing agents
need an acidic gastric environment for dissolution and/or (ARAs), such as Histamine H1-receptor antagonists and proton
release may be poorly assimilated in elder subjects. pump inhibitors (PPIs) [49]. These interactions can sometimes
In 12 healthy children aged 8 ~ 14 years, the mean fasting be clinically relevant – exposures of weakly basic drugs can be
stomach pH was 1.5 and the duodenal pH was 6.4 [45]. The pH reduced by an order of magnitude resulting in loss of efficacy.
gradually rose down the small intestine, reaching a peak value For instance, the exposure of erlotinib decreased 50% when it
of 7.4 in the distal ileum, then dropped to 5.9 in the cecum but was co-administered with omeprazole [50]. Erlotinib labeling
8 L. GAOHUA ET AL.

recommends patients avoiding concomitant use of erlotinib The pH change along the GI tract from pH of 1.5 at the
and PPIs if possible. FDA recently released its draft guidance stomach to pH 7 at the colon has been used in colonic drug
about drug–drug interaction mediated by stomach-pH chan­ delivery, using pH-responsive polymer coatings. For instance,
ging due to ARAs [35]. Eudragit S (a polymethacrylic acid methyl methacrylate ester
Compared to the Histamin H2-receptor antagonists, PPIs copolymer with a dissolution threshold of pH 7) is introduced
represent the gold-standard therapy in the gastric acid- to effect release the drug substance from the formulation as
related disorders [51]. Omeprazole (a weakly basic compound, the intestinal pH increases distally to a maximum at the ileo­
pKa ~4) is a gastric anti-secretory agent. At low stomach pH, cecal junction [59].
greater than 99% omeprazole is ionized. In a study with eight
subjects treated with 40 mg QD for 5 days, the gastric pH rose
5.4. pH and drug instability
from 1.9 to 5 with morning dosage or 4.5 with evening
dosage, corresponding to a greater than 99% reduction in Drug stability investigation is one of the most important compo­
hydrogen ion activity [52]. It was observed that the relative nents of drug quality evaluation. Hydrolysis is one common
bioavailability of omeprazole increased 2 ~ 3-fold from day 1 degradation reaction and source of instability. Water, either as
to day 5 of treatment, suggesting that increased absorption of a liquid solvent or from moisture in the air, contacts the pharma­
this acid-labile drug occurs with increasing inhibition of acid ceutical dosage forms to some degree, thence the potential for
secretion. PPIs, acting as perpetrators by gastric acid suppres­ this degradation pathway exists for most drugs and excipients. In
sion, can inhibit drug oral absorption, for instance, the expo­ a study of the hydrolytic behavior of pravastatin at different pHs
sures of atazanavir dropped 95% when it was co-administered and temperatures, it was observed that degradation of pravas­
with lansoprazole [53]. tatin was dramatically influenced by both pH and temperature
Due to the serious clinical consequences of pH-dependent [60]. At a high temperature of 80°C, the degradation constant
drug–drug interaction, prediction methods would be crucial to decreased between pH 3 ~ 9 although it increased between pH 9
assess the drug–drug interaction risk associated with and 12. Therefore, it is reasonable to conclude that, within the
a particular victim drug. Recent publications have presented physiological pH range, the stability of pravastatin increases
various methods applied across a range of drugs to simulate concomitantly with the increasing of pH. It should be noted
gastric pH-dependent drug–drug interaction [54] as well as however that the result may not be generalizable to other drugs.
food effects [55,56]. Oral absorption modeling is becoming
more impactful to respond to regulatory questions.
5.5. pH and intestinal microflora
The intestinal microflora is a complex ecosystem consisting of
5.3. pH-triggered disintegration/deaggregation
over 400 bacterial species within the lower bowel. It was shown
Disintegration is a physical process related to the mechanical that under in vitro conditions, pH was the strongest driver of
breakdown of a tablet into fine particles/granules before dis­ microbial community structure and function and microbial and
solving (Figure 3). Disintegration and its following disaggrega­ metabolic interactions among pH-sensitive fermentative species
tion help to increase the surface area of particles in contact [61]. The balance between bicarbonate alkalinity and formation
with the medium, therefore increasing the dissolution rate of fatty acids by fermentation determined the pH, which con­
from the solid form [57]. trolled microbial community structure.
In an in vitro study about the influence of the medium pH
on the disintegration time of commercially available phos­
5.6. Modeling of dissolution rate
phate binder formulations, it was observed that the pH sig­
nificantly affected in vitro disintegration in the majority of In most oral absorption models, the intrinsic dissolution rate of
phosphate binders tested [58]. In this study, seven calcium a drug is described by the classical Noyes-Whitney equation or
carbonate tablet formulations, two calcium acetate tablet for­ Nernst-Brunner equation [62]. However, these equations were
mulations, three aluminum hydroxide tablet formulations, and developed for a planar surface where the concentration gra­
three aluminum hydroxide capsule formulations were tested dient in the diffusion layer is linear at steady state. To allow for
at three pH conditions (distilled water, pH 5.1; simulated gas­ the fact that the concentration gradient around a spherical
tric fluid, pH 1.5; and simulated intestinal fluid, pH 7.5) using particle is not linear under pseudo-steady-state conditions,
the USP standard disintegration apparatus at the time. It was Wang and Flanagan proposed a general solution of the diffu­
shown that 9 out of the 15 products tested were sensitive to sion layer model (DLM) for dissolution from mono-dispersed
changes in pH, showing significant differences in disintegra­ spherical particles under sink and non-sink conditions, and the
tion time. The shortest disintegration time of 1 min was model was validated experimentally [63,64].
observed in all three of the aluminum hydroxide tablets at When applying the Wang and Flanagan DLM to describe
all pH tested, whilst the longest disintegration time of 99 min the drug dissolution in GI luminal fluid, the ionization of
was observed in one of the 3 aluminum hydroxide capsules in a drug within the luminal fluid plays two significant roles.
the distilled water at pH 5.1. This study reveals how the First, the driving force for drug dissolution in the DLM is the
disintegration time can vary depending on the immersion concentration gradient between the aqueous solubility at the
medium pH. surface of particles and the bulk concentration in the luminal
EXPERT OPINION ON DRUG METABOLISM & TOXICOLOGY 9

fluid. The aqueous solubility at the surface of particles 6. pH effects on drug distribution
depends on the pKa of the drug and the micro-
6.1. Distribution process
environmental pH at the surface of particles, following the
aforementioned Henderson-Hasselbalch equations. Secondly, After being absorbed from the GI luminal fluid into the sys­
the effective diffusion coefficient used in the DLM to define temic circulation, the process of drug movement from blood
the dissolution rate is also affected by the ionization, because into various tissue and organ compartments like the interstitial
the binding of ionized drugs to the bile micelle is supposed to space and the intracellular space is called drug distribution,
be lower than their unionized counterpart. Therefore, if the another important process resulting in exposure of the target
drug is highly ionized, the diffusion coefficient may be lower. organ to the drug. Various factors influence the dynamics of
The total impact of ionization on dissolution rate is case-by- drug distribution, including cardiac output, regional blood
case, as the solubility and the diffusion coefficient show dif­ flow, alterations in capillary permeability, plasma protein bind­
ferent directions when pH is changing. ing, and affinity of tissue protein binding. Active transporters
and metabolic enzymes may contribute to the drug distribu­
tion and disposition within the tissue/organ as well.
5.7. Passive permeability
Drug distribution within plasma, tissue, and organ is pre­
Based on the pH-partition hypothesis, only free (unbound dominantly determined by the chemical pKa and the physio­
unionized) drug can passively penetrate across the cell mem­ logical pH if the drug is not a substrate of the transporter or
brane. It is generally accepted that the unbound unionized enzyme. As mentioned previously, the pH within the human
molecule has higher permeability (normally 3 ~ 4 log unit) body is heterogeneous, with the plasma pH slightly varying
than its ionized counterpart [65]. Hence, to get across most around 7.4, while local pH at various tissues can vary signifi­
membranes, the drug must be relatively nonpolar. cantly (Figure 2). Within a solid tissue such as the liver, the
Although the penetration of unbound unionized species extracellular pH is almost the same as that of the perfusing
would be higher than the unbound ionized species, the plasma; however, the intracellular pH is lower, around 7, and
unbound ionized drug molecules do penetrate across the the lymphatic pH is between 7 and 7.4. Due to the ionization
cell membrane of enterocytes, which have −40 mV mem­ difference across the cell membrane and based on the pH-
brane potential, following the Nernst-Planck equation. partition hypothesis, it is expected that cations accumulate
Ionized molecules at physiological pH tend to interact with intracellularly and anions extracellularly. At steady state, how­
negatively charged lipid membranes, generally resulting in ever, the free (unbound unionized) concentration should be in
a low permeability in an order of neutrals, bases, zwitter­ equilibrium across any semi-permeable membranes, regard­
ions, and then acids [10]. Increasing lipophilicity may less of the type of tissue and organ, cell or its organelle.
increase permeability, especially for ionizable drugs, i.e. Notably, different tissues and fluids within the human body
acids, bases, and zwitterions. have different pH, and these are mostly time-dependent as
The penetration of unbound unionized drug molecules well (as discussed in Section 2). These pH differences are
across cell membranes, specifically the apical membrane of important especially for tissue-targeted drug delivery. It is
the enterocytes in the GI tract, is also dependent on the critical to consider the ionization and pH gradient when devel­
gradient of unbound unionized concentration across the oping PBPK models for tissue-targeted drug delivery.
membrane. Due to the constant pH gradient between the
unstirred water layer and the intracellular fluid of the enter­
6.2. Ionization and tissue distribution
ocytes (~6.5 vs. ~7), the weakly basic drugs are more ionizable
in the unstirred water layer than in the intracellular fluid. The Only minor differences were observed between the binding
difference results in about 0.5 log unit (~3-fold) high ionized of acidic, basic, neutral, and zwitterionic substances to var­
fraction of drug in the unstirred water layer, even though the ious tissues, although basic drugs tend to be stored in
unbound unionized drug concentration in the unstirred water tissues with a pH that is lower than their pKa values (e.g.
layer is in equilibrium with that in the enterocyte. Therefore, lung) [10]. For tissues with a lower pH, a greater fraction of
the unstirred water layer with a constant pH of 6.5 could be ionized basic species can electrostatically interact with the
a natural barrier of the enterocytes. negatively charged cell constituents (i.e. membrane phos­
Notably, the drug penetration across the apical membrane pholipids and/or acidic cellular compartments in which
of enterocyte is also dependent on the drug binding in the accumulation may occur). Binding of bases was stronger to
unstirred water layer and in the enterocyte. If the drug is hepatocytes compared to neutral or acidic molecules for
significantly bound in the enterocyte such as by lysosomal a given lipophilicity, although tissue distribution of basic
trapping, it is expected that the unbound unionized fraction drugs is highly dependent on lipophilicity. Only minor dif­
will be lower in the enterocyte than that in the unstirred water ferences in binding to brain tissue were found between
layer with less binding proteins and therefore resulting in an acidic, basic, neutral, and zwitterionic substances, which
accumulation of drug within the enterocyte. Resultantly, low were more influenced by lipophilicity. This is in contrast to
Cmax and large Tmax will be observed in the plasma concentra­ overall CNS penetration for which basic substances showed
tion-time profile, though the drug has been well penetrated greater brain exposures than neutrals followed by zwitter­
from the gut lumen into the enterocyte. ions and then acids [26].
10 L. GAOHUA ET AL.

6.3. Tissue/plasma partition coefficient (Kp) prediction a low Vss for acidic compounds, compared to the classic
Rodgers and Rowland’s method.
The pH of intracellular fluid is typically slightly lower than that
Acids generally have lower Vss values, also due to higher
of the extracellular fluid (approximately 7 vs. 7.4) (Figure 5).
plasma protein binding. Basic molecules often have higher Vss
This drives a slight pH partitioning in accordance with the pKa
values (due to lower protein binding) favoring increased half-
of the drug and the Henderson-Hasselbalch relationship.
life. These compounds can show significant inter-organ varia­
Likewise, mammalian cells maintain an electric potential
tion [10].
(approximately −40 mV) which can drive electrochemical par­
titioning in accordance with the degree of ionization and
relative intrinsic permeability of the neutral and ionized 6.4. Ion trapping
species. Another example to add an extra complex mechanism to Kp
Various approaches have been applied to define the tissue/ prediction is to consider the ionization difference in the intra­
plasma partition coefficient (Kp) and the steady-state distribu­ cellular organelles and the intracellular plasma due to their
tion volume (Vss), two key parameters used in PBPK modeling different pH (Figure 5). A typical phenomenon is known as
and simulation. The mechanistic model of Kp prediction was lysosomal trapping.
first developed by Poulin and Theil based on the equilibrium Lysosomal trapping of ions has been considered as one of
of unbound concentration between plasma and tissue, with­ the major mechanisms for high Vss of basic compounds [65].
out consideration of ionization difference between plasma Both GastroPlusTM and Simcyp® have incorporated the lysoso­
and tissue [66]. The model was later expanded by Rodgers mal trapping into their mechanistic models to predict Kp and
and Rowland to incorporate the ionization difference in the Vss for ionizable compounds.
intracellular vs. extracellular space, following the pH-partition As evidenced by the hepatic accumulation of chloroquine,
hypothesis [67–69]. Depending on the intracellular pH, which the acidic interior of lysosomes is able to trap the positively
is about half log unit lower than plasma pH, positively charged charged cations [71]. However, it is important to consider that
cations accumulate intracellularly. Furthermore, the intracellu­ the accumulation of basic compounds into lysosomes would
lar acidic phospholipid is negatively charged; therefore, an raise the intra-lysosomal pH; therefore, the capability of lyso­
electrochemical interaction with the positively charged cations somal accumulation of these basic compounds will be
introduces more accumulation of basic compounds within reduced when those cations accumulate within lysosomes.
a cell. In the method of Rodgers and Rowland, the ratio of Although greater pH and electrochemical partitioning can
unbound unionized concentration in the intracellular vs. that occur in subcellular compartments where larger gradients
in the extracellular space, known as Kpuu, is 1 at the steady- exist (lysosome pH 5, mitochondrial membrane potential of
state, in accordance with pH-partition hypothesis. approximately −170 mV), the impact of these effects to overall
Additional mechanisms can be incorporated in the predic­ Vss can often be negated by the dominant relative effect of
tion of Kp and Vss. One example is the impact of electrostatic the plasma unbound fraction (fup) and the relatively small
potential of cell membranes (Figure 5). A negatively charged aqueous volumes of these intracellular organelle spaces [72].
cell membrane will benefit the passive permeation of posi­
tively charged cations into the intracellular space [70]. In this
case, Kpuu is larger than 1 for basic compounds and less than 1 6.5. pH and fup
for acidic compounds. In other words, the cell membrane General trends for plasma protein binding are well estab­
potential and the passive permeability of ionized compounds lished, in order from greatest to least: acids, neutrals, zwitter­
would result in a high Vss predicted for basic compounds but ions, and then bases [10]. Albumin and alpha-1-acid

Figure 5. Physiological pH in tissue compartments.


EXPERT OPINION ON DRUG METABOLISM & TOXICOLOGY 11

glycoprotein are major drug-binding proteins in serum. was observed that changes in plasma protein binding due to
Albumin is an alkaline protein, so acidic and neutral drugs pH increases, which often occurred during the equilibrium
primarily bind to it while basic drugs are predominantly dialysis experiment, were not species-specific but drug-
bound to alpha1-acid glycoprotein, and this binding is often specific, though cationic compounds had a higher likelihood
high and is very pH sensitive. Since acids typically have higher of displaying pH-dependent binding. The study emphasized
plasma protein binding and thus lower distribution volume, the importance of effectively controlling pH in plasma protein
they may require high metabolic stability in order to obtain binding studies in vitro and in vivo. Similarly, it is essential in
acceptable half-lives. Increasing lipophilicity may also increase the PBPK modeling and simulation to consider a calibration of
protein binding regardless of ionization class. the unbound fraction fup measured in the in vitro system of
In general, the unbound concentration in the whole blood a pH different from the physiological pH.
or in the plasma is the most important determinant of PK/PD
of a drug. Many PK analyses require unbound blood or plasma
6.6. pH and blood/plasma partition
concentrations, including prediction of clearance, volume of
distribution, drug–drug interactions, tissue-specific uptake, Cytosol of the red blood cells (RBCs) has a pH range between
and so on [73]. 7.1 and 7.3, slightly lower than the plasma pH 7.4. Such a small
Plasma protein binding, i.e. the free fraction fup, is pH- difference in pH is accountable for the passage of slightly
dependent. Based on literature data, Hinderling and positively charged electrolytic drug into the RBCs from plasma
Hartmann have analyzed the relationship between unbound as acidic conditions favor the ionization of basic drugs [75].
fraction and pH [74]. The basic drugs showed a consistent Like other cells, ionization within RBCs prevents the transloca­
behavior with the unbound fraction decreasing with increas­ tion of ionized species from the cell, based on the pH-partition
ing pH, and a large majority of the acids displayed pH- hypothesis and low membrane permeability of ionized spe­
dependent binding as well – the binding increased for some cies. It was observed in an in vitro study that for both quinine
and decreased for others with increasing pH. The authors also and cryptolepine, a decrease in pH caused an increase in RBCs
observed a large difference in the pH sensitivity of the plasma partitioning, due to various mechanisms related to permeabil­
protein binding among individual compounds. The unbound ity, solubility, lipophilicity, and stability. For acidosis, the dose
fraction in plasma for some bases and acids increased up to for both quinine and cryptolepine should be decreased to
136% and 95%, respectively, at pH values seen in severe prevent accumulation in the RBCs [74].
acidemia or alkemia. These changes in unbound fraction are
unlikely to have significant effect on unbound plasma drug
6.7. pH and transporters
concentration. However, they could have clinical relevance
with respect to the interpretation of changes in total plasma The expression and activity of transporter P-glycoprotein
drug concentration. Since only unbound drugs have pharma­ (P-gp) plays a role in the multidrug resistance of tumors [76].
cological effects and can be hepatic metabolized or renal Growing solid tumors often show pronounced hypoxia or
excreted, these changes can be clinically relevant with drugs extracellular acidosis. The impact of an acidic environment
having narrow therapeutic windows. on the expression and activity of P-gp and on the cytotoxicity
In practice, measuring the total concentration of drug in of chemotherapeutic agents has been studied, using prostate
plasma is more convenient than measuring its unbound con­ carcinoma cells which were exposed to an acidic extracellular
centration. To arrive at unbound plasma concentration from environment (pH 6.6) for up to 24 h. It was observed that P-gp
the measured total plasma concentration, a separate in vitro activity more than doubled after 3 ~ 6 h of incubation in acidic
determination of the plasma protein binding of a drug is medium, whereas cellular P-gp expression remained constant,
usually carried out in serum or in plasma in the laboratory. indicating that increased transport rate was the result of func­
The plasma PK profiles and parameters are mathematically tional modulation. In parallel, the cytotoxic efficacy of daunor­
adjusted using the in vitro measured unbound fraction. In ubicin showed pronounced reduction at low pH, an effect that
vitro measurement of plasma protein binding or the drug was reversible upon co-incubation with a P-gp inhibitor.
unbound fraction is known to be affected by protein, drug, Indeed, a reduction of intracellular Ca2+ concentration by
free fatty acid concentrations, lipoprotein partitioning, tem­ 35% under acidic conditions induced a higher transport rate
perature, pH, and the presence or absence of other drugs or of P-gp, an effect comparable to that found on inhibition of
displacing agents within plasma samples. Experimental errors protein kinase C. These data indicate that P-gp activity is
in the determination of unbound fraction caused by lack of increased by acidic pH presumably as a result of lowered
adequate pH control in in vitro assays will have unpredictable intracellular Ca2+ levels and inhibition of protein kinase
impacts on the PK calculations. C. These findings may explain the reduced cytotoxicity of
Konchansky and colleagues examined the effect of variable chemotherapeutic agents in hypoxic/acidic tumors.
pH on measured values of fup during equilibrium dialysis Organic anion-transporting polypeptide (OATP) 1A2 is
experiments using a diverse set of 55 drugs [73]. The ratio of expressed on the apical sides of intestinal and renal epithelial
the unbound fraction at pH 7.4 (10% CO2) vs. pH 8.7 (air) was cells and is considered to be involved in the intestinal absorp­
more than two-fold for 40% of the 55 compounds tested. Only tion and renal reabsorption of drugs [77]. The pH-dependency
one of the 55 compounds tested had a ratio less than 0.9. It of transport kinetics of human OATP1A2 was observed in
12 L. GAOHUA ET AL.

HEK293 cells. Indeed, OATP1A2 exhibited bimodal saturation a wide variety of nucleophilic compounds containing
kinetics at pH 6.3 and 7.4. Compared with that seen at pH 6.3 a nitrogen, sulfur, phosphorous, or selenium atom [85,86].
(5.62 mM), the Km value of the high-affinity site was ~8-fold Basic amines tend to be FMO substrates although there is
higher at pH 7.4 (43.2 mM). Therefore, transport properties of some overlap with CYPs (e.g. CYP2D6), whereas non-basic
OATP1A2 under lower pH conditions, such as those found in compounds tend to be oxidized by CYPs, with the degree of
the microenvironments of the small intestinal mucosa and N-substitution also playing a distinct role. Tertiary amines are
distal tubules, may differ from those seen under neutral pH generally FMO substrates in contrast to primary amines that
conditions. are generally better CYP substrates; secondary amines are less
clear-cut [10]. In the case of N-oxidation, when a molecule
possesses more than one basic nitrogen atom, the preferred
7. pH effects on drug metabolism
site is generally the most basic center.
7.1. Ionization and enzyme substrate
A major group of drug-metabolizing enzymes is the microso­ 7.2. pH and enzyme activity
mal cytochrome P450 (CYP450) family. CYP3A4, the most
Various environmental factors can affect the rate of enzyme-
abundant P450 enzyme in the liver, is responsible for meta­
catalyzed reactions through reversible or irreversible changes
bolizing approximately 40 ~ 50% of the drugs used in clinical
in the protein structure. Most enzymes have a characteristic
practice. Among the FDA-approved small molecule drugs
optimum pH at which the velocity of the catalyzed reaction is
(2005 − 2016), Hu et al. found that neutral and basic com­
maximal, and above and below which the velocity
pounds with molecular weight (MW) ≥ 360 g/mol tend to be
declines [87].
primarily metabolized by CYP3A4, whereas acidic compounds
As the pH changes, the ionization of groups both at the
with MW < 360 g/mol are most likely to be primarily metabo­
enzyme’s active site and on the substrate can alter, influencing
lized by other CYP enzymes [78].
the rate of binding of the drug molecule to the active site of
CYP2D6, which is expressed not only in the liver but also in
the enzyme. These effects are often reversible. Noteworthy,
the brain, is responsible for the metabolism of the second
the optimum pH of an enzyme might not be identical to that
highest number of drugs metabolized by P450 enzymes after
of its normal intracellular surroundings. In other words, the
CYP3A4. CYP2D6 tends to oxidize substrates that contain
local pH can exert a controlling influence on enzyme activity.
a protonated basic nitrogen at physiological pH and planar
The pH optimum of cytochrome P-450 reactions usually is
aromatic ring [79].
reported to be at 7.4 [88]. However, Jensen and Dalgaard
CYP2C9, which is expressed in a number of tissues through­
studied the metabolism of M1-muscarinic agonist Lu 25–109
out the body including GI tract, prefers neutral and acidic
in human liver microsomes [89]. The authors observed several
substrates, and the majority of substrates are weakly acidic
fold increases of the metabolites mediated by CYP1A2 at pH
compounds, although CYP2C9 also catalyzes the
8.5 compared to pH 7.4. This phenomenon was also reported
N-demethylation of a number of basic drugs (e.g. amitripty­
for cyclosporin A and CYP3A4, for benzphetamine and CYP2B/
line, fluoxetine, and zopiclone) [80]. CYP2C9 and CYP2C19 are
3A4 in human liver microsomes [90], and for testosterone and
structurally similar, but CYP2C19 does not have the same
CYP3A4 in rat liver microsomes [91].
affinity for acids as that found for CYP2C9 [10].
A general characteristic of FMO forms is that maximal
Monoamine oxidases (MAOs) are flavin-containing enzymes
enzyme activity is typically observed for a pH range of
that are responsible for catalyzing the deamination of bio­
8.5 ~ 11 [92–97]. This is substantially higher than the pH
genic and xenobiotic amines. MAO exists in two isoforms in
used for the determination of activity for other microsomal
most tissues, namely MAO-A and MAO-B. The substrates of
enzymes such as P450 between 7.2 and 7.6 [88]. Human FMO5
MAOs range from weakly basic to highly basic (primary
showed a broader range and greater functional activity from
amines, also some secondary and tertiary amines). MAO-A
pH 6 to 11, and its pH dependence profile for functional
preferentially metabolizes bulkier amines, while MAO-B prefers
activity differed significantly from that of other FMO enzymes
to metabolize smaller amines.
[98]. Taniguchi-Takizawa et al. used benzydamine
Flavin-containing monooxygenases (FMOs) are present
N-oxygenation as an index for FMO activity and found that
widely in both hepatic and extrahepatic tissues including the
high benzydamine N-oxygenation activities of recombinant
liver, kidney, lungs, skin, brain, and other tissues, with the
hFMO1 and hFMO3 and human kidney microsomes were
highest concentration occurring in liver, kidney, and lungs
observed at pH 8.4 compared to pH 7.4 [99]. Similarly, the
[81,82]. Multiple FMO forms have been identified in most
pH optimum of hFMO1 and hFMO3 for the oxidation of thia­
mammalian species, including humans, with the gene family
cetazone was found to be pH ~9 [100]. Krueger et al. reported
consisting of five functionally active members (FMO1-5) that
hFMO2.1 having a maximal activity at pH 9.5, and authors also
exhibit at least 80% amino acid identity for orthologous forms
confirmed its presence in lung tissue from a heterozygous
(i.e. human FMO1 and rat FMO1) and 51 to 58% identity for
individual by Western analysis [95].
homologous forms (i.e. human FMO1 and human FMO3)
Many psychoactive drugs (e.g. imipramine) are substrates
[83,84]. FMOs catalyze the NADPH-dependent oxidation of
of the brain FMO, so the FMO-mediated metabolism of these
EXPERT OPINION ON DRUG METABOLISM & TOXICOLOGY 13

drugs might contribute to local pharmacodynamic modulation with pH ranging 6.5 ~ 8, whereas SULT1A3 displayed a distinct
within the human brain. Bhagwat et al. reported the optimum optimum at pH 9.5. SULT1C4 showed a pH optimum ranging
pH for N-oxidation of imipramine was found to be 8.5 in 7 ~ 8.5. The differential pH-dependence profiles of these three
human brain microsomes, so it is conceivable that FMOs are SULTs may reflect their distinct substrate binding and/or cat­
significantly involved in the local metabolism and modulation alytic residues involved in mediating the sulfation of
of pharmacological and/or toxic effects of psychoactive acetaminophen.
drugs [101].
For the FMOs in animals, the purified FMOs from mouse
7.3. Hepatic intracellular pH
kidney and liver showed pH optima of 9.2 [102]. The activity of
dog FMO1 showed a 4.5-fold increase in S-oxidase activity as Depending on the cell and tissue type, pH of intracellular fluid
the incubation pH increased from 7.6 to 9.0 [103]. Krueger can vary in the range of 6.8–7.2, with a relatively consistent
et al. reported that monkey FMO demonstrated a maximal intracellular fluid pH of 7 in the hepatocytes in humans [108].
activity at pH 9.5 [95]. Physiologically, pH of the intracellular fluid can be affected by
Glucuronosyl transferases are a family of enzymes that many factors, such as metabolic inhibition, temperature, trans­
catalyze the transfer of glucuronic acid from UDP-glucuronic port of weakly basic/acidic compounds into the cell [109].
acid to acceptor molecules containing hydroxyl, phenol, car­ Intracellular pH can also be directly modulated by extra­
boxylic acid, thiol, or amine groups. UDP- cellular pH. It was observed that intracellular pH of cultured
glucuronosyltransferases (UGTs) are expressed almost ubiqui­ hepatocytes in bicarbonate:CO2 medium was relatively con­
tously throughout the body, with high expression in the liver stant at 6.85–7.05 over the external pH range of 7–8 [110].
and the intestines [104]. Below an external pH of 7, intracellular pH fell below 6.8.
Chang et al. explored the impact of incubation pH on the Varying the pressure of CO2 between 15 and 40 mmHg did
glucuronidation of raloxifene, mycophenolic acid, and ezeti­ not alter the extracellular versus intracellular pH curve.
mibe, which are basic, acidic, and neutral compounds, respec­ An extracellular pH of 7.4 and an intracellular pH of 7 result
tively [105]. The glucuronidation was examined in human liver in a half-log unit gradient of pH between plasma and the
microsomal incubations by monitoring the production of glu­ intracellular environment, which theoretically would lead to
curonide metabolites at a pH range of 5.4 ~ 9.4. Compared to a pKa-dependent change in the ratio between intracellular
physiological pH, unbound intrinsic clearance was 11- and 12- and extracellular concentrations. The liver is the major route
fold higher at pH 9.4 for raloxifene 4′-glucuronide and ralox­ of drug elimination. Given changing ionization based on the
ifene 6-glucuronide, respectively, whereas a 10-fold increase chemical properties (such as compound class and its pKa’s) of
was observed at pH 5.4 for mycophenolic acid glucuronide. In the drug and the differences in the physiological environment
contrast, ezetimibe glucuronidation did not vary as the pH (such as pH and proteins) of intracellular and extracellular fluid
deviated from 7.4. One explanation for the phenomena may across the membrane of hepatocytes, it has been proposed
be the ionization of key amino acids making up the catalytic that hepatic models for clearance should be modified to
site is sensitive to shifting incubation pH; thus, it affects sub­ account for changes in hepatic intracellular concentra­
strate affinity for UGTs. Secondly, this may be due to mem­ tion [111].
brane permeability changes, as adjusting the incubation pH
can neutralize the charge on the acidic and basic compounds.
7.4. Well-stirred liver model
Neutralizing compounds would make them more readily able
to move across biological membranes. Shifting the incubation In the approach of in vitro-in vivo extrapolation (IVIVE) which
pH may also enhance membrane permeability by changing empowers the translational PBPK modeling and simulation,
the substrate binding to membrane channels. Glucuronidation the in vivo hepatic clearance is derived from the in vitro intrin­
in microsomes often under predicts in vivo glucuronidation, sic clearance. In vitro clearance is determined from various
the higher rate of glucuronidation may give better in vivo assays such as hepatocytes, microsomes, S9 or recombinants,
correlation. The results suggest that not only the assay condi­ with various liver tissue models such as the well-stirred, par­
tions of microsomal glucuronidation should be considered allel-tube, dispersion, or multiple-compartmental liver models
while conducting PBPK modeling on the drugs cleared by [112,113]. Although the well-stirred liver model is the most
glucuronidation, but also the impact of pH on extrahepatic used model, often the IVIVE with the well-stirred liver model
metabolism should be investigated. under-predicts the in vivo human hepatic clearance. Among
Sulfate conjugation as catalyzed by the cytosolic sulfotrans­ various approaches to improve the IVIVE clearance prediction,
ferases (SULTs) is known to be involved in the biotransforma­ correction of ionization has shown the most success, which
tion of a variety of drugs. There are thirteen SULTs that are was based on the observation that pH difference exists in the
classified into four distinct gene families, designated as SULT1, extracellular and intracellular fluid (7.4 and 7, respectively)
SULT2, SULT4, and SULT6 [106]. Yamamoto et al. studied pH [111]. An ionization factor FI, defined as the ratio of the union­
dependence of the major acetaminophen SULTs, SULT1A1, ized fractions in plasma and that in intracellular fluid, has been
SULT1A3, and SULT1C4 [107]. The pH-dependence of the sul­ introduced to the traditional well-stirred liver model. It was
fation of acetaminophen differed considerably among the demonstrated that accounting for this ionization factor FI may
three human SULTs. SULT1A1 showed a high level of activity yield the IVIVE-predicted hepatic clearance up to 6.3-fold
14 L. GAOHUA ET AL.

greater than that obtained using the traditional equation for kidney stone formation. Manissorn and coworkers used arti­
the strong diprotic basic compounds, and up to 6.3-fold smal­ ficial urine to systematically evaluated effects of pH (ranging
ler for the strong diprotic acidic compounds. For triprotic acids 4 ~ 8) on kidney stone formation, including calcium oxalate
and bases, the difference could be as much as 15-fold. It has (CaOx) crystallization, crystal-cell adhesion, crystal internali­
been demonstrated that the model performance of predicting zation into renal tubular cells, and binding of apical mem­
both clearance and drug–drug interactions can be significantly brane proteins to the crystals [118]. Microscopic
improved after considering the ionization correction in the examination revealed that CaOx monohydrate (COM) was
PBPK model [108,114]. crystallized with greatest size, number, and total mass at
pH 4 and least crystallized at pH 8, whereas CaOx dihydrate
(COD or weddellite) was crystallized with the vice versa
8. pH effects on drug excretion
order. The greatest degree of crystal-cell adhesion was
8.1. Urine pH modification and renal excretion observed at the most acidic pH and least at the most
basic pH. Crystal internalization into renal tubular cells was
On average, urine pH is close to 6.3. The extremes of urine pH
maximal at the neutral pH 7. No significant difference was
can be 4.5 and 7.5 under forced acidification and alkaliniza­
observed in binding capacity of the crystals to apical mem­
tion, respectively. These extremes contrast with the narrow
brane proteins at different pH. The authors concluded that
range of plasma pH of 7.3–7.5, thus a large pH gradient may
the acidic urine pH may promote CaOx kidney stone forma­
exist between plasma and urine. Urine pH can be altered by
tion, whereas the basic urine pH (i.e. by alkalinization) may
diet, drugs, and the clinical state of a patient. In addition, urine
help to prevent CaOx kidney stone disease.
pH varies through the day.
While only the unbound drug in the renal perfusing
Renal clearance itself is a complicated PK parameter con­
blood can be filtrated through the glomerulus, the
sisting of filtration, secretion, and reabsorption. Both acids and
unbound fraction in blood or plasma is pH-dependent, as
bases were subject to significantly greater renal clearance than
aforementioned. Therefore, the renal clearance depends on
neutral or zwitterionic molecules [10]. The effect of urine pH
the systemic plasma pH. The more the unbound drug, the
on renal excretion and systemic disposition has been observed
more the renal filtration. Compared to the systemic plasma
for many drugs [115]. For ionizable drugs, mostly being weak
pH, however, it is the renal tubular pH that directly and
acids and weak bases, urine pH is one of the key parameters
significantly affects the drug secretion and reabsorption
affecting reabsorption and therefore drug PK, PD, and toxi­
(Figure 6). To explore the interplay of various mechanisms
city [116].
involved in renal excretion, mechanistic PBPK models have
Changing the urine pH will directly affect the drug ioniza­
been developed for kidney in the last decade [119,120].
tion in the renal tubules; therefore, the passive secretion and
reabsorption between tubular urine and tubular cells resul­
tantly influence the renal clearance and systemic exposure of 8.3. Modeling of renal excretion
drugs. Urine pH effect on drug disposition has not been well To the best of our knowledge, the mechanistic kidney model
characterized in humans. (Mech-KiM) implemented in the Simcyp® simulator was the
Memantine, a weak base with a pKa of 10.27, is predomi­ first PBPK model to integrate drug characteristics to renal
nantly excreted unchanged via the kidneys. It has been physiology in predicting the contributions of glomerular filtra­
demonstrated that memantine plasma concentrations depend tion, active and passive secretion, active and passive reabsorp­
on urine pH [117]. Alkaline urine pH results in a reduced renal tion, and metabolism to renal elimination and excretion [119].
excretion and renal clearance compared to acidic urine pH. The Mech-KiM model has been used by industry and regula­
The reduced renal clearance at alkaline urine pH can be tors in the PBPK modeling of various market drugs, such as
explained by pH-dependent tubular reabsorption under oseltamivir carboxylate, cidofovir, cefuroxime, and peme­
these alkaline conditions because the ratio of unionized mem­ trexed [121,122].
antine in alkaline urine (e.g. pH 8) is considerably higher Huang and Isoherranen also developed a 35-
(0.005) than in acidic urine (e.g. pH 5), where the ratio of compartmental kidney model to simulate the renal clearance,
unionized drug is very low (0.000005). Under acidic conditions, including filtration and pH-dependent passive reabsorption,
tubular reabsorption seems to be unlikely, and tubular secre­ from in vitro permeability data [120]. The model was verified
tion must be considered. As a result, the renal clearance of using 46 compounds, including neutrals, acids, bases, and
memantine at acidic pH will exceed the expected glomerular zwitterions. The feasibility of incorporating active secretion
filtration rate. and pH-dependent bidirectional passive diffusion into the
model was demonstrated using para-aminohippuric acid,
8.2. pH in kidney and tubule cimetidine, memantine, and salicylic acid. The developed
model enabled the simulation of renal clearance from in vitro
Though less relevant to PBPK modeling and simulation, permeability data, with predicted renal clearance within two­
urine pH is one of the major factors that can modulate fold of observed for 87% of the drugs assessed.
EXPERT OPINION ON DRUG METABOLISM & TOXICOLOGY 15

Figure 6. Physiological pH in renal tubule.

9. pH and toxicity/safety bioaccumulation were higher for acids at lower pH values,


whereas the opposite was true for bases. The effect of pH
9.1. pH and pharmacodynamics
was most pronounced when pH-pKa was in the range of
In cellular assays, the zwitterions tend to be more active than −1 ~ 3 for acids, and −3 ~ 1 for bases. It was observed that
basic and neutral compounds while the acids generally tend the factor by which toxicity and bioaccumulation changed
to be less active, though there are still a significant number of with pH was correlated with the lipophilicity of the com­
acids with potent cellular activity [10]. pound. For both acids and bases, the correlation was posi­
tive, albeit significant only for acids. Limited information was
available for amphoteric and zwitterionic compounds.
9.2. pH and receptor binding
It has been demonstrated in Chinese hamster ovary (CHO) cells 9.4. Ionization and safety
that the affinity and binding kinetics of Histamine H1 receptor
antagonists (levocetirizine, fexofenadine, and desloratadine) are 9.4.1. Selectivity
differently affected by an acidic environment (pH 5.8) that can be Basic compounds are generally known to bind to a variety of
encountered during inflammation status [18]. Simulations based receptors and transporters. Highly basic positively charged
upon clinical data indicated that, albeit apparently modest in compounds are generally more prone to promiscuity than
amplitude, these changes in binding properties could have sig­ neutral, zwitterionic, or acidic compounds. Increased lipophili­
nificant effects on the PD of antihistamines in the clinical situa­ city also generally contributes to overall promiscuity [10].
tion. Indeed, under acidic conditions and at their therapeutically
active doses, the levocetirizine binding profile would be 9.4.2. Cellular distribution
enhanced, that of fexofenadine would remain largely unchanged Represented by lysosomal trapping, bases often become
and that of desloratadine would be impaired. sequestered in acidic organelles of many different cell types
and may thereby contribute to various toxicities.
Lipophilic basic molecules, especially cationic amphiphiles,
9.3. pH and toxicity can cause phospholipidosis by distributing to membranes and
pH of exposure mediums can influence the toxicity and lysosomes and can also be influenced by overall compound
bioaccumulation of ionizable compounds [123]. An ionizable lipophilicity.
compound is generally more toxic and more bioaccumula­ Both strongly acidic and basic drugs have been associated
tive when in the neutral state. The change in lipophilicity with mitochondrial dysfunction. Strong lipophilic acids tend to
when a neutral compound becomes ionized, the electrical have the greatest propensity to cause uncoupling of oxidative
attraction, and the ion trap have been identified as 3 major phosphorylation, while basic drugs can accumulate into the
processes to explain the behavior of ionizable compounds acidic mitochondrial cytosolic space in tissues such as liver
with changing pH. Literature data indicated that toxicity and and pancreas.
16 L. GAOHUA ET AL.

10. Further considerations 10.2. pH and pKa manipulation


10.1. Impacts of uncertain pKa or pH 10.2.1. Change pH for better PK
The pHs of the injectables are always adjusted for better
10.1.1. Uncertainty in pH and pKa measurements
performance. Phenytoin sodium is a typical example. Dilantin
As one of the major physicochemical parameters of drugs,
(phenytoin sodium injection, USP) is a ready-mixed solution of
various methods have been developed to measure pKa [124].
phenytoin sodium in a vehicle containing 40% propylene
The physiological pH can also be measured using various
glycol and 10% alcohol in water for injection, adjusted to pH
methods, non-invasively and continuously. For example,
12 with sodium hydroxide [30]. A fall in plasma levels may
a telemetric SmartPill® capsule system can be used to gener­
occur when switching from oral to intramuscular administra­
ate the intra-gastric pH profile over time, while the capsule is
tion. Such a drop in exposure is possibly due to slower absorp­
transited from stomach to colon [125].
tion from the precipitates at the injection site, due to the poor
pH is the negative of the base 10 logarithm of the hydro­
water solubility of phenytoin.
gen ion concentration in the medium and pKa is the pH of the
medium where the drug is 50% ionized. Since both pH and
10.2.2. Change pKa for better PK
pKa are logarithmic measurement, one-unit difference in the
Fluorination, namely introducing fluorine into a drug mole­
values of pH or pKa corresponds to 10-fold difference in
cule, can alter electron distribution, which can impact on
ionization. Slight uncertainty in the pH and pKa measurements
the pKa, dipole moment and even the chemical reactivity
may have significant impacts on the calculation of ionization;
and stability of neighboring functional groups [129].
therefore, the aqueous solubility, passive permeation, and
Fluorine can reduce the basicity of compounds, which
further drug ADMET. This is true especially in the case where
can improve bioavailability due to better membrane per­
the physiological pH and the chemical pKa are comparable. If
meation of the compound. Recently, Velcicky et al.
the difference between pH and pKa is big enough, the impact
reported their approach based on specifically placed fluor­
of uncertain pH and pKa may not be such significant, as the
ine within a pyrrole-based MK2 (MAP-activated protein
ionization is reaching its limitations (either fully ionized or fully
kinase 2) inhibitor scaffold for manipulation of its physico­
unionized).
chemical and ADME properties [130]. While preserving tar­
get potency, the novel fluoro-derivatives showed greatly
10.1.2. Temperature matters
improved permeability as well as enhanced solubility and
While the human body has a representative temperature dif­
reduced in vivo clearance, leading to significantly increased
ferent from the environmental room temperature, when
oral exposure.
applying pKa measured in in vitro systems under the room
temperature 20 ~ 25°C to in silico PBPK model of the human
body with a representative temperature 37°C, one needs to 10.3. Extension of Henderson-Hasselbalch equation
consider the difference in temperature, as pKa itself is tem­
10.3.1. Common ion effect and salt effect
perature dependent [124,126]. In general, pKa of an acidic
Henderson-Hasselbalch equation is the basis to define the
compound or a basic compound will increase or decrease
drug ionization within a medium. Changing the medium pH
and then the acidic or basic drug is more ionizable at
shifts the equilibrium of the fraction of drug that is ionized
a specific pH and more soluble when temperature is increased.
and unionized. For instance, more basic drug molecules are
For instance, the pKa of fentanyl, a weak base, decreased from
ionized when the environmental pH is decreased by adding
8.6 to 7.89 as the temperature was increased from 15°C to
hydrochloric acid (HCl), resulting in high total concentration
47.5°C [126]. Such data indicated the necessity to calibrate the
in the medium, corresponding to enhanced solubility due to
solubility measured at room temperature 20 ~ 25°C to that at
acidification of the medium. However, due to the common
37°C before applying to PBPK model.
ion effect, a decrease of solubility for a basic salt is frequently
observed when adding extremely high HCl to adjust the
10.1.3. Circadian rhythm of physiological pH
medium pH in the pH-solubility experiments. The same is
The physiological pH in plasma and several fluid/tissue/organs
true for an acidic salt when using sodium hydroxide to adjust
is tightly regulated within narrow ranges. GI is a typical case
the medium pH. Common ions (Cl− and Na+) added to the
where pH is changing tremendously, depending on location,
medium in general shift the acid-base equilibrium to the
feeding, age, gender, and drug itself and so on. Plasma pH has
reactant’s side causing precipitation, a mechanism used to
a circadian rhythm with a peak in the afternoon (e.g. 3 PM)
purify the compound in drug discovery and development.
and a trough in the middle-night (e.g. 3 AM) [127]. Urinary pH
The common ion effect generally decreases the solubility of
has a diurnal pattern as well [128]. To the best of our knowl­
the salt form (but not the free base, which does not have
edge, however, this detailed pH variation remains to be incor­
common ion).
porated into PBPK model, even though it might be one of the
In contrast, adding a new, uncommon ion to the medium
critical physiological parameters that contribute to the intra-
can facilitate strong ionic interactions with the products. This
individual (or inter-occasion) variabilities.
EXPERT OPINION ON DRUG METABOLISM & TOXICOLOGY 17

‘salt effect’ therefore shifts from the reactants to the ionized 10.4.2. New mechanisms
products, therefore increasing solubility of the drug. Both the There are several additional developments beyond the tradi­
common ion effect and salt effect are examples of using Le tional pH-partition hypothesis in the PBPK modeling area. One
Chatelier’s principle to modulate solubility. concerns the permeation of bound drugs across cell mem­
brane. Tsao and colleagues have constructed a ‘facilitated-
10.3.2. Critical micellization concentration and pH dissociation’ model in which the interaction of the ligand-
Oral absorption is a pathway of primary interest in many PBPK albumin complex with the cell surface enhanced the dissocia­
applications. Bile acids are amphipathic molecules formed tion of that complex to provide unbound ligand for uptake
from cholesterol metabolism that play an important role in into the hepatocytes [133]. The ‘facilitated-dissociation’ model
oral absorption. Bile salt solubilization is one of the major enabled an accurate description of the enhanced update
mechanisms contributing to solubility enhancement of lipo­ clearance in the in vitro hepatocyte experiment with various
philic weakly basic compounds in the proximal GI, but not in amounts of albumin added [134,135]. Poulin and coworkers
the distal GI. It is well known that the unconjugated bile acids have summarized several recent literatures supporting the
are insoluble at neutral pH while their conjugated salts are protein-facilitated uptake mechanism(s) and further discussed
highly soluble. From the Henderson-Hasselbalch equation, the the interplay among the effects of pH gradient, protein-
solubility of both the unconjugated and conjugated forms facilitated uptake, metabolism, active transport, and permea­
rises exponentially with increasing pH. As the concentration tion limitation in cells [136,137].
of the conjugated anions reaches the critical micellization The second development of the pH-partition hypothesis is
concentration (CMC), micelle formation occurs and solubility the passive permeation of unbound ionized species across the
becomes practically unlimited [131]. Various bile acids and bile cell membrane. As aforementioned, the cell membrane is nor­
salts have different CMCs, which depend on the pKa of bile mally negatively charged; therefore, the positive charged
acids and bile salts as well as the environmental pH. In gen­ cations can penetrate the cell membrane because of the elec­
eral, CMC increases with the increasing pH until it reaches the tronic potential, following the Nernst-Planck equation. Various
critical micellization pH, above which the CMC is pH- PBPK models have considered the passive permeation of ionized
independent. Lipophilic basic drugs tend to precipitate when species across cell membrane [36,70,138,139]. To the best of our
the dissolved species move from the strong acidic stomach to knowledge, however, such a penetration of unbound ionized
the weakly acidic jejunum and ileum. However, the bile salts drug molecules across the apical membrane of enterocytes
are secreted from the bile duct into the duodenum, transit remains to be considered in mechanistic oral absorption model.
through the small intestine and are re-absorbed at the end of
ileum. Bile salts allow the lipophilic drugs to be dissolved
10.5. Feedback from PD to PK in PBPK modeling and
within micelles, preventing precipitation. It is therefore impor­
simulation
tant to define the correct CMC for each bile salt at each GI
segment, enabling a proper description of solubilization, dis­ As mentioned, pH in the GI tract has a significant impact on
solution and precipitation in the PBPK modeling and the oral absorption for all ionizable drugs administrated orally.
simulation. However, some drugs like PPIs (such as omeprazole) are mod­
ulating the GI pH, while their PKs are depending on GI pH
itself. For this reason, a feedback loop from PD to PK is
10.4. Extension of pH-partition hypothesis
required in the PBPK/PD modeling and simulation to model
10.4.1. Transporters PK and PD simultaneously.
The current review on pH-pKa crosstalk, its impacts on drug Another example where a PD-PK feedback loop is impor­
ADMET and its incorporation in the PBPK modeling and simu­ tant for quantitative accuracy is for those bile acid transporter
lation work is mostly based on the traditional pH-partition inhibitors (iBATs, such as elobixibat), where the solubility of
hypothesis established in the 1950s for the passive permea­ the iBATs in the GI tract is enhanced by bile salts while the bile
tion process [5–8]. However, Kell recently proposed ‘no trans­ acid re-absorption is inhibited by iBATs.
porter means no transport’ based on recent insights from Some urinary pH modifiers (such as sodium bicarbonate) can
experiments using targeted CRISPR-Cas9 deletion of solute change the urinary pH, therefore affecting the renal section and/
carries genes [132]. In the perspective article from Kell, drugs or re-absorption, which may impact on the PK of drug itself.
pass through the cell membrane by hitchhiking on membrane
transporters may be substantially underappreciated, implying
10.6. Biologics
‘passive diffusion’ through membrane bilayers is negligible.
Such a novel hypothesis may fundamentally change the cur­ The neonatal Fc receptor (FcRn) is characterized by its high
rent approach for PBPK modeling and simulation and the affinity to bind albumin and immunoglobulin G (IgG) at low
paradigm of drug discovery and development. Given the dra­ pH. It plays an important role in regulating the half-lives of
matic implications and outstanding quantitative questions the both plasma albumin and IgG through a pH-dependent recy­
hypothesis raises, it remains challenging to entirely accept and cling mechanism [140]. The endosome is slightly more acid­
adopt at this moment. ified than the blood (pH 6 vs. 7.4). The FcRn-bound albumin
18 L. GAOHUA ET AL.

and IgG in the acidified endosome are protected from degra­ model (V.18.2) uses the total dissolved amount to define the
dation and can be recycled when these albumin and IgG meet passive oral absorption [142].
the physiological pH again on the membrane surface of The OrBiTo project, a multi-million EU-funded international
endothelial cells. The half-lives of antibodies can often be public–private partnership established in late 2012, has high­
long, around 21 days for IgG, due in substantial part to the pH- lighted multiple issues in these models that ultimately led to
dependent FcRn recycling mechanism [141]. unsatisfactory accuracy in the prediction of drug oral bioavail­
ability [143]. However, it should have been recognized that none
of the oral absorption models compared in OrBiTo project has
11. Conclusion
properly implemented the basic pH-partition hypothesis for the
Although it is not an exhaustive review, we have collated and passive process of oral absorption/exsorption. One of our con­
compiled the majority of information in the literature about cerns is, without systemic and consistent incorporation of pH-
the interaction of the physiological pH and the chemical pKa pKa crosstalk and pH-partition hypothesis in the existing oral
and its impacts on drug ionization and ADMET from the view­ absorption and PBPK models, that any conclusion and sugges­
point of PBPK modeling and simulation. Our analysis of pH- tion from the comparison of model performances may be mis­
pKa crosstalk has been focused mainly on absorption pro­ leading. We suggest the ‘industry standard’ oral absorption
cesses, where drug solubility, dissolution, and passive penetra­ models should be revisited to review their fundamental model­
tion are directly impacted by ionization. Beyond absorption, ing assumptions, as missing pH-partition hypothesis in the mod­
the pH-pKa interaction also has a significant impact on drug els should be aware to the industrial end-users while these
disposition, where the tissue/plasma partition coefficient (Kp) models are increasingly used by the pharmaceutical companies
and the ion trapping within organelles are normally predicted and regulatory agencies in drug discovery and development.
using various mechanistic models based on pH-partition Our understandings on pH-pKa crosstalk and ionization in
hypothesis. Additional impacts of pH-pKa crosstalk on meta­ defining drug ADMET are accumulated in the past centuries
bolism, excretion, and toxicity are being increasingly recog­ and will be continued in the future with efforts from different
nized. The pH-partition hypothesis has been the theoretical disciplines. As evidenced in this review, the impacts of ioniza­
basis for PBPK modeling and simulation when describing tion on oral absorption (through modifying solubility, dissolu­
these ADMET processes mathematically and mechanistically. tion, disintegration, and permeation and so on) and tissue
A better understanding of the interaction of physiological pH distribution (through modifying free fractions across the cell
and chemical pKa therefore enables a full consideration of the membrane) have been well explored, whilst the impacts of
ionization of drugs within the physiological systems, regard­ ionization on drug metabolism, transporter-mediated elimina­
less of whether modeling healthy or disease states, when tion and tissue accumulation and toxicity remain further study.
developing a PBPK model to support drug discovery and Undoubtedly, PBPK model is providing us a theoretical tool/
development. platform to integrate our understandings systemically, consis­
tently and mechanistically. Among all the impacts the pH-pKa
crosstalk has on drug ADMET, we need to identify the key
12. Expert opinion
mechanisms where pH-pKa crosstalk plays the most important
Majority of our experiences on pH-pKa crosstalk and pH- roles and to implement them into PBPK models first. In our
partition hypothesis in PBPK modeling and simulation are opinion, independent and scattered works on ionization
related to the passive processes of oral absorption and tissue impacts on distribution volume, IVIVE of drug metabolism
distribution, both of which are determined by the free con­ and drug–drug interaction prediction will benefit from
centration gradients across cell membrane. Ideally, all the a consortium by bringing all scientific thoughts together in
mechanistic oral absorption and distribution PBPK models a systemic and consistent way.
should have been based on the same assumption of pH- Looking forward, we expect more emphasis on the funda­
partition hypothesis for the passive membrane permeation. mental understanding of pH-pKa crosstalk and pH-partition
However, to the best of our knowledge, many PBPK models hypothesis in PBPK modeling and simulation, where new
have not fully incorporated the pH-pKa crosstalk and pH- aspects and functionalities are continuously being incorpo­
partition hypothesis systemically and consistently. For exam­ rated in the next 5–10 years. More confidence will be achieved
ple, the ‘industry standard’ GastroPlusTM and Simcyp® PBPK about our PBPK model and its prediction when we know our
platforms predict the tissue/plasma partition coefficient (Kp), model basic assumptions are correct.
a key parameter defining tissue concentration kinetics and
steady-state distribution, using mechanistic models that are
Acknowledgments
based on pH-partition hypothesis. However, GastroPlusTM
ACAT model (V.9.7) is using the gradient of the total luminal The views and opinions expressed here are those of the authors and do
concentration and the unbound enterocyte concentration to not necessarily reflect the official policy, practice, or position of the
organizations that the authors work. The authors acknowledge
define the passive oral absorption. In contrast, Simcyp® ADAM
EXPERT OPINION ON DRUG METABOLISM & TOXICOLOGY 19

proofreading by and scientific discussions with Dr. Brian Schmidt of BMS, 13. Abelson MB, Udell IJ, Weston JH. Normal human tear pH by direct
and Drs. Claire Beaumont, Simon Teague, and Kunal Taskar of GSK. measurement. Arch Ophthalmol. 1981 Feb;99(2):301.
14. Valentin J. Basic anatomical and physiological data for use in radi­
ological protection: reference values. A report of age- and
Declaration of interest gender-related differences in the anatomical and physiological
characteristics of reference individuals. ICRP publication 89. Ann
The authors have no relevant affiliations or financial involvement with any
ICRP. 2002;32(3–4):5–265.
organization or entity with a financial interest in or financial conflict with
15. Coles WH, Jaros PA. Dynamics of ocular surface pH. Br
the subject matter or materials discussed in the manuscript. This includes
J Ophthalmol. 1984 Aug;68(8):549–552.
employment, consultancies, honoraria, stock ownership or options, expert
16. Lang F, Vallon V, Knipper M, et al. Functional significance of chan­
testimony, grants or patents received or pending, or royalties.
nels and transporters expressed in the inner ear and kidney. Am
J Physiol Cell Physiol. 2007;293(4):C1187–C1208.
17. Gillies R, Garcia-Martin ML, Gatenby RA pH imaging: a review of pH
Reviewer disclosures
measurement methods and applications in cancers. IEEE
Peer reviewers on this manuscript have no relevant financial or other Engineering in Medicine and Biology Magazine. 2004;September
relationships to disclose. 57–64.
18. Gillard M, Chatelain P. Changes in pH differently affect the binding
properties of histamine H1 receptor antagonists. Eur J Pharmacol.
ORCID 2006 Jan;530(3):205–214.
19. Shen DD, Artru AA, Adkison KK. Principles and applicability of CSF
Lu Gaohua http://orcid.org/0000-0003-1701-492X sampling for the assessment of CNS drug delivery and
Xiusheng Miao http://orcid.org/0000-0001-7667-6910 pharmacodynamics. Adv Drug Deliv Rev. 2004 Oct;56
(12):1825–1857.
20. Gaohua L, Neuhoff S, Johnson TN, et al. Development of a
Reference permeability-limited model of the human brain and cerebrospinal
fluid (CSF) to integrate known physiological and biological knowl­
Papers of special note have been highlighted as either of interest (•) or of
edge: estimating time varying CSF drug concentrations and their
considerable interest (••) to readers.
variability using in vitro data. Drug Metab Pharmacokinet. 2016
1. Guyton AC, Hall JE. Textbook of medical physiology. 11th ed.
Jun;31(3):224–233.
Pennsylvania: W.B. Saunders Company; 2000.
21. Gaohua L, Wedagedera J, Small BG, et al. Development of
2. Redman M. Cocaine: What is the crack? A brief history of the use of
a multicompartment permeability-Limited Lung PBPK model and
Cocaine as an Anesthetic. Anesth Pain Med. 2011;1(2):95–97.
its application in predict ing pulmonary pharmacokinetics of anti­
3. Watson PJ. The mode of action of local anaesthetics. J Pharm
tuberculosis drugs. CPT Pharmacometrics Syst Pharmacol. 2015;4
Pharmacol. 1960 May;12(1):257–292.
(10):605–613.
4. Wood HCS. Ionization, pH and biological activity. Nature. 1952;170
22. Morriss FH Jr., Brewer ED, Spedale SB, et al. Relationship of human
(4337):994–996.
milk pH during course of lactation to concentrations of citrate and
5. Hogben CA, Schanker LS, Tocco DJ, et al. Absorption of drugs from
fatty acids. Pediatrics. 1986 Sep;78(3):458–464.
the stomach. II. The human. J Pharmacol Exp Ther. 1957 Aug;120
23. Lui CY, Amidon GL, Berardi RR, et al. Comparison of gastrointestinal
(4):540–545.
pH in dogs and humans: implications on the use of the beagle dog
6. Schanker LS, Shore PA, Brodie BB, et al. Absorption of drugs from
as a model for oral absorption in humans. J Pharm Sci. 1986 Mar;75
the stomach. I. The rat. J Pharmacol Exp Ther. 1957 Aug;120
(3):271–274.
(4):528–539.
24. McConnell EL, Basit AW, Murdan S. Measurements of rat and
7. Schanker LS, Tocco DJ, Brodie BB, et al. Absorption of drugs from
mouse gastrointestinal pH, fluid and lymphoid tissue, and implica­
the rat small intestine. J Pharmacol Exp Ther. 1958 May;123
tions for in-vivo experiments. J Pharm Pharmacol. 2008 Jan;60
(1):81–88.
(1):63–70.
8. Shore PA, Brodie BB, Hogben CA. The gastric secretion of drugs:
25. Lombardo F, Berellini G, Obach RS. Trend analysis of a database of
a pH partition hypothesis. J Pharmacol Exp Ther. 1957 Mar;119
intravenous pharmacokinetic parameters in humans for 1352 drug
(3):361–369.
compounds. Drug Metab Dispos. 2018 Aug;46(11):1466–1477.
9. Avdeef A. ed.. Absorption and drug development: solubility, perme­
26. Manallack DT. The pK(a) distribution of drugs: application to drug
ability, and charge. 2nd. ed. New Jersey: John Wiley & Sons, Inc. 2012.
discovery. Perspect Medicin Chem. 2007 Sep;17(1):25–38.
•• A textbook from an expert of solution chemistry with the
27. Manallack DT. The acid-base profile of a contemporary set of drugs:
current information needed to thoroughly investigate and
implications for drug discovery. SAR QSAR Environ Res. 2009 Oct;20
accurately determine a compound's pharmaceutical properties
(7–8):611–655.
and their effects on drug absorption.
28. Manallack DT, Prankerd RJ, Yuriev E, et al. The significance of acid/
10. Charifson PS, Walters WP. Acidic and basic drugs in medicinal
base properties in drug discovery. Chem Soc Rev. 2013 Jan;42
chemistry: a perspective. J Med Chem. 2014 Dec;57(23):9701–9717.
(2):485–496.
•• A review paper on potential advantages and disadvantages of
29. Liu X, Testa B, Fahr A. Lipophilicity and its relationship with passive
both acidic and basic drugs, with new analyses based on avail­
drug permeation. Pharm Res. 2011 May;28(5):962–977.
able public data on the effects of ionization state on drug meta­
30. FDA. NDA 010151 Dilantin (phenytoin sodium) injection FDA
bolism and PK (DMPK) properties including solubility,
approved labeling text dated 10/2011. 2011. p. 14.
permeability, oral bioavailability, metabolism and clearance, tis­
31. Washington N, Washington C, Wilson CG. ed.. Physiological phar­
sue distribution, volume of distribution, and protein binding.
maceutics - barriers to drug absorption. 2nd. ed. London: Taylor &
11. WHO. Laboratory manual for the examination of human semen and
Francis;2001.
sperm-cervical mucus interaction. Singapore: Cambridge University
32. Mitra A, Kesisoglou F. Impaired drug absorption due to high sto­
Press; 1999.
mach pH: a review of strategies for mitigation of such effect to
12. DuckF A. Physical properties of tissues: a comprehensive reference
enable pharmaceutical product development. Mol Pharm. 2013
book. London: Academic press; 2013.
Nov;10(11):3970–3979.
20 L. GAOHUA ET AL.

33. Fang YH, Yang YH, Hsieh MJ, et al. Concurrent proton-pump inhi­ 51. Herszenyi L, Bakucz T, Barabas L, et al. Pharmacological approach
bitors increase risk of death for lung cancer patients receiving to gastric acid suppression: past, present, and future. Dig Dis.
1st-line gefitinib treatment - a nationwide population-based 2020;38(2):104–111.
study. Cancer Manag Res. 2019;11:8539–8546. 52. Prichard PJ, Yeomans ND, Mihaly GW, et al. Omeprazole: a study of
34. Kwok WC, Man HJC, Leung Lam DC, et al. Clinical impact of Gastric its inhibition of gastric pH and oral pharmacokinetics after morning
acid suppressants use on the efficacy of Gefitinib in patients with or evening dosage. Gastroenterology. 1985 Jan;88(1 Pt 1):64–69.
advanced adenocarcinoma of the lung harboring common EGFR 53. Tomilo DL, Smith PF, Ogundele AB, et al. Inhibition of atazanavir
mutations. Clin Cancer Drugs. 2020;7(1):57–61. oral absorption by lansoprazole gastric acid suppression in healthy
35. FDA. Evaluation of Gastric pH dependent drug interactions with volunteers. Pharmacotherapy. 2006 Mar;26(3):341–346.
acid-reducing agents: study design, data analysis, and clinical impli­ 54. Mitra A, Parrott N, Miller N, et al. Prediction of pH-dependent
cations guidance for industry. 2020. [cited 2021 Apr 12]. Available drug-drug interactions for basic drugs using physiologically based
from: https://www.fda.gov/media/144026/download. biopharmaceutics modeling: industry case studies. J Pharm Sci.
• The recent regulatory guidance on gastric pH dependent drug 2020 Nov;109(3):1380–1394.
interactions with acid-reducing agents. 55. Li M, Zhao P, Pan Y, et al. Predictive performance of physiologically
36. Ghosh A, Maurer TS, Litchfield JE, et al. Towards a unified model of based pharmacokinetic models for the effect of food on oral drug
passive drug permeation II: the physiochemical determinants of absorption: current status. CPT Pharmacometrics Syst Pharmacol.
unbound tissue distribution with applications to the design of 2018 Feb;7(2):82–89.
hepatoselective Glucokinase activators. Drug Metab Dispos. 2014 56. Tistaert C, Heimbach T, Xia B, et al. Food effect projections via
Jul;42(10):1599–1610. physiologically based pharmacokinetic modeling: predictive case
• Mechanistic model for tissue distribution to account for pH and studies. J Pharm Sci. 2019 Jun;108(1):592–602.
electrochemical potential across cellular membranes, ioniza­ 57. Markl D, Zeitler JA. A review of disintegration mechanisms and
tion according to Henderson-Hasselbalch, passive permeation measurement techniques. Pharm Res. 2017 May;34(5):890–917.
of the neutral species using Fick’s law, and passive permeation 58. Stamatakis MK, Alderman JM, Meyer-Stout PJ. Influence of pH on
of the ionized species using the Nernst-Planck equation. in vitro disintegration of phosphate binders. Am J Kidney Dis. 1998
37. Evans DF, Pye G, Bramley R, et al. Measurement of gastrointestinal Nov;32(5):808–812.
pH profiles in normal ambulant human subjects. Gut. 1988 Aug;29 59. Ibekwe VC, Khela MK, Evans DF, et al. A new concept in colonic
(8):1035–1041. drug targeting: a combined pH-responsive and
38. Fallingborg J. Intraluminal pH of the human gastrointestinal tract. bacterially-triggered drug delivery technology. Aliment Pharmacol
Dan Med Bull. 1999 Jun;46(3):183–196. Ther. 2008 Oct;28(7):911–916.
39. Avdeef A. Physicochemical profiling (solubility, permeability and 60. Brain-Isasi S, Requena C, Álvarez-Lueje A. Brain-Isasi 2008 pravasta­
charge state). Curr Top Med Chem. 2001 Sep;1(4):277–351. tin stability. J Chil Chem Soc. 2008;53(4):1684–1688.
40. Sugano K. Possible reduction of effective thickness of intestinal 61. Ilhan ZE, Marcus AK, Kang DW, et al. pH-Mediated microbial and
unstirred water layer by particle drifting effect. Int J Pharm. 2010 metabolic interactions in fecal enrichment cultures. mSphere. 2017
Mar;387(1–2):103–109. May-Jun;2(3). doi:10.1128/mSphere.00047-17.
41. Wang YT, Mohammed SD, Farmer AD, et al. Regional gastrointest­ 62. Jamei M, Turner D, Yang J, et al. Population-based mechanistic
inal transit and pH studied in 215 healthy volunteers using the prediction of oral drug absorption. AAPS J. 2009 Jun;11(2):225–237.
wireless motility capsule: influence of age, gender, study country 63. Wang J, Flanagan DR. General solution for diffusion-controlled
and testing protocol. Aliment Pharmacol Ther. 2015 Sep;42 dissolution of spherical particles. 1. Theory. J Pharm Sci. 1999
(6):761–772. Jul;88(7):731–738.
42. Soldin OP, Chung SH, Mattison DR. Sex differences in drug 64. Wang J, Flanagan DR. General solution for diffusion-controlled
disposition. J Biomed Biotechnol. 2011;2011:187103. dissolution of spherical particles. 2. evaluation of experimental
43. Feldman M, Barnett C. Fasting gastric pH and its relationship to data. J Pharm Sci. 2002 Feb;91(2):534–542.
true hypochlorhydria in humans. Dig Dis Sci. 1991 Jul;36 65. Trapp S, Rosania GR, Horobin RW, et al. Quantitative modeling of
(7):866–869. selective lysosomal targeting for drug design. European Biophysics
44. Russell TL, Berardi RR, Barnett JL, et al. Upper gastrointestinal pH in Journal. 2008;37(8):1317–1328.
seventy-nine healthy, elderly, North American men and women. 66. Poulin P, Theil FP. A priori prediction of tissue: plasma partition
Pharm Res. 1993 Feb;10(2):187–196. coefficients of drugs to facilitate the use of physiologically-based
45. Fallingborg J, Christensen LA, Ingeman-Nielsen M, et al. pharmacokinetic models in drug discovery. J Pharm Sci. 2000;89
Measurement of gastrointestinal pH and regional transit times in (1):16–35.
normal children. J Pediatr Gastroenterol Nutr. 1990 Aug;11 67. Rodgers T, Leahy D, Rowland M. Physiologically based pharmaco­
(2):211–214. kinetic modeling 1: predicting the tissue distribution of
46. Fallingborg J, Christensen LA, Ingeman-Nielsen M, et al. moderate-to-strong bases. J Pharm Sci. 2005 Jun;94(6):1259–1276.
Gastrointestinal pH and transit times in healthy subjects with 68. Rodgers T, Rowland M. Physiologically based pharmacokinetic mod­
ileostomy. Aliment Pharmacol Ther. 1990 Jun;4(3):247–253. elling 2: predicting the tissue distribution of acids, very weak bases,
47. Roman C, Carriere F, Villeneuve P, et al. Quantitative and qualitative neutrals and zwitterions. J Pharm Sci. 2006 June;95(6):1238–1257.
study of gastric lipolysis in premature infants: do MCT-enriched 69. Rodgers T, Rowland M. Mechanistic approaches to volume of dis­
infant formulas improve fat digestion? Pediatr Res. 2007 Jan;61 tribution predictions: understanding the processes. Pharm Res.
(1):83–88. 2007 May;24(5):918–933.
48. Zimmer J, Lange B, Frick JS, et al. A vegan or vegetarian diet • Theoretical framework for mechanistic prediction of tissue/
substantially alters the human colonic faecal microbiota. Eur plasma partition coefficients used in PBPK models.
J Clin Nutr. 2012 Jan;66(1):53–60. 70. Gaohua L, Turner D, Fisher C, et al. A novel mechanistic approach
49. Dong Z, Li J, Wu F, et al. Application of physiologically-based to predict the steady state volume of distribution (Vss) using the
pharmacokinetic modeling to predict Gastric pH-dependent Fick-Nernst-Planck Equation. The Population Approach Group in
drug-drug interactions for weak base drugs. CPT Europe (PAGE) Meeting 2016; 2016; Lisboa, Portugal.
Pharmacometrics Syst Pharmacol. 2020 Jul;9(8):456–465. 71. MacIntyre AC, Cutler DJ. Role of lysosomes in hepatic accumulation
50. Kletzl H, Giraudon M, Ducray PS, et al. Effect of gastric pH on of chloroquine. J Pharm Sci. 1988 March;77(3):196–199.
erlotinib pharmacokinetics in healthy individuals: omeprazole and 72. Smith DA, Beaumont K, Maurer TS, et al. Volume of distribution in
ranitidine. Anticancer Drugs. 2015 Jun;26(5):565–572. drug design. J Med Chem. 2015 Aug;58(15):5691–5698.
EXPERT OPINION ON DRUG METABOLISM & TOXICOLOGY 21

73. Kochansky CJ, McMasters DR, Lu P, et al. Impact of pH on plasma 94. Krueger SK, Yueh MF, Martin SR, et al. Characterization of expressed
protein binding in equilibrium dialysis. Mol Pharm. 2008 May-Jun;5 full-length and truncated FMO2 from rhesus monkey. Drug Metab
(3):438–448. Dispos. 2001 May;29(5):693–700.
74. Hinderling PH, Hartmann D. The pH dependency of the binding of 95. Krueger SK, Martin SR, Yueh MF, et al. Identification of active
drugs to plasma proteins in man. Ther Drug Monit. 2005 Feb;27 flavin-containing monooxygenase isoform 2 in human lung and
(1):71–85. characterization of expressed protein. Drug Metab Dispos. 2002
75. Kwakye RA, Kuntworbe N, Ofori-Kwakye K. et al. Detection, quanti­ Jan;30(1):34–41.
fication, and investigation of the red blood cell partitioning of 96. McManus ME, Stupans I, Burgess W, et al. Flavin-containing mono­
cryptolepine hydrochloride. Journal of Pharmacy & oxygenase activity in human liver microsomes. Drug Metab Dispos.
Pharmacognosy Research. 2018;6(4):260–270. 1987 Mar-Apr;15(2):256–261.
76. Thews O, Gassner B, Kelleher DK, et al. Impact of extracellular 97. Rodriguez RJ, Acosta D Jr. Metabolism of ketoconazole and deace­
acidity on the activity of P-glycoprotein and the cytotoxicity of tylated ketoconazole by rat hepatic microsomes and
chemotherapeutic drugs. Neoplasia. 2006 Feb;8(2):143–152. flavin-containing monooxygenases. Drug Metab Dispos. 1997
77. Morita T, Akiyoshi T, Sato R, et al. pH-dependent transport kinetics Jun;25(6):772–777.
of the human organic anion-transporting polypeptide 1A2. Drug 98. Motika MS, Zhang J, Ralph EC, et al. pH dependence on functional
Metab Pharmacokinet. 2020 Apr;35(2):220–227. activity of human and mouse flavin-containing monooxygenase 5.
78. Hu B, Zhou X, Mohutsky MA, et al. Structure-property relationships Biochem Pharmacol. 2012 Apr;83(7):962–968.
and machine learning models for addressing CYP3A4-mediated 99. Taniguchi-Takizawa T, Shimizu M, Kume T, et al. Benzydamine
victim drug-drug interaction risk in drug discovery. Mol Pharm. N-oxygenation as an index for flavin-containing monooxygenase
2020 Sep;17(9):3600–3608. activity and benzydamine N-demethylation by cytochrome P450
79. de Groot MJ, Wakenhut F, Whitlock G, et al. Understanding CYP2D6 enzymes in liver microsomes from rats, dogs, monkeys, and
interactions. Drug Discov Today. 2009 Oct;14(19–20):964–972. humans. Drug Metab Pharmacokinet. 2015 Feb;30(1):64–69.
80. Daly AK, Rettie AE, Fowler DM, et al. Pharmacogenomics of CYP2C9: 100. Qian L, Ortiz de Montellano PR. Oxidative activation of thiaceta­
functional and clinical considerations. J Pers Med. 2017 Dec;8(1):1. zone by the Mycobacterium tuberculosis flavin monooxygenase
81. Dannan GA, Guengerich FP. Immunochemical comparison and EtaA and human FMO1 and FMO3. Chem Res Toxicol. 2006
quantitation of microsomal flavin-containing monooxygenases in Mar;19(3):443–449.
various hog, mouse, rat, rabbit, dog, and human tissues. Mol 101. Venkatesh K, Levi PE, Hodgson E. The flavin-containing monoox­
Pharmacol. 1982 Nov;22(3):787–794. ygenase of mouse kidney. A comparison with the liver enzyme.
82. Tynes RE, Philpot RM. Tissue- and species-dependent expression of Biochem Pharmacol. 1991 Sep;42(7):1411–1420.
multiple forms of mammalian microsomal flavin-containing 102. Bhagwat SV, Bhamre S, Boyd MR, et al. Cerebral metabolism of
monooxygenase. Mol Pharmacol. 1987 Jun;31(6):569–574. imipramine and a purified flavin-containing monooxygenase from
83. Hernandez D, Janmohamed A, Chandan P, et al. Organization and human brain. Neuropsychopharmacology. 1996 Aug;15(2):133–142.
evolution of the flavin-containing monooxygenase genes of 103. Stevens JC, Melton RJ, Zaya MJ, et al. Expression and characteriza­
human and mouse: identification of novel gene and pseudogene tion of functional dog flavin-containing monooxygenase 1. Mol
clusters. Pharmacogenetics. 2004 Feb;14(2):117–130. Pharmacol. 2003 Feb;63(2):271–275.
84. Lawton MP, Cashman JR, Cresteil T, et al. A nomenclature for the 104. Tukey RH, Strassburg CP. Human UDP-glucuronosyltransferases:
mammalian flavin-containing monooxygenase gene family based metabolism, expression, and disease. Annu Rev Pharmacol
on amino acid sequence identities. Arch Biochem Biophys. 1994 Toxicol. 2000;40(1):581–616.
Jan;308(1):254–257. 105. Chang JH, Yoo P, Lee T, et al. The role of pH in the glucuronidation
85. Cashman JR, Zhang J. Human flavin-containing monooxygenases. of raloxifene, mycophenolic acid and ezetimibe. Mol Pharm. 2009
Annu Rev Pharmacol Toxicol. 2006;46(1):65–100. Jul-Aug;6(4):1216–1227.
86. Krueger SK, Williams DE. Mammalian flavin-containing monooxy­ 106. Blanchard RL, Freimuth RR, Buck J, et al. A proposed nomenclature
genases: structure/function, genetic polymorphisms and role in system for the cytosolic sulfotransferase (SULT) superfamily.
drug metabolism. Pharmacol Ther. 2005 Jun;106(3):357–387. Pharmacogenetics. 2004 Mar;14(3):199–211.
87. Robinson PK. Enzymes: principles and biotechnological 107. Yamamoto A, Liu MY, Kurogi K, et al. Sulphation of acetaminophen
applications. Essays Biochem. 2015;59:1–41. by the human cytosolic sulfotransferases: a systematic analysis.
88. Sato R, Omura T, editors. Cytochrome: p-450. 1st ed. New York: J Biochem. 2015 Dec;158(6):497–504.
Academic Press; 1978. 108. Rougee LR, Mohutsky MA, Bedwell DW, et al. The impact of the
89. Jensen KG, Dalgaard L. In vitro metabolism of the M1-muscarinic hepatocyte-to-plasma pH Gradient on the prediction of hepatic
agonist 5-(2-ethyl-2H-tetrazol-5-yl)-1-methyl-1,2,3,6-tetrahydropyri­ clearance and drug-drug interactions for CYP2D6 substrates. Drug
dine by human hepatic cytochromes P-450 determined at pH 7.4 Metab Dispos. 2016 Nov;44(11):1819–1827.
and 8.5. Drug Metab Dispos. 1999 Jan;27(1):125–132. 109. Strazzabosco M, Boyer JL. Regulation of intracellular pH in the
90. Ahmed SS, Napoli KL, Strobel HW. Oxygen radical formation during hepatocyte. Mechanisms and physiological implications.
cytochrome P450-catalyzed cyclosporine metabolism in rat and J Hepatol. 1996 May;24(5):631–644.
human liver microsomes at varying hydrogen ion concentrations. 110. Pollock AS. Intracellular pH of hepatocytes in primary monolayer
Mol Cell Biochem. 1995 Oct;151(2):131–140. culture. Am J Physiol. 1984 May;246(5 Pt 2):F738–44.
91. Gemzik B, Halvorson MR, Parkinson A. Pronounced and differential 111. Berezhkovskiy LM. The corrected traditional equations for calcula­
effects of ionic strength and pH on testosterone oxidation by tion of hepatic clearance that account for the difference in drug
membrane-bound and purified forms of rat liver microsomal cyto­ ionization in extracellular and intracellular tissue water and the
chrome P-450. J Steroid Biochem. 1990 Mar;35(3–4):429–440. corresponding corrected PBPK equation. J Pharm Sci. 2011
92. Hoskins J, Shenfield G, Murray M, et al. Characterization of moclo­ Mar;100(3):1167–1183.
bemide N-oxidation in human liver microsomes. Xenobiotica. 2001 •• Ionization factor (FI) was introduced to PBPK model to
Jul;31(7):387–397. account for the pH difference in extracellular and intracellular
93. Itagaki K, Carver GT, Philpot RM. Expression and characterization of water.
a modified flavin-containing monooxygenase 4 from humans. J Biol 112. Ito K, Houston JB. Comparison of the use of liver models for
Chem. 1996 Aug;271(33):20102–20107. predicting drug clearance using in vitro kinetic data from hepatic
22 L. GAOHUA ET AL.

microsomes and isolated hepatocytes. Pharm Res. 2004 May;21 129. Shah P, Westwell AD. The role of fluorine in medicinal chemistry.
(5):785–792. J Enzyme Inhib Med Chem. 2007 Oct;22(5):527–540.
113. Watanabe T, Kusuhara H, Maeda K, et al. Physiologically based 130. Velcicky J, Schlapbach A, Heng R, et al. Modulating ADME
pharmacokinetic modeling to predict transporter-mediated clear­ Properties by Fluorination: MK2 Inhibitors with Improved Oral
ance and distribution of pravastatin in humans. J Pharmacol Exp Exposure. ACS Med Chem Lett. 2018 Apr;9(4):392–396.
Ther. 2009 Nov;328(2):652–662. 131. Hofmann AF, Mysels KJ. Bile acid solubility and precipitation
114. Rougee LRA, Mohutsky MA, Bedwell DW, et al. The impact of the in vitro and in vivo: the role of conjugation, pH, and Ca2+ ions.
hepatocyte-to-plasma pH gradient on the prediction of hepatic J Lipid Res. 1992;33(5):617–626.
clearance and drug-drug interactions for CYP2C9 and CYP3A4 132. Kell DB. Hitchhiking into the cell. Nat Chem Biol. 2020 Apr;16
substrates. Drug Metab Dispos. 2017 Jul;45(9):1008–1018. (4):367–368.
115. Huang W, Czuba LC, Isoherranen N. Mechanistic PBPK modeling of 133. Tsao SC, Sugiyama Y, Sawada Y, et al. Kinetic analysis of
urine pH effect on renal and systemic disposition of albumin-mediated uptake of warfarin by perfused rat liver.
Methamphetamine and Amphetamine. J Pharmacol Exp Ther. J Pharmacokinet Biopharm. 1988 Apr;16(2):165–181.
2020 Jun;373(3):488–501. 134. Fukuchi Y, Toshimoto K, Mori T, et al. Analysis of nonlinear phar­
116. Rowland M, Tozer TN. Clinical pharmacokinetics: concepts and macokinetics of a highly Albumin-Bound compound: contribution
applications. Pennsylvania: Lippincott Willians & Wilkins; 1995. of Albumin-Mediated hepatic uptake mechanism. J Pharm Sci. 2017
117. Freudenthaler S, Meineke I, Schreeb KH, et al. Influence of urine pH Sep;106(9):2704–2714.
and urinary flow on the renal excretion of memantine. Br J Clin 135. Miyauchi S, Masuda M, Kim SJ, et al. The phenomenon of
Pharmacol. 1998 Dec;46(6):541–546. “albumin-mediated” hepatic uptake of organic anion transport
118. Manissorn J, Fong-Ngern K, Peerapen P, et al. Systematic evaluation polypeptide substrates: prediction of the in vivo uptake clear­
for effects of urine pH on calcium oxalate crystallization, crystal-cell ance from the in vitro uptake by isolated hepatocytes using
adhesion and internalization into renal tubular cells. Sci Rep. 2017 a “facilitated-dissociation” model. Drug Metab Dispos. 2018
May;7(1):1798. Jan;3(46):259–267.
119. Neuhoff S, Gaohua L, Burt H, et al. Accounting for transporters in 136. Poulin P, Burczynski FJ, Haddad S. The role of extracellular binding
renal clearance: towards a mechanistic kidney model (Mech KiM). proteins in the cellular uptake of drugs: impact on quantitative in
In: Sugiyama Y, Steffansen Beditors, Transporters in Drug vitro-to-in vivo extrapolations of toxicity and efficacy in physiolo­
Development. AAPS Advances in the Pharmaceutical Sciences gically based pharmacokinetic-pharmacodynamic research.
Series Vol. 7 New York: Springer; 2013. 155–177. J Pharm Sci. 2016 Jul;105(2):497–508.
120. Huang W, Isoherranen N. Development of a dynamic physiologi­ • Overview of interplay among the effects of pH gradient, pro­
cally based mechanistic kidney model to predict renal clearance. tein-facilitated uptake, metabolism, active transport, and per­
CPT Pharmacometrics Syst Pharmacol. 2018 Sep;7(9):593–602. meation limitation in cells.
121. Hsu V, De LT, Vieira M, et al. Towards quantitation of the effects of 137. Poulin P, Bteich M, Haddad S. Supplemental analysis of the predic­
renal impairment and probenecid inhibition on kidney uptake and tion of hepatic clearance of binary mixtures of bisphenol A and
Efflux Transporters, using physiologically based pharmacokinetic Naproxen determined in an isolated perfused rat liver model to
modelling and simulations. Clin Pharmacokinet. 2013 Nov;53 promote the understanding of potential albumin-facilitated hepa­
(3):283–293. tic uptake mechanism. J Pharm Sci. 2017 Aug;106(11):3207–3214.
122. Posada MM, Bacon JA, Schneck KB, et al. Prediction of renal trans­ 138. Trapp S, Horobin RW. A predictive model for the selective accumu­
porter mediated drug-drug interactions for pemetrexed using phy­ lation of chemicals in tumor cells. Eur Biophys J. 2005 Oct;34
siologically based Pharmacokinetic modeling. Drug Metab Dispos. (7):959–966.
2014;43(3):325–334. 139. Chien HC, Zur AA, Maurer TS, et al. Rapid method to determine
123. Rendal C, Kusk KO, Trapp S. Optimal choice of pH for toxicity and intracellular drug concentrations in cellular uptake assays: applica­
bioaccumulation studies of ionizing organic chemicals. tion to metformin in organic cation transporter 1-transfected
Environmental toxicology and chemistry/SETAC. 2011 Nov;30 human embryonic kidney 293 cells. Drug Metab Dispos. 2016
(11):2395–2406. Mar;44(3):356–364.
124. Reijenga J, Van Hoof A, Van Loon A, et al. Development of methods for 140. Pyzik M, Rath T, Lencer WI, et al. FcRn: the architect behind the
the determination of pKa values. Anal Chem Insights. 2013;8:53–71. immune and nonimmune functions of IgG and albumin.
125. Koziolek M, Schneider F, Grimm M, et al. Intragastric pH and pressure J Immunol. 2015 May;194(10):4595–4603.
profiles after intake of the high-caloric, high-fat meal as used for food 141 Deng R, Jin F, Prabhu S, et al. Monoclonal antibodies: what are the
effect studies. J Control Release. 2015 Dec;220(Pt A):71–78. pharmacokinetic and pharmacodynamic considerations for drug
126. Thurlkill RL, Cross DA, Scholtz JM, et al. pKa of fentanyl varies with development? Expert Opin Drug Metab Toxicol. 2012 Feb;8(2):141–160.
temperature: implications for acid-base management during 142 Gaohua L, Pilla Reddy V. Passive intestinal absorption: systematic
extremes of body temperature. J Cardiothorac Vasc Anesth. 2005 comparison of 4 mechanistic oral absorption models in
Dec;19(6):759–762. GastroPlusTM and in Simcyp®. The 3rd meeting of the European
127. Latenkov VP. Diurnal rhythm of acid-base equilibrium and Network on Understanding Gastrointestinal Absorption-related
blood gas composition. Biull Eksp Biol Med. 1986 May;101 Processes (UNGAP), UNGAP spring meeting.; 2020 Feb. 2020;
(5):614–616. Ljubljana, Slovenia.
128. Cameron M, Maalouf NM, Poindexter J, et al. The diurnal 143 Ahmad A, Pepin X, Aarons L, et al. IMI - Oral biopharmaceutics tools
variation in urine acidification differs between normal indivi­ project - evaluation of bottom-up PBPK prediction success part 4:
duals and uric acid stone formers. Kidney Int. 2012 Jun;81 prediction accuracy and software comparisons with improved data
(11):1123–1130. and modelling strategies. Eur J Pharm Biopharm. 2020 Aug;156:50–63.

You might also like