You are on page 1of 23

Feature Review

Steroidogenesis: Unanswered
Questions
Walter L. Miller1,*
Until the mid-1980s studies of steroidogenesis largely depended on identifying
Trends
steroid structures and measuring steroid concentrations in body fluids. The
The principal pathways of steroido-
molecular biology revolution radically revolutionized studies of steroidogenesis genesis, the identities of the principal
with the cloning of known steroidogenic enzymes, by identifying novel factors, steroidogenic enzymes, and genetic
disorders of most of these enzymes
and delineating the genetic basis of known and newly discovered diseases. have been elucidated. Nevertheless,
Unfortunately, this dramatic success has led many young research-oriented new pathways, new enzymes, and
endocrinologists to regard steroidogenesis as a ‘solved area’. However, many new steroids have been discovered,
and require further investigation.
important and exciting questions remain, especially concerning the mecha-
nisms of cholesterol delivery to the steroidogenic machinery, the biochemistry Much remains to be learned about the
biochemistry and cell biology of several
of androgen synthesis, the regulation and biological role of adrenarche,
steroidogenic factors, notably StAR,
fetal adrenal development and involution, the roles of steroids made in 3bHSD, and P450c17.
‘extraglandular’ cells, and the search for genetic disorders. This review outlines
The cellular mechanisms of fetal adre-
some of these questions, but this list is necessarily incomplete. nal involution, adrenarche, and adre-
nopause remain unknown.
Current knowledge about steroidogenesis (see Glossary) is summarized in Box 1 and
detailed in [1]. When the Endocrine Society asked me to speak in conjunction with receiving The roles of extraglandular steroido-
the 2017 Fred Conrad Koch Lifetime Achievement Award, instead of presenting past work I genesis remain poorly understood.
tried to give a prospective lecture, outlining some unanswered questions in steroidogenesis
Despite decades of work, the genetics
that might motivate future work, and could even stimulate funding agencies. Thus this is not a of these factors remain incompletely
traditional review, but an outline of areas that have puzzled me and remain largely unanswered. understood, and genetic disorders
This is therefore a personal perspective, and is not intended to be comprehensive; other items have not been found for many steroi-
dogenic genes.
could certainly be added to this list. I have tried to emphasize newer papers, and I apologize to
colleagues whose work is only cited by inference.

The Steroidogenic Acute Regulatory Protein, StAR


Question 1: How Does StAR Work?

Importance
The steroidogenic acute regulatory protein (StAR) mediates the acute steroidogenic
response which is needed for adrenal stress responses, for the rapid production of gonadal
steroids, and possibly for other physiologic responses. Despite more than 20 years of intense
research, the mechanism of StAR action remains wholly unknown [2,3].

Background
The first and rate-limiting step in steroidogenesis is the conversion of cholesterol to pregneno-
lone by the cholesterol side-chain cleavage enzyme, P450scc (CYP11A1), a cytochrome 1
Center for Reproductive Sciences,
P450 (CYP) enzyme that lies on the inner mitochondrial membrane (IMM). The intracellular University of California, San Francisco
processing of cholesterol is complex and remains under investigation (Figure 1) [2,3]. StAR (UCSF), San Francisco, CA 94143-
triggers the flow of cholesterol from the outer mitochondrial membrane (OMM) to P450scc on 0556, USA

the IMM (Box 2). Ferguson and Garren identified a short-lived protein needed to initiate
steroidogenesis; Orme-Johnson and Stocco independently characterized it by 2D gel elec- *Correspondence:
trophoresis; then the laboratory of Douglas Stocco cloned mouse StAR [4]. The indispensable wlmlab@ucsf.edu (W.L. Miller).

Trends in Endocrinology & Metabolism, November 2017, Vol. 28, No. 11 http://dx.doi.org/10.1016/j.tem.2017.09.002 771
© 2017 Elsevier Ltd. All rights reserved.
Box 1. Principal Pathways of Steroidogenesis
Glossary
Figure incorporates pathways from all three zones of the adrenal as well as gonadal steroidogenesis (see [1] for details).
Adrenarche: the gonadotropin-
Steroidogenesis begins from cholesterol; StAR facilitates a rapid flux of cholesterol into steroidogenic mitochondria, but
independent increase in adrenal
low levels of steroidogenesis can occur in its absence. The left-hand column shows the D5 pathway in which steroids
androgen production, marked by
retain the double bond between carbon atoms 5 and 6 in the cholesterol B-ring. Mitochondrial P450scc removes the 6-
serum DHEAS. Adrenarche typically
carbon side-chain of cholesterol to yield pregnenolone, a C21 (21-carbon) steroid. Hormonally inactive D5-steroids are
begins at least 1 year before the
converted to the corresponding D4-steroids by 3b-hydroxysteroid dehydrogenase, type 2 (3bHSD2, HSD3B2 gene) in
onset of puberty, but DHEAS levels
the adrenal and gonad, or by the closely related 3bHSD1 in placenta and peripheral tissues. The adrenal zona
do not reach maximal values until
glomerulosa does not express P450c17, permitting progesterone to be 21-hydroxylated by microsomal P450c21
young adulthood.
(CYP21A2 gene). Mitochondrial P450c11AS (aldosterone synthase, CYP11B2 gene) then catalyzes 11-hydroxylase,
Aldo-keto reductases (AKRs): a
18-hydroxylase, and 18-methyl oxidase activities to yield aldosterone (Aldo). Expression of P450c17 (17-hydroxylase/
subgroup of hydroxysteroid
17,20 lyase, CYP17A1 gene) in the adrenal zona fasciculata permits synthesis of 17OH-progesterone (17OHP), which is
dehydrogenases.
converted to cortisol by P450c11b (11-hydroxylase, CYP11B1 gene). To catalyze 17,20 lyase activity, which converts
Cholesterol side-chain cleavage
C21 steroids to C19 steroids, P450c17 requires allosteric action of cytochrome b5 (b5, CYB5 gene), which is essentially
enzyme (P450scc): the enzyme that
confined to androgenic tissues (adrenal zona reticularis and testicular Leydig cells). Even with the assistance of b5,
converts cholesterol to
human P450c17 converts 17OHP to androstenedione with only 2–3% of its activity in converting 17OH-pregnenolone
pregnenolone. It is encoded by the
to DHEA, such that most human testosterone synthesis proceeds via DHEA and not via 17OHP; nevertheless, P450c17
CYP11A1 gene and is associated
from many other mammals catalyzes this step efficiently. In the testis, 17bHSD3 readily converts DHEA to andros-
with the matrix side of the inner
tenediol and androstenedione to testosterone; low levels of 17bHSD5 (AKR1C3) in the adrenal zona reticularis permit
mitochondrial membrane (IMM).
the adrenal to make small amounts of testosterone. In ovarian granulosa cells (and in some peripheral tissues, especially
Congenital adrenal hyperplasia
adipocytes) P450aro (aromatase, CYP19A1 gene) converts androstenedione to estrone and testosterone to estradiol;
(CAH): a group of genetic disorders
in estrogenic tissues (ovary, breast, fat), 17bHSD1 converts estrone to estradiol. In genital skin (and possibly in the
in adrenal cortisol synthesis that are
testis), 5a-reductase type 2 (5aRed2, SRD5A2 gene) further activates testosterone to dihydrotestosterone. Not all
caused by mutations of
intermediate steroids, pathways, and enzymes are shown.
steroidogenic enzymes. When
cortisol is low, the pituitary makes
more ACTH (adrenocorticotropic
Cholesterol hormone, corticotropin), resulting in
P450scc adrenal growth. About 95% of CAH
+ StAR P450c11β is caused by 21-hydroxylase
3βHSD2 P450c21 P450c11AS P450c11AS P450c11AS deficiency.
Pregnenolone Pregesterone DOC Corcosterone 18OH Aldo Cytochrome b5 (b5): a small,
Corcost 18 kDa heme-containing protein,
P450c17 P450c17 encoded by the CYB5 gene, that
serves diverse functions.
3βHSD2 P450c21 P450c11β
17OH-Preg 17OH-Progesterone 11-Deoxycorsol Corsol Cytochrome P450: a family of
(17OHP) proteins encoded by CYP genes that
include both microsomal and
P450c17+ b5 mitochondrial members acting as
3βHSD2 P450aro biosynthetic and xenobiotic-
DHEA Androstenedione Estrone metabolizing enzymes. The name
‘P450’ refers to spectral
17βHSD3/5 17βHSD3/5 17βHSD1 characteristics of the proteins
(absorption at 450 nm). The 57
P450aro
Androstenediol 3βHSD2 Testosterone Estradiol human P450s are about 60 kDa and
contain a heme group. Several
5αRed2 steroidogenic enzymes are P450s.
DHT Dehydroepiandrosterone (DHEA):
DHEA and its sulfate (DHEAS) are
19-carbon (C19) steroids that are
abundantly produced by the adrenal.
Figure. Pathways of Steroidogenesis. Abbreviations: 17OH-Preg, 17-hydroxypregnenolone; 18-OH Corticost, DHEA(S) designates either DHEA or
18-hydroxycorticosterone; ALDO, aldosterone; DHEA, dehydroepiandrosterone; DHT, dihydrotestosterone; DOC, DHEAS.
11-deoxycorticosterone; P450scc, cholesterol side-chain cleavage enzyme. © WL Miller. Disordered sexual development
(DSD): any situation in which the
external genitalia develop in a sex-
atypical form (formerly termed
role of StAR in steroidogenesis is highlighted by the complex and often lethal outcome of ‘ambiguous genitalia’).
Hydroxysteroid dehydrogenase
its mutations that cause congenital lipoid adrenal hyperplasia, and by similar findings in StAR- (HSD): a group of non-P450
knockout mice [2]. Studies on the biophysics and cell biology of the StAR protein [2,3] have steroidogenic enzymes that include
shown the following: (i) StAR is synthesized as a nominal 37 kDa protein that is imported into the short-chain dehydrogenases and
members of AKR family.
mitochondria, accompanied by cleavage to a 30 kDa intramitochondrial form. (ii) Maximal StAR
NCI-H295 cells: the only currently
activity requires phosphorylation of Ser195. (iii) StAR is not a major transporter of cholesterol to available immortalized human adrenal
the OMM  the StAR mitochondrial leader peptide irreversibly affixes StAR to the OMM, so that

772 Trends in Endocrinology & Metabolism, November 2017, Vol. 28, No. 11
cell line that expresses all the adrenal
steroidogenic machinery and makes
all the adrenal steroids.
LDL receptor P450c17: the microsomal P450 that
SRB1 sequentially catalyzes 17-hydroxylase
Endosome
activity and (in the presence of b5)
LAL 17,20 lyase activity. It is encoded by
CE C the CYP17A1 gene and is expressed
NPC1 NPC2
in the adrenals and gonads, but not
CC in placenta.
ACAT P450 oxidoreductase (POR): a
two-flavin protein that transfers
electrons to microsomal (but not
‘Free’ HSL
cholesterol Lipid mitochondrial) P450s, and is required
o
nov s
for their catalytic activity.
droplet
e
D thesi
Polycystic ovary syndrome
n (PCOS): a group of idiopathic,
sy Transport proteins hyperandrogenic disorders of women
u ic

characterized by increased ovarian


ul m

as m StAR Lipid
o pl c
and adrenal C19 steroid secretion,
d droplet ovarian cysts, and mild virilism.
n re  Steroidogenesis: the study of
E

steroid hormone synthesis. Many


cells and tissues (e.g., liver) can
P450scc metabolize one steroid to another,
SNARE but a cell is said to be
proteins ‘steroidogenic’ only if it can initiate
the process by expressing P450scc,
MAM which initiates steroidogenesis.
Tethering Steroidogenic acute regulatory
proteins C protein (StAR): a rapidly induced,
short-lived, 37 kDa protein that acts
on the outer mitochondrial
membrane (OMM) to stimulate the
Figure 1. Intracellular Cholesterol Trafficking. Steroidogenic cells take up circulating low-density lipoproteins (LDL) by flow of cholesterol from the OMM to
receptor-mediated endocytosis, directing cholesterol to endosomes, or high-density lipoproteins (HDL) via scavenger the IMM. At the IMM, cholesterol is
receptor B1 (SRB1). Cholesterol is also synthesized de novo in the endoplasmic reticulum (ER). In late endosomes, taken up by P450scc and converted
lysosomal acid lipase (LAL) cleaves the ester group of cholesterol esters bound by NPC2; unesterified cholesterol is then to cholesterol.
transferred to NPC1, inserted into the endosomal membrane, and exported. The MLN64/MENTHO system and the NPC
system colocalize to the same endosomes; roles for MLN64/MENTHO in cholesterol trafficking seem likely but remain
unproven. Cholesterol can be re-esterified by acyl-CoA:cholesterol transferase (ACAT) and stored in lipid droplets, then
de-esterified by hormone-sensitive lipase (HSL). Cholesterol may reach the outer mitochondrial membrane (OMM) by at
least three mechanisms. First, domains of the ER called the ‘mitochondria-associated membrane’ (MAM) may associate
with the OMM via ‘tethering proteins’ (e.g., sigma receptor), permitting non-vesicular flux of cholesterol to the OMM.
Second, cholesterol-binding transport proteins, such as those belonging to the STARD4/D5/D6 family, may deliver
cholesterol to the OMM. Third, lipid droplets may associate with the OMM via SNARE (soluble NSF attachment protein
receptor) proteins, permitting vesicular transport of cholesterol to the OMM. Irrespective of how the cholesterol reaches the
OMM, maximal steroidogenesis requires StAR (steroidogenic acute regulatory protein), which appears to interact with
OMM proteins (the ‘transduceosome’ complex) to trigger the rapid movement of cholesterol from the OMM to the IMM
where it is converted to pregnenolone by P450scc. © WL Miller.

this is a one-way trip: each StAR molecule can only transport one molecule of cholesterol to the
OMM. (iv) Nevertheless, each molecule of StAR is responsible for the movement of >100
molecules of cholesterol from the OMM to the IMM. (v) StAR acts exclusively on the OMM. This
is shown by the constitutive activity of constructs that immobilize it on the OMM, and by the
inactivity of constructs that localize it to the IMM or the intramembrane space. (vi) StAR must
undergo a series of conformational changes (molten globule transition) on the OMM to be
active. (vii) StAR interacts with other OMM proteins including VDAC1 (voltage-dependent anion
channel 1) and the OMM phosphate carrier protein. (viii) The peripheral benzodiazepine
receptor (also known as translocator protein, TSPO) which was long thought to be involved
in mitochondrial cholesterol import, is not required for StAR action [5–8]. New work indicates
that StAR interacts with the mitochondria-associated membrane (MAM) of the endoplasmic

Trends in Endocrinology & Metabolism, November 2017, Vol. 28, No. 11 773
Box 2. The Steroidogenic Acute Regulatory Protein, StAR
ACTH stimulates adrenal steroidogenesis in three time-frames. First, over weeks and months, ACTH stimulates adrenal
hypertrophy and hyperplasia, providing the cellular location for steroidogenesis. Second, over days, ACTH via cAMP
stimulates transcription of genes encoding steroidogenic enzymes, providing the steroidogenic machinery. Third, over
the course of 15–60 minutes, ACTH via StAR stimulates the delivery of cholesterol to that machinery. StAR is mainly
expressed in cells that exhibit a rapid steroidogenic response to tropic stimulation (adrenal, gonad, but not placenta); it is
unclear whether the low levels of StAR produced in extraglandular steroidogenic cells serve an essential role because
these tissues appear to be intact when StAR is mutated. StAR is produced as a 285 amino acid protein with a
mitochondrial leader peptide that is cleaved off upon its entry into mitochondria; extramitochondrial 37 kDa StAR has a
half-life of minutes, whereas intramitochondrial 30 kDa StAR has a half-life of several hours. Nevertheless, it is the
extramitochondrial protein that is active because StAR must interact with the OMM to stimulate steroidogenesis; the
intramitochondrial 30 kDa protein is inactive because of its location, and not because it has been processed during
mitochondrial import. Steroidogenic cells (i.e., those expressing P450scc) can make small amounts of steroids in the
absence of StAR. The mechanism(s) of this StAR-independent steroidogenesis are unclear: the placenta lacks StAR but
produces the progesterone needed to maintain pregnancy. Mutations in StAR are characterized by adrenal insufficiency
with salt loss and 46,XY sex reversal secondary to a severe defect in Leydig cell steroidogenesis; because the ovary is
steroidogenically quiescent until puberty, gonadal manifestations appear at adolescence in females. Both sexes
accumulate cholesterol and cholesterol esters in the adrenals, leading to the name ‘congenital lipoid adrenal hyper-
plasia’. Placental steroidogenesis is normal; affected fetuses reach term, have normal parturition, and have no
anomalies other than impaired steroidogenesis. ACTH and LH, via cAMP, stimulate phosphorylation of Ser195 to
maximize StAR activity, and also stimulate very rapid transcription of the STAR gene. StAR then acts with an OMM
protein complex (which remains under investigation, but does not include PBR/TSPO) to increase the flow of cholesterol
from the OMM to the IMM, where P450scc resides. Thus chronic regulation of steroidogenesis is based on establishing
the cellular steroidogenic machinery, whereas acute regulation is based on the delivery of cholesterol to that machinery
[2,3].

reticulum (ER) [9] and that StAR requires a chaperone protein called GRP78 for proper folding
and activity [10]. Thus we have learned much about the StAR protein, but we still have no idea
how this protein triggers a flood of cholesterol moving from the OMM to the IMM [3].

Potential Approaches
Perhaps more attention should be focused on cholesterol itself. If we can follow a cholesterol
molecule from its intracellular sources to its binding to P450scc on the IMM, we would have to
encounter the StAR protein during that journey. What then is the ‘molecular itinerary’ of an
intracellular cholesterol molecule? What proteins or other molecules does it interact with? How
does one molecule of StAR, sitting on the cytoplasmic aspect of the OMM, move >100
molecules of cholesterol to the IMM? Could this really be as simple as the creation of ‘contact
sites’ between the OMM and IMM, permitting cholesterol to flow down a concentration
gradient? And if so, are there specific proteins with which StAR interacts to induce contact
site formation? It appears that steroidogenic cholesterol is in a kinetically different ‘pool’ than
the structural cholesterol that is an integral part of membranes [2], and that this labile pool of
steroidogenic cholesterol has multiple sources, including: (i) low-density lipoprotein (LDL)
cholesterol that is liberated from esters by lysosomal acid lipase and traffics through late
endosomes via the NPC proteins (so-named because their mutations cause Niemann–Pick
type C disease); (ii) high-density lipoprotein (HDL) cholesterol that is de-esterified by
hormone-sensitive lipase and transferred to the OMM by the association of mitochondria
and lipid droplets via SNARE proteins; (iii) free, de-esterified, insoluble cholesterol trans-
ported through the cytoplasm bound to transfer proteins such as STARD4, STARD5, and
STARD6; (iv) cholesterol newly synthesized in the ER brought to the OMM by association of
the MAM and mitochondria via tethering proteins including the sigma receptor (Figure 1) [3].
In non-steroidogenic cells, about 30% of cholesterol trafficking is via energy-requiring
vesicular transport (membrane associations) and 70% represents non-vesicular transport
via binding proteins [11], but it seems unlikely that these percentages will hold in steroido-
genic cells where vastly greater amounts of cholesterol are in flux, and where the ultimate
destination is the OMM. The interactions of cholesterol with each of these factors require
further study.

774 Trends in Endocrinology & Metabolism, November 2017, Vol. 28, No. 11
Phosphorylation of P450c17
Question 2: Does Phosphorylation of P450c17 Play a Role in Mediating 17,20 Lyase
Activity In Vivo?
Importance
The 17,20 lyase activity of P450c17 is the only known mechanism for the biosynthesis of 19-
carbon (C19) steroids, which are the precursors of all androgens and estrogens. P450c17 is of
central importance in reproduction, sex-steroid-dependent malignancies (e.g., breast and
prostate cancers), disordered sexual development (DSD), and hyperandrogenic disorders
including polycystic ovary syndrome (PCOS).

Background
The synthesis of cortisol, a 17-hydroxy C21 steroid, requires the 17a-hydroxylase (17OHase)
activity of P450c17 (CYP17A1). Serum cortisol concentrations remain constant throughout the
lifespan, whereas the levels of dehydroepiandrosterone (DHEA) and its sulfate, (DHEAS) rise
nearly 100-fold during adrenarche (Figure 2). This clinical observation illustrates a complex
biochemical fact: although a single enzyme catalyzes both 17OHase and 17,20 lyase activities,
these two activities are independently regulated. Synthesis of DHEA(S) requires the 17,20 lyase
activity of P450c17 that cleaves the 17,20 carbon bond, converting C21 steroids to C19
steroids (Box 3). Like all microsomal P450 enzymes, P450c17 mediates catalysis via a single
active site by using electrons from NADPH donated via the flavoprotein P450 oxidoreductase
(POR); the sequential 17OHase and 17,20 lyase activities each require a pair of electrons,
meaning that two POR molecules must sequentially donate two single electrons. Current data

3.5

3
Males
2.5 Females
Serum
DHEAS 2 Corsol
(mg/l) O
1.5 (17OHase)

1
DHEA(S)
0.5
(17,20 Iyase)
0
0.1 1 10 100
Age (years)

Figure 2. Adrenarche and Adrenopause. The graph displays normal serum concentrations of dehydroepiandros-
terone (DHEA) sulfate (DHEAS) on a linear scale against age on a logarithmic scale. DHEA(S) concentrations are high in the
fetus and newborn, fall rapidly in the first month of life, and remain fairly low during childhood. The adrenal zona reticularis
becomes histologically identifiable at around age 3 years and makes small amounts of DHEA(S). Around age 8 years, and
before the rise in serum gonadotropins and the onset of puberty, secretion of DHEA(S) rises, reaching maximal levels
around age 25. After age 40–50 years, DHEA(S) levels fall slowly, approaching childhood levels in the elderly (adreno-
pause). DHEAS is typically lower in females, who have two copies of the X-linked gene for steroid sulfatase, compared to
men, who have only one copy of this gene. By contrast, serum cortisol concentrations, when averaged over their 24 h
diurnal rhythm, remain approximately constant throughout the human lifespan. Cortisol, a 17-hydroxy C21 steroid, is a
marker for the 17-hydroxylase activity of P450c17, whereas DHEA and DHEAS, which are C19 steroids, are markers for
the 17,20 lyase activity of P450c17. © WL Miller.

Trends in Endocrinology & Metabolism, November 2017, Vol. 28, No. 11 775
Box 3. 17,20 Lyase Activity
It was once thought that 17-hydroxylase activity and 17,20 lyase activity (which removes carbons 20 and 21 from C21
steroids to form C19 precursors of sex steroids) were separate enzymes: the cell-type distribution of these activities
differed, and distinct diseases were reported for each. The cloning of bovine and human P450c17 enzymes and their
expression in vitro proved that P450c17 catalyzes both activities. This discovery prompted the genetic and biochemical
search for factors that confer 17,20 lyase activity on P450c17. Genetic studies of patients having hormonal evidence for
normal 17-hydroxylase activity but deficient 17,20 lyase activity have found defects in P450c17 itself, in P450
oxidoreductase (POR), and in b5; oddly the first family described clinically as having 17,20 lyase deficiency was found
to have a defects in AKR1C2 and AKR1C4 in the backdoor pathway of androgen synthesis, and not in 17,20 lyase
activity [23]. Biochemical studies confirmed the role of b5 and identified phosphorylation of P450c17, but genetic
searches have yet to confirm a role for P450c17 phosphorylation. Thus 17,20 lyase deficiency is a syndrome with
multiple causes and is not a single specific disease [107].

identify four factors that favor 17,20 lyase activity: (i) a high molar ratio of POR to P450c17;
(ii) correct charge–charge interactions between the acidic residues on the surface of the
electron-donating domain of POR and the basic residues of the redox-partner binding site
of P450c17; (iii) the presence of cytochrome b5 (b5), which appears to act as an allosteric
factor rather than as an alternative electron donor (although this remains a topic of study); (iv)
the phosphorylation of P450c17 on serine and threonine (but not tyrosine) residues [1,12].

Many questions are contained in these observations. Cytochrome b5 has minimal effect on
17OHase but increases the maximum velocity (Vmax) of 17,20 lyase (which is much slower than
17OHase) without affecting the Michaelis–Menten constant (Km). While the precise mechanism
of b5 action remains under investigation, it appears to alter the dynamic behavior of the active
site of P450c17, increasing its catalytic efficiency [13]. In addition, b5 does not promote 17,20
lyase activity equally well with all substrates: b5 strongly stimulates conversion of 17OH-
pregnenolone to DHEA, but conversion of 17OH-allopregnanolone to androsterone proceeds
robustly in the absence of b5. The ‘allosteric’ action of b5 was initially thought to require the
transient formation of a P450c17/POR/b5 ternary complex [14], but no evidence for such a
complex has been found to date, and it appears that the same residues of P450c17 interact
with both POR and b5. It may be that P450c17 forms a dimer, with one oligomer interacting
with b5 while the other interacts with POR [15]; detailed structural studies will be necessary to
resolve the mechanism by which b5 increases the Vmax for 17,20 lyase.

Independently of b5, phosphorylation of P450c17 also promotes 17,20 lyase activity [12]. Both
Ser/Thr phosphorylation of P450c17 and the presence of b5 can independently maximize
17,20 lyase activity [16]. P450c17 phosphorylation is stimulated by cAMP, but does not involve
the ERK1/2, MEK1/2, PKA/PI3K/Akt, or Ca/calmodulin/MEK pathways, is opposed by PP2A
but not PP4, and is promoted by phosphoprotein SET [14]. P450c17 can be phosphorylated
and its 17,20 lyase activity augmented by p38a (MAPK14), but not by p38b, p38d, or p38g
(MAPK11–13), in human adrenal NCI-H295A cells or by purified proteins in vitro [17].
However, many questions remain about P450c17 phosphorylation: (i) MAP kinases, such
as p38a, respond to cellular stimuli via a cascade of kinases and phosphorylation events that
amplify the initial signal [12], but what are the upstream kinases that activate adrenal p38a? (ii)
Which P450c17 residues are phosphorylated? Mutagenesis of the consensus phosphorylation
site for p38a in human P450c17 did not affect the ability of p38a to phosphorylate P450c17 or
increase its 17,20 lyase activity. (iii) How does P450c17 phosphorylation promote 17,20 lyase
activity? Both b5 and phosphorylation by p38a increase the Vmax of 17,20 lyase activity without
altering the Km; the actions of phosphorylation and b5 on lyase activity are not additive, suggesting
that both act by the same mechanism, probably by promoting interaction with POR, but this
remains unproven. (iv) There is no proof that P450c17 phosphorylation or p38a promote 17,20
lyase activity in normal human physiology. (v) It remains possible that other kinase(s) may be more
important. (vi) Does P450c17 phosphorylation contribute to the hyperandrogenism of premature
exaggerated adrenarche, PCOS, or other hyperandrogenic states? A potential link between

776 Trends in Endocrinology & Metabolism, November 2017, Vol. 28, No. 11
P450c17 phosphorylation causing hyperandrogenism and insulin receptor phosphorylation
causing insulin resistance in PCOS was proposed in the initial report of P450c17 phosphorylation;
additional data are consistent with but have not proven this link [18]; further studies are needed.

Potential Approaches
Mass spectrometric (MS) studies of proteolytically cleaved phospho-P450c17 could identify its
phosphorylated residues. Our (unpublished) efforts toward this goal using immunologically
purified P450c17 from NCI-H295A cells were frustrated by impure protein, incomplete cover-
age of proteolytic peptides, and an apparently low molar abundance of phospho-P450c17 in
cAMP-treated NCI-H295A cells. Pure, bacterially expressed p38a phosphorylates pure, bac-
terially expressed human P450c17 in vitro, conferring 17,20 lyase activity [17]; this could be an
optimal source of phospho-P450c17 for study.

Determination that p38a is the specific kinase involved, or at least an important kinase, will be
difficult. siRNA knockdown of p38a in NCI-H295A cells reduced 17,20 lyase activity by only
half, suggesting there may be another kinase(s) with similar activity. Alternatively, this residual
17,20 lyase activity may reflect the action of b5. A central role for b5 in 17,20 lyase activity is
supported by its robust expression in cells that make C19 steroids, and by the finding that its
mutations disturb C19 steroid synthesis; no similar evidence illuminates the potential role of
p38a (or of any other kinase). Immunocytochemical study of the distribution of expression of
p38a or other candidate kinases in the post-adrenarchal human adrenal may be informative, as
it has been for b5. It is possible that b5 is the principal physiologic mechanism promoting 17,20
lyase activity and that a role for P450c17 phosphorylation is only seen in pathologic, hyper-
androgenic states. Unfortunately, animals are poor models for the human adrenal because the
level of adrenal 17,20 lyase activity in non-primate mammalian species is low. Laboratory
rodents do not express P450c17 in their adrenals; Leydig cell-specific mouse knockout of b5
yields phenotypically and reproductively normal mice, although an impairment of testicular
17,20 lyase activity is seen with gonadotropin stimulation [19].

The ‘Backdoor Pathway’ of Androgen Synthesis


Question 3: Does the Backdoor Pathway Play an Essential Role in Human Male Sexual
Differentiation or Reproductive Function?
Importance
Understanding androgen synthesis is essential for understanding human genital development
and its disorders, for understanding hyperandrogenic disorders such as PCOS and the virilizing
forms of congenital adrenal hyperplasia (CAH), and for understanding the biology of sex-
steroid-dependent malignancies.

Background
Human embryonic genital anlage differentiate into female external genitalia unless acted upon
by androgens. Until recently the ‘conventional dogma’ was that the human fetal testis produces
testosterone, which is then converted to the more powerful androgen dihydrotestosterone
(DHT) by the action of 5a-reductase within the genital skin, which then induces fusion of the
labioscrotal folds to close the perineum. Consistent with this model, mutations in the SRD5A2
gene encoding 5a-reductase type 2 cause a form of DSD. Studies of fetal genital endocrinology
are difficult. The group of Marilyn Renfree developed a marsupial, the Tamar wallaby, as a
unique animal system for such studies; these animals are born at a stage of development that
precedes differentiation of the external genitalia [20]. Genital development proceeds while the
young are in the maternal pouch, permitting comparatively easy access and the ability to follow
the development of an individual longitudinally. Studies of the Tamar wallaby showed that these
animals produce DHT by an alternative ‘backdoor’ pathway (Figure 3) wherein 17OH-proges-
terone (17OHP) is converted to DHT without the intermediacy of DHEA, androstenedione, and

Trends in Endocrinology & Metabolism, November 2017, Vol. 28, No. 11 777
Cholesterol
P450scc
+ StAR

Pregnenolone
P450c17 Reducve
3αHSD
3βHSD2 17OHP 5αRed1 5α-Pregnan- ? AKR1C2 17OH-
17OH-Preg 17α-ol-3,20-dione Allopregnanolone

P450c17+ b5 (P450c17+ b5) P450c17

3βHSD2
DHEA Androstenedione Androsterone

17βHSD3/5 17βHSD3/5 Oxidave 17βHSD3/5


3αHSD
3βHSD2 5αRed2 ? RoDH
Androstanediol
Androstenediol Testosterone DHT

Figure 3. The Backdo or Pathway of Androgen Synthesis. As in the generic pathway shown in Box 1, the left-hand
column shows the D5 pathway; these steroids may be acted on by 3bHSD2 to yield the corresponding D4 steroids. In the
backdoor pathway, 17OH-progesterone (17OHP) is first 5a-reduced by 5aRed1 (SRD5A1 gene) to 5a-pregnan-17a-
ol-3,20-dione, which is then 3a-reduced, probably by AKR1C2, to yield 17a-allopregnanolone. When 17a-allopreg-
nanolone is the substrate, P450c17 can catalyze its 17,20 lyase activity without the allosteric action of b5 [108], yielding
androsterone, which then can be acted on by testicular 17bHSD3 or adrenal 17bHSD5 (AKR1C3) to yield andros-
tanediol (not to be confused with the corresponding D5 steroid, androstenediol). Androstanediol may then be 3a-
oxidized, possibly by 17bHSD6 (also known as retinol dehydrogenase, RoDH, encoded by HSD17B6) or possibly by
AKR1C4 [23,24]. The identities of the enzymes catalyzing the reductive and oxidative 3aHSD reactions remain
under investigation. Abbreviations: b5, cytochrome b5; DHEA, dehydroepiandrosterone; DHT, dihydrotestosterone;
17OH-Preg, 17-hydroxypregnenolone. © WL Miller.

testosterone [1,21]. Evidence for a role for this pathway in human physiology and disease is
beginning to emerge. First, studies of the urinary metabolites of steroidal intermediates unique
to the backdoor pathway show that it is robustly active in 21-hydroxylase deficiency [22].
Second, mutation of the AKR1C2 and AKR1C4 genes, which appear to participate uniquely in
the backdoor pathway, is a newly described cause of DSD, suggesting that both the classical
pathway of androgen synthesis and the backdoor pathway are needed for normal human male
sexual differentiation [23,24], Third, the backdoor pathway appears to be activated in the male
minipuberty of infancy [25]. Finally, the enzymes of this pathway are also expressed in the post-
pubertal human ovary and appear to be upregulated in PCOS [26].

Potential Approaches
Much remains to be learned about the role of the backdoor pathway in human development,
normal physiology, and disease. Studies examining urinary metabolites of the steroidal inter-
mediates of this pathway are comparatively simple and are non-invasive, and are hence readily
approved by research boards. Such studies will eventually need to be complemented by
measurements of the relevant steroids in serum, spinal fluid, and other body compartments.
Immortalized cell models that express this pathway are needed, but studies are first necessary to
see to what extent, if any, the backdoor pathway is expressed in standard cellular models of
steroidogenesis such as mouse Leydig MA-10 cells or human adrenal NCI-H295A cells. Such
studies should clarify the identities of all the enzymes in the pathway and identify which enzymatic
steps may be catalyzed by more than one enzyme. Stably transfected variants of MA-10 or NCI-
H295A cells could be built once the missing enzymes are identified definitively. The identification
of the role of the backdoor pathway in normal human male sexual development derives from a

778 Trends in Endocrinology & Metabolism, November 2017, Vol. 28, No. 11
single report of two families that were studied genetically but not hormonally [23]. Additional
families need to be identified and studied thoroughly at the clinical, hormonal, and genetic levels;
agnostic studies of patients with DSD by whole-exome sequencing or a similar tactic need to pay
special attention to the known and candidate genes in the backdoor pathway.

New Adrenal Androgens


Question 4: What is the Significance of ‘Alternative Adrenal Androgens’?
Importance
For about 60 years studies of ‘adrenal androgens’ have focused on DHEA(S), androstenedione
(D4A), and occasionally testosterone (T). New data suggest that the principal androgen made
by the human adrenal is 11-ketotestosterone (11KT), a rarely studied steroid. If 11KT and
related steroids are the principal androgens secreted by the human adrenal, then virtually all
clinical studies of the adrenal in hyperandrogenic disorders will need to be repeated.

Background
Mitochondrial 11b-hydroxylase (P450c11b, encoded by CYP11B1) is well known for its role in
cortisol synthesis, but is not usually considered in androgen synthesis. Studies in 1973
identified 11b-hydroxyandrostenedione (11OHD4A) as an adrenally produced steroid, but
only in 2013 was substantial attention directed to adrenal 11-hydroxy androgens [27–29];
recent reports expand those observations [30,31,32,33–35]. The pathways involved are shown

Cholesterol
P450scc
+ StAR
Pregnenolone
P450c17
17OH-Preg
P450c17+ b5

DHEA
3βHSD2
5αRed2 11βHSD2 P450c11β
11Keto- 11Keto–Δ4 11βOH–Δ4 Androstenedione
Androstanedione
17βHSD3/5
5αRed2 11βHSD2 P450c11β
11Keto-DHT 11Keto–T 11βOH–T Testosterone (T)

Figure 4. Novel Adrenal Androgens. Adrenal androgen synthesis is generally described according to the pathway
[cholesterol ! pregnenolone ! 17OH-pregnenolone (17OH-Preg) ! DHEA ! androstenedione] shown on the right.
Although DHEA and androstenedione are usually termed ‘adrenal androgens’, they are more properly simply termed
C19 (19-carbon) precursors of androgens because these steroids have virtually no capacity to bind to and transactivate
the androgen receptor. Small amounts of androstenedione may be converted to testosterone in the adrenal by 17bHSD5
(AKR1C3). However, recent data show that the principal androgenic steroid secreted by the adrenal is 11-ketotestoster-
one (11Keto-T). Both androstenedione and testosterone may undergo 11-hydroxylation catalyzed by P450c11b
(CYP11B1 gene) to yield 11OH-androstenedione (11OH-D4) and 11OH-testosterone (11OH-T), respectively. These
11-hydroxysteroids may be oxidized by 11b-hydroxysteroid dehydrogenase type 2 (which is more familiar for its role
in the oxidation of cortisol to cortisone) to 11-ketoandrostenedione (11Keto-D4) and 11-ketotestosterone (11Keto-T),
respectively. These 11-keto steroids may then be 5a-reduced by 5a-reductase type 2 (SRD5A2 gene) in peripheral
tissues, and possibly also by 5a-reductase type 1 (SRD5A1 gene) in the adrenal itself, to 5a-androstanedione
and 5a-dihydrotestosterone, respectively. The quantitatively predominant pathway is androstenedione ! 11OH-
D4 ! 11Keto-D4 ! 11-Keto-T, as indicated by the bold arrows. © WL Miller.

Trends in Endocrinology & Metabolism, November 2017, Vol. 28, No. 11 779
in Figure 4. Both 11KT and 11KDHT are bona fide androgens that bind to and transactivate the
androgen receptor [27,28,36,37]. Whereas most studies have addressed the synthesis of
these steroids in castration-resistant prostate cancer, Turcu et al. reported that 11OHD4A,
11KD4A, 11OHT, and 11KT were elevated 3–4-fold in patients with 21-hydroxylase deficiency,
were higher in the adrenal vein than in the inferior vena cava, rose in response to intravenous
ACTH in adrenally intact individuals [31], and that these steroids correlate with adrenal volume
and predict testicular rest tumors in CAH [38]. Thus prior studies of the potential role(s) of
‘adrenal androgens’ in premature, exaggerated adrenarche, PCOS, and other hyperandro-
genic conditions need to be reconsidered.

Potential Approaches
The above studies of 11-hydroxy adrenal androgens used modern assays of urinary steroid
metabolites by gas chromatography followed by MS (GC/MS) or of serum steroids by liquid
chromatography followed by tandem MS (LC-MS/MS). Urinary steroids by GC/MS are reliable
but not widely available because throughput is limited; nevertheless, this approach is a very
useful research tool [39]. LC-MS/MS requires more elaborate and expensive equipment, and
more highly trained personnel, but is vastly more sensitive and accurate than conventional
immune-based systems, and is not vulnerable to ‘interfering substances’ or immunologic
overlap of analytes [40]. LC-MS/MS has become the ‘gold standard’ for steroid research
[41] and is being applied to studies of the ‘backdoor pathway’ and to new adrenal androgens.
These ‘new’ adrenal androgens should be measured by these technologies in all adrenal
hyperandrogenic disorders.

The Fetal Adrenal


Question 5A: Does the Human Fetal Adrenal Serve a Physiologic Role? Question 5B: What
Triggers Involution of the Fetal Adrenal at Birth?

Importance
Unlike other mammals, primates, especially old world primates including H. sapiens, have very
large adrenals during fetal life that make large amounts of DHEA(S). Despite decades of
research, no essential role for the human fetal adrenal has been found.

Background
The human fetal adrenals grow throughout gestation, reaching a combined weight of about 8 g
at birth, which is about equal to the combined weight of the adult adrenals. Combined fetal
adrenal weight then decreases precipitously to about 2 g via apoptosis of the fetal zone within a
few weeks following birth. The fetal adrenals represent 0.4% of total body weight from the 3rd to
6th months of gestation, diminishing to about 0.2% at birth, but represent only 0.01% of adult
weight (Figure 5) [42]. The fetal adrenal consists of a large fetal zone that involutes following
birth, and a small definitive zone that gives rise to the adult adrenal. Fetal adrenal steroidogen-
esis begins at around 50–52 days post-fertilization; the fetal adrenal expresses 3b hydrox-
ysteroid dehydrogenase (3bHSD2, encoded by HSD3B2) and produces cortisol at weeks 8–
10, but HSD3B2 expression and cortisol synthesis wane thereafter. This transient cortisol
synthesis may suppress adrenal synthesis of androgenic C19 steroids during the time when
genital differentiation is occurring [43,44]. After 12 weeks the fetal zone is morphologically
distinct, lacks 3bHSD, but expresses CYP11A1, CYP17A1, CYB5, and the steroid sulfotrans-
ferase SULT2A. Consequently the fetal circulation contains high concentrations of DHEA and
DHEAS, which fall rapidly after birth. The fetal adrenal also expresses some 17bHSD5
(AKR1C3) which can convert androstenedione to testosterone. Fetal adrenal DHEA(S) can
be converted to estriol by the placenta, hence maternal estriol levels reflect fetoplacental
steroidogenesis. However, various genetic disorders of steroidogenesis, which constitute
human ‘gene knockout experiments of nature’, indicate that fetal adrenal steroidogenesis,
and the fetal adrenal itself, are not essential for fetal development, survival, or parturition.

780 Trends in Endocrinology & Metabolism, November 2017, Vol. 28, No. 11
Birth

8 4
Combined adrenal weights

Relave adrenal weight


(g adrenal/kg body)
6 3
(g)

o
4 2

2 1

0 0

1 2 3 4 5 67 8 9 1 3 5 7 9 11 13 15 17
Fetal Postnatal
(months) (years)

Figure 5. Adrenal Growth (and Involution) Over the Human Lifespan. Combined adrenal weight (closed symbols)
and the ratio of adrenal weight to body weight (open symbols) are shown as a function of fetal postnatal age. The fetal
adrenals begin to grow rapidly between the second and third months of gestation as a result of hypertrophy of the fetal
zone, and reach 8 g/kg body weight from months 3–6 of gestation. Within 2 months of birth the fetal zone involutes and
adrenal weight falls precipitously; adrenal weight then increases slowly but remains roughly constant compared to body
weight thereafter. Reprinted, with permission, from [42].

Because regulation of fetal salt and water balance is handled by the placenta, CYP11B2
mutations that disrupt aldosterone synthesis do not affect the fetus, and the fetus does not
need mineralocorticoids until birth. Fetal serum cortisol concentrations are only 5–10% of those
in the mother, and cortisol appears to be unnecessary for fetal development because defects in
STAR and CYP11A1 impair all cortisol synthesis but do not affect development, and one fetus
has been described who was homozygous for null mutations in the glucocorticoid receptor but
developed its lungs and other organs normally (although the newborn had severe hypoglycemia
and hypertension) [45]. Absent expression of CYP17A1 excludes production of DHEA(S), but
such fetuses reach term, are normally developed, and undergo normal parturition. Thus
steroidogenic gene-knockout experiments of nature cast substantial doubt on any proposed
essential roles for fetal adrenal steroids. These observations are reinforced by the observations
that patients with X-linked adrenal hypoplasia congenita (AHC, caused by mutations in DAX1),
with IMAGe syndrome (caused by mutations in CDKN1C), or with other congenital adrenal
hypoplasia syndromes have normal organogenesis and are physiologically intact until after
birth.

At birth the fetal zone of the adrenal undergoes apoptosis and involutes rapidly, with rapid falls
in circulating DHEA(S); the event(s) triggering fetal adrenal involution, cellular remodeling, and
the changed pattern of steroidogenesis remain unclear. Parturition itself, instead of some 40
week timing mechanism, appears to be the triggering event, and it has been proposed that
withdrawal of a hypothetical placental growth factor is responsible [46], but no such factor has
been identified. An alternative proposal is that the profound change in fetal adrenal oxygenation
at birth triggers this event. The fetal abdominal viscera receive hypoxic blood with a partial
pressure of oxygen of only 2 kPa (20–23 mm Hg), which rises abruptly to 20 kPa at birth

Trends in Endocrinology & Metabolism, November 2017, Vol. 28, No. 11 781
with the change from fetal to adult circulation. Microarray analysis of gene expression in
human fetal adrenal cells or adrenal NCI-H295A cells grown in 2% O2 and then ‘delivered’ to
20% O2 revealed profound changes in the transcriptome, including HSD3B2, STAR, and
CYP17A1, suggesting that the transition from hypoxia to normoxia may be a triggering event
[47]. Nevertheless, detailed mechanistic support for either of these hypotheses remains
lacking.

Potential Approaches
The proposal that withdrawal of a placental growth factor triggers fetal adrenal involution might
be addressed by studying the growth, apoptosis, and steroidogenic patterns of fetal adrenal
cells cultured with fetal versus adult serum, and by co-culture experiments with fresh placental
tissue from the same pregnancy (to obviate potential immunologic problems). Candidate
growth factors will need to be confirmed by activity with pure molecules and the identification
of a signal transduction pathway. These are classic endocrine approaches. The proposal that
the hypoxic–normoxic shift acts as a trigger requires identification of genes activated by this
shift that lead to apoptosis pathways. Although NCI-H295A cells model many features specific
to the fetal adrenal, and showed some changes in genes in the HIF-1a pathway [47], they may
not model this transition accurately, and future studies should therefore emphasize the use of
primary cultures of human fetal adrenal cells. Finally, it should be considered that these two
mechanistic proposals are not mutually exclusive.

Adrenarche and DHEA(S)


Question 6: Does DHEA Do Anything Important?
Importance
DHEA is a precursor of androgens and estrogens, but whether it exerts important physiologic
actions on its own, without further metabolic modification, remains controversial.

Background
Primates appear to be unique among mammals in having adrenarche and making large
amounts of DHEA(S) (Figure 2) [48,49]. DHEA is a precursor steroid in pathways to androgens
and estrogens, but it is unclear whether DHEA has any function without subsequent steroido-
genic transformation. Effects on the rodent brain are well documented [50], but genetic
disruption of human CYP17A1, which is required for the synthesis of DHEA(S), causes no
apparent human neuropsychiatric disorder. Premature, exaggerated adrenarche may result in
a modest degree of virilization, and appears to be a precursor of PCOS [51]. Many claims have
been made that DHEA acts directly to prolong life and vitality, and to stimulate the immune
system, but these reports are based on clinical and epidemiologic observations, and are
subject to observer bias; mechanistic evidence for DHEA action as a hormone itself is lacking.

Circulating concentrations of DHEA(S) fall rapidly after birth with the involution of the fetal
adrenal, rise slowly in early childhood as the zona reticularis differentiates, and accelerate
before the onset of puberty (Figure 2). Concentrations of DHEAS are 100–1000-fold higher than
those of DHEA, principally because most DHEAS is protein-bound; consequently, DHEA
concentrations are acutely responsive to ACTH whereas DHEAS concentrations are not.
DHEA(S) synthesis is largely confined to the zona reticularis. Immunocytochemical studies
of human and monkey adrenals show that expression of b5 and SULT2A1 increase during
adrenarche and are confined almost exclusively to the zona reticularis, while expression of
HSD3B2 in the reticularis falls to very low levels [52,53]. Studies of CYB5 gene transcription
have not identified transcription factors that might participate in CYB5 expression that do not
also participate in CYP17A1 expression [54]. Thus, while many molecular events have been
described that are needed for adrenarche to occur, a molecular trigger has not been identified;
a role for IGF1 has been proposed but remains unproven.

782 Trends in Endocrinology & Metabolism, November 2017, Vol. 28, No. 11
Potential Approaches
The identification of the molecular trigger(s) and mechanism(s) of adrenarchal progression and
of the differentiation and growth of the zona reticularis is needed. The temporal dynamics and
roles of new adrenal androgens (see the earlier section on New Adrenal Androgens) need to be
studied across the developmental spectrum. Two fundamental problems have limited research
about adrenarche: the lack of suitable animal models and the almost exclusive reliance on
clinical studies; more in vitro cell biology is needed. NCI-H295 adrenocorticocarcinoma cells
are the only available cell culture system for the human adrenal. Although these cells make
DHEA they do not model the morphologic changes in the developing adrenal zona retic-
ularis or the evolving rise in DHEA synthesis. Additional cell lines might be helpful. Con-
struction of transgenic mice that express CYP17A1 in their adrenals may not help because
mice silence their endogenous Cyp17a1 gene by DNA methylation [55], and would probably
methylate a CYP17A1 transgene. Information about zone-specific expression of CYB5 is
essential.

P450 Oxidoreductase
Question 7: Do Polymorphisms in the POR Gene Affect Puberty, Drug Metabolism, or
Disease Risk?
Importance
All human disease is multifactorial: monogenic diseases caused by the same mutations may
yield different phenotypes depending on ‘modifying genes’. More-complex disorders, such as
hypertension, metabolic syndrome, neurodegenerative disorders, etc., reflect the interplay of
many genes and environmental factors. The identification of such factors, and the quantitation
of their associated risk(s), is central to understanding and treating disease. POR serves as the
electron donor for all microsomal cytochrome P450 enzymes and for several other proteins,
including heme oxygenase, fatty acid elongase, squalene monoxygenase, and b5 [56], hence
POR merits consideration in disorders beyond those concerning steroidogenesis.

Background
The human genome encodes only 57 cytochrome P450 (CYP) enzymes, of which 50 are
microsomal and require POR (Figure 6). These microsomal P450 enzymes include steroido-
genic P450c17 (CYP17A1 gene), P450c21 (CYP21A2 gene), and P450aro (aromatase,
CYP19A1 gene), the principal hepatic drug-metabolizing enzymes, and enzymes involved in
the synthesis of oxysterols, eicosanoids, and leukotrienes [57]. About 94% of hepatic drug
metabolism is catalyzed by only eight CYP enzymes, and CYP2D6 and CYP3A4 account for
metabolism of >50% of clinically used drugs [58]. All P450 enzymes must receive electrons
from NADPH to permit catalysis; the 50 microsomal P450 enzymes receive these electrons via
POR (Figure 7) [59]. Mutation of POR results in a complex form of adrenal hyperplasia, with
partial deficiencies of P450c17, P450c21, and P450aro activities [60]. The POR gene is highly
polymorphic, with the frequency of specific polymorphisms varying greatly among ethnic
groups [61]. The impact of POR amino acid sequence variants depends on the CYP enzyme
receiving the electrons, the reaction being catalyzed, and the size and shape of the substrate
[62]. For example, the activity of human P450c17 is severely compromised by the common
POR mutations A287P and R457H; whereas the activity of P450aro is impaired by R457H but
not by A287P [63], and POR A287P has little effect on P450c21 [64]. The common variant
A503V reduces the activity of most, but not all, microsomal P450 enzymes in vitro. Clinical
studies found that A503V impairs bupropion metabolism by CYP2B6 [65], but A503V did not
alter the metabolism of erythromycin or midazolam by CYP3A4 [66] or affect warfarin metabo-
lism [67], and is not a modifier gene for CAH caused by CYP21A2 mutations [68]. Studies in
Chinese subjects found that A503V is associated with increased risk of bladder cancer [69],
and that several non-coding POR polymorphisms correlate with failure to quit smoking [70]. It is
not known if such associations will be confirmed in other groups.

Trends in Endocrinology & Metabolism, November 2017, Vol. 28, No. 11 783
57 Human CYP genes
7 50

Type I mitochondria Type II


(bacteria) Endoplasmic reculum
15 25 10

Sterol Drug Biosynthesis: Orphan


biosynthesis metabolism cholesterol,
fay acids,
eicosanoids,
steroids

Figure 6. Human Cytochrome P450 Enzymes. The human genome has only 57 CYP genes for cytochrome P450
enzymes; by contrast, Drosophila melanogaster has 90, mice have 102, and Arabidopsis thalina has 244. P450s may be
type I, targeted to mitochondria, or type II, that are targeted to the endoplasmic reticulum (ER). Type I enzymes mediate
catalysis using electrons from NADPH transferred via ferredoxin reductase and ferredoxin (Figure 7A). The seven human
type I P450s are P450scc (CYP11A1, the cholesterol side-chain cleavage enzyme), P450c11b (CYP11B1, 11b-hydro-
xylase), P450c11AS (CYP11B2, aldosterone synthase), P450c24 (CYP24, vitamin D 24-hydroxylase), P450c27
(CYP27A1, a bile acid 27-hydroxylase), P450c1a (CYP27B1, vitamin D 1a-hydroxylase); and CYP27C1 (which converts
vitamin A1 to A2). Type II P450s use electrons from NADPH transferred via P450 oxidoreductase (POR; Figure 7B). The 50
human Type II P450s fall into three functional groups: about 15 are involved in xenobiotic metabolism; about 25 participate
in biosynthesis of sterols, steroids, eicosanoids, and fatty acids; and about 10 are ‘orphans’ of uncertain function [57].
There is redundancy and overlap in the catalytic capacities of these enzymes. The human steroidogenic type II P450s are
P450c17 (CYP17A1, 17-hydroxylase/17,20-lyase), P450c21 (CYP21A2, 21-hydroxylase), and P450aro (CYP19A1,
aromatase). © WL Miller.

Potential Approaches
Few studies have considered the effects of POR mutations/polymorphisms on drug metabo-
lism in vivo. Test drugs are widely used to examine the activities of hepatic enzymes. The
metabolic ratio of orally administered dextromethorphan to dextrorphan is widely used to study
metabolism by CYP2D6 [71]. Studies of hepatic CYP3A4 typically use intravenous midazolam
to bypass intestinal CYP3A5 because most CYP3A4 substrates are also metabolized by highly
polymorphic intestinal CYP3A5 [72]. CYP3A4 yields both 1-hydroxy- and 4-hydroxymidazolam
in about a 4:1 ratio; because this ratio may vary with POR polymorphisms [73], both products
must be assessed. One patient with POR A287P had impaired disposal of test drugs metabo-
lized by CYP1A2, CYP2C9, CYP2D6, and CYP3A4 [74]. Further studies of POR-deficient
patients are needed. Much larger studies of drug metabolism are needed in clinically healthy
individuals who have been genotyped for both POR and for hepatic CYP polymorphisms and
variants. Such studies should consider all POR polymorphisms, not only those changing amino
acid sequences or transcription in vitro. Notably, a clinical study found that the common variant
A503V did not affect the doses of warfarin needed in patients, but three POR promoter
polymorphisms (173C > A, 208C > T, and rs2868177) did [75], even though these poly-
morphisms did not affect POR promoter activity as assessed by promoter/reporter studies in
transfected cells [76].

Extraglandular Steroidogenesis
Question 8: How Important is Extraglandular Steroidogenesis?

Importance
Extraglandular steroidogenesis has been studied for about 40 years. However, although many
tissues make steroids, and many functions have been proposed for extraglandular steroids, no
essential roles have been proven for these steroids because disruption of their synthesis by
Mendelian disorders of steroidogenesis does not lead to apparent extraglandular disorders.

784 Trends in Endocrinology & Metabolism, November 2017, Vol. 28, No. 11
(A) NADPH

NADP+
P450
e– e–
– ––
FAD –
– – Fe
– –
Fedx
FeRed

(B) NADPH
– P O4
P450
N A D P+ –

FMN –– Fe
FAD
b5
POR

Figure 7. Electron Transport to P450 Enzymes. (A) Mitochondrial cytochrome P450 enzymes receive electrons from
NADPH via two proteins; ferredoxin reductase (FeRed), which is bound to the inner mitochondrial membrane (IMM), and
ferredoxin (Fedx), which is loosely associated with the IMM or may be free in the mitochondrial matrix. The FAD moiety of
FeRed accepts two electrons from NADPH, yielding NADP+. These electrons pass to the iron–sulfur cluster (Fe2S2,
quadrilateral with dots) of Fedx, which then donates the electrons to the iron atom in the heme group of the P450 (square
with Fe). Negatively charged residues in Fedx () guide docking and electron transfer with positively charged residues (+) in
both FeRed and the P450. (B) Microsomal cytochrome P450 enzymes receive electrons from NADPH via the bi-lobed
protein P450 oxidoreductase (POR). Donation of electrons (e) to the FAD moiety elicits a conformational change, bringing
the isoalloxazine rings of the FAD and FMN (flavin mononucleotide) moieties sufficiently close together to permit electron
passage from the FAD to the FMN moiety. POR then returns to its original orientation, permitting the FMN domain of POR
to dock with the redox-partner binding site of the P450. The electrons then reach the P450 heme group to mediate
catalysis. The interaction between POR and the P450 is coordinated by electrostatic interactions, similarly to the
interaction of Fedx with mitochondrial P450s. For all P450s, the substrate-binding pocket lies on the opposite side of
heme ring from the redox-partner binding site. © WL Miller.

Background
In addition to steroidogenesis in the adrenals, gonads, and placenta, there is well-documented
extraglandular steroidogenesis in the brain [1,77], skin [78–80], adipocytes, leukocytes, gut,
lung, bone, heart, and thymus. To qualify as ‘extraglandular steroidogenesis’, a non-endocrine
cell must express P450scc, which is the only known enzyme that can remove the side chain of

Trends in Endocrinology & Metabolism, November 2017, Vol. 28, No. 11 785
cholesterol to yield pregnenolone. This differs from extraglandular modification of steroids that
are made elsewhere, such as the conversion of T to DHT by SRD5A2 in genital skin, or the
aromatization of androgens to estrogens in fat. Extra-adrenal 21-hydroxylation was docu-
mented in the early 1980s and was shown not to be catalyzed by P450c21; recent work
has identified CYP2C19 and CYP3A4 as two hepatic 21-hydroxylases, but the presence of
21-hydroxylase activity in many tissues, especially in the fetus, suggests that other enzymes
may also catalyze extra-adrenal 21-hydroxylation [81]. By contrast, steroidogenesis in the
brain, skin, and other tissues is catalyzed by the same enzymes that make steroids in the
adrenals and gonads [80,82]. It is this later form of extraglandular steroidogenesis that is most
vexing. For example, the mammalian brain makes allopregnanolone (3a,5a-tetrahydroproges-
terone), which acts as an anesthetic, anxiolytic, and anticonvulsant by acting on GABAA
receptors [83], and plays a major role in the reward systems of addiction to alcohol [84],
but no neuropsychiatric disorder has been identified in Mendelian disorders that should disrupt
its synthesis. Similarly, DHEA is an excitatory neurosteroid made in the central nervous system,
but its absence in the brains of patients with 17-hydroxylase deficiency appears to be benign.
Perhaps neurosteroids exert actions that overlap with those of other neurotransmitters, such
that disrupted neurosteroid biosynthesis may not be sufficient to cause a phenotype. Proving
the roles of factors that serve redundant functions is challenging.

Potential Approaches
The apparent actions of extraglandular steroidogenesis have been studied in greatest detail in
the skin and in the central nervous system. However, no studies of patients with steroidogenic
defects have addressed either steroidogenesis in these tissues or any potential, hypothetically
predicted clinical manifestations in these tissues. One approach is the direct study of extra-
glandular steroidogenic tissues from patients with well-defined steroidogenic defects, but such
tissues will rarely become available. However, other clinical studies can also be performed: for
example, P450c17 is expressed in the developing mouse and human nervous system, but
neuropsychiatric studies of behavior and neurologic function in 17-hydroxylase deficiency have
not been reported. Such studies should be a first step in assessing whether or not the steroids
made in these tissues serve any essential role. Assessing whether or not they might serve roles
that are merely modulatory, rather than truly essential, will be more difficult.

Congenital Adrenal Hyperplasia


Question 9: Why Is CAH Rare among African-Americans?

Importance
The genetics of CAH are complicated by the location of the duplicated C4/CYP21/TNX genes
within the human leukocyte antigen (HLA) locus, resulting in a very high rate of genetic
recombination. The high rate of recombination between the functional CYP21A2 gene and
the closely linked, non-functional CYP21A1P pseudogene results in the high incidence of
CAH and its unusual genetics, driven by gene conversions rather than by point mutations. The
apparently very low incidence of CAH among African-Americans is inconsistent with this
genetics and requires explanation.

Background
21-Hydroxylase deficiency (21OHD) is caused by mutations in CYP21A2, and is 10–100-fold
more common than forms of CAH cased by mutations in other genes. The CYP21A2 gene, the
closely linked C4 genes (encoding the fourth component of serum complement), and the TNX
gene (encoding the extracellular matrix protein tenascin-X) lie in a duplicated 30 kb unit in the
HLA locus on chromosome 6p21.33 (Figure 8). Both the C4A and C4B genes are functional but
only CYP21A2 encodes a functional 21-hydroxylase. The duplicated units share 98.5%
sequence identity and, because they are in the HLA locus, undergo a much higher rate of
genetic recombination than other genomic regions. This high rate of recombination inserts

786 Trends in Endocrinology & Metabolism, November 2017, Vol. 28, No. 11
Class I Class II

A C B TNF DR DQ DP GLO
Short arm of 250 kb 390 kb >300 kb
chromosome 6

0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 kb

C2 Bf RD G11/RP XB

G11/RP C4A 21A XA C4B 21B XB-S XB

ZA YA ZB YB

Figure 8. The CYP21 Gene Locus. (Top line) Macro-view of 2.5 Mb of the HLA locus on chromosome 6p21.33; the telomere is to the left and the centromere to
the right. The class III region containing the CYP21 genes lies between the class I and class II regions. The middle diagram (with a scale above in kb) shows (from
telomere to centromere) the genes for complement factor C2, properedin factor Bf, RD (NEFLE) and G11/RP (STK19), and the duplicated C4/CYP21/TNX loci; arrows
indicate transcriptional orientation, and the vertical dotted lines identify the duplication boundaries; the duplicated regions are colloquially termed A and B. The
expanded scale on the bottom line shows the C4A and C4B genes encoding the fourth component of complement, the inactive CYP21A1P pseudogene (21A), and the
active CYP21A2 gene (21B) that encodes P450c21. XA, YA, and YB are adrenal-specific transcripts that lack open reading frames. The XB gene (TNXB) encodes
tenascin-X, an extracellular matrix protein; XB-S (XB-short) encodes a truncated adrenal-specific form of tenascin-X that appears to interact with mitotic motor kinesin
Eg5 [109]. Promoter elements required for the transcription of the CYP21A1P and CYP21A2 genes lie 4.8 kb upstream within intron 35 of the corresponding C4 gene,
and also give rise to the adrenal-specific ZA and ZB transcripts that have open reading frames, but it is not known if they are translated. The arrows indicate the
orientation of transcription. © WL Miller.

sequences from the CYP21A1P pseudogene into the functional locus, providing the genetic
basis of CAH [1]. Duplication of the C4/CYP21/TNX unit post-dates mammalian speciation: the
mouse and human loci have different duplication borders, and some mammals do not have
duplicated loci. Dysfunction of the encoded 21-hydroxylase enzyme results in elevated con-
centrations of 17OHP, permitting reliable newborn screening. Such screening is now done in
most developed countries.

The incidence of CAH has been ascertained in many studies. Among 1.9 million newborns
screened in Texas from 1989 to 1995, the overall incidence was 1:16 008; the incidence was
similar among Caucasians (1:15 583) and Hispanics (1:14 417), but was much lower among
African-Americans (1:42 309) [85]. Screening 2.7 million newborns in Sweden yielded an
incidence of 1:11 853 [86], and three studies in Japan found an incidence of 1:20 000
[87]. Studies from Brazil have not subdivided the population by ethnicity, and it is therefore not
known if the incidence in Afro-Brazilians differs from that in Euro-Brazilians [88–91]. Unfortu-
nately, as of 2015 there are no direct data on the incidence of CAH in sub-Saharan African
countries [92]. Based on the incidence of 1:42 000 for African-Americans and the data that
24% of the African-American gene pool is of European origin [93], the ‘true’ incidence of CAH in

Trends in Endocrinology & Metabolism, November 2017, Vol. 28, No. 11 787
the African ancestors of African-Americans would be calculated to be about 1:250 000. There
is no apparent reason why the incidence in people of African ancestry should differ from that in
other ethnic groups.

Potential Approaches
Studies of CAH in sub-Saharan Africa are needed; such studies will need to incorporate
detailed ethnic, linguistic, and genetic data because human genetic diversity in sub-Saharan
Africa is substantially greater than it is throughout the rest of the world combined [94,95]. This is
the expected finding because human beings emerged ‘out of Africa’ only 50 000 to 100 000
years ago, long after modern H. sapiens evolved in Africa. Because other diseases are more
life-threatening, endocrinology is not yet well developed in Africa. However, newborn screening
programs for sickle-cell anemia and for congenital hypothyroidism are beginning in several sub-
Saharan African countries [92], hence the needed samples may become available in the
foreseeable future. A simple but unanswered question is whether the duplication of the
30 kb C4/CYP21/TNX unit is found in all human populations.

3b-Hydroxysteroid Dehydrogenase
Question 10A: What Is the Basis of Mild, Apparent 3bHSD Deficiency without HSD3B2
Mutations? Question 10B: Is the Subcellular Distribution of 3bHSD Regulated?
Importance
3bHSD lacks a leader peptide and is not proteolytically cleaved, and thus resembles a cytosolic
enzyme. However, 3bHSD is found primarily in the mitochondria. Is the subcellular location of
3bHSD regulated? If so, how? And if so, do different subcellular distributions influence
steroidogenic patterns?

Background
3bHSD catalyzes the conversion of D5 steroids to the corresponding D4 steroids (Box 1) [1].
Whereas the steroidogenic P450 enzymes are either bound to the cytoplasmic aspect of the ER
or the IMM, the subcellular distribution of the HSD enzymes is more varied (Box 4). Notably,

Box 4. Hydroxysteroid Dehydrogenases


HSDs lack heme groups and require NADH/NAD+ or NADPH/NADP+ to be able to reduce or oxidize steroids, and often
catalyze more than one reaction. HSDs include two 3bHSDs, multiple 3aHSDs, two 11bHSDs, and many 17bHSDs.
HSDs can be short-chain dehydrogenase/reductases (SDRs) or aldo-keto reductases (AKRs), both of which contain
paired Tyr/Lys residues in the active site. The SDR enzymes include the 3bHSDs, 11bHSDs, and 17bHSD-1, 2, and
3; the AKR enzymes include 17bHSD5 (AKR1C3) and the AKR1C2/4 enzymes in the backdoor pathway. Biochemi-
cally, the HSDs are dehydrogenases, using NAD+ to oxidize hydroxysteroids to ketosteroids, or reductases, using
NADPH to reduce ketosteroids to hydroxysteroids. These enzymes are typically bidirectional in vitro but unidirectional in
vivo, with directionality based on their affinities for, and the availability of, NAD(H) versus NADP(H): some ‘reductive’
enzymes become oxidases when NADPH is depleted or mutations impair NADPH binding [1]. 3bHSD converts
pregnenolone to progesterone, 17OH-pregnenolone to 17OHP, DHEA to androstenedione, and androstenediol to
testosterone, with Michaelis–Menten constants (Km) of 5.5 mM, which is substantially higher than the 0.8 mM Km for
P450c17, thus favoring the D5 pathway. There are two 3bHSD enzymes with 93.5% sequence identity: 3bHSD1 in
placenta, breast, liver, and brain, and 3bHSD2 in the adrenals and gonads. The fetal adrenal contains little 3bHSD2, and
its expression in the zona reticularis declines during adrenarche. 17bHSDs catalyze multiple reactions and may be
preferential oxidases or reductases; they vary widely in size, structure, substrate specificity, cofactor utilization, and
physiologic functions. Some 17bHSDs mainly catalyze reactions other than 17bHSD activity. The 11bHSDs have both
oxidase and reductase activities, depending on whether NADP+ or NADPH is available. 11bHSD1, which requires
micromolar steroid concentrations, catalyzes oxidation of cortisol to cortisone using NADP+ (Km 1–2 mM) and reduction
of cortisone to cortisol using NADPH (Km 0.1–0.3 mM); the reaction catalyzed depends on which cofactor is available.
Prednisone and cortisone must be reduced to their 11b-hydroxy forms by hepatic 11bHSD1 to become biologically
active. 11bHSD1 resides in the lumen of the ER where abundant NADPH is maintained by hexose-6-phosphate
dehydrogenase. 11bHSD2 is only 21% identical to 11bHSD1, catalyzes the oxidation of cortisol to cortisone using
NAD+, and functions at nanomolar steroid concentrations. 11bHSD2 in the distal nephron inactivates cortisol, which
would otherwise bind to and overwhelm mineralocorticoid receptors, permitting low concentrations of aldosterone to
elicit specific mineralocorticoid activity. Placental 11bHSD prevents maternal cortisol from reaching the fetus.

788 Trends in Endocrinology & Metabolism, November 2017, Vol. 28, No. 11
3bHSD enzymes have been found in the cytoplasm, ER, and in the mitochondrial matrix [96].
Recent work in human adrenal NCI-H295 cells and mouse Leydig MA-10 cells indicates that
the mitochondrion is the preferred location; mitochondrial import required the mitochondrial
translocase Tom22, and its knockdown eliminated steroidogenesis without inhibiting import of
P450scc [97]. The 3bHSD enzymes lack classic leader peptides and must interact with
chaperone proteins to reach their subcellular localization(s) [10], hence the subcellular location
of 3bHSD may be regulated, and may differ in different cell types, influencing the array of
steroids produced. The action of 3bHSD on pregnenolone probably occurs in the mitochon-
dria, whereas its actions on 17OH-pregnenolone and DHEA are more likely to be in the
cytoplasm, thus the intracellular distribution of 3bHSD would be expected to influence the
relative abundances of the steroids secreted by a steroidogenic cell. If so, changes in the
cellular distribution of 3bHSD might result in phenotypic variation. This intriguing possibility
merits study. The importance of studying the potentially regulated subcellular distribution of
3bHSD is highlighted by the recent observation that b5 can promote 3bHSD activity by
increasing its affinity for NAD, thus increasing the Vmax of the enzyme, resulting in a modest
(15–25%), increase in activity [98].

Many young females are evaluated clinically for premature adrenarche with mild hirsutism and
virilism. In response to intravenously administered ACTH, their ratios of D5 to D4 steroids often
rise >3 SD above the mean expected rise, prompting a diagnosis of ‘partial 3bHSD deficiency’.
However, these patients do not have mutations in their HSD3B2 genes; in fact, the ratios of D5
to D4 steroids must exceed 8 SD above the mean if one is to find HSD3B2 mutations [99].
However, why such individuals have hyper-responsive adrenal D5 steroids and why they have
hyperandrogenism remain unknown. An intriguing possibility is that the subcellular distribution
of a structurally normal 3bHSD2 enzyme may be different in the zona reticularis of such
individuals.

Potential Approaches
As with studies of adrenarche, research is limited by the lack of animal or cellular models.
Careful immunocytochemistry characterizing the subcellular localization of 3bHSD needs to be
done in cells under different conditions, in hormonally manipulated animals, and in human
pathologic specimens. Finding families with heritable ‘apparent partial 3bHSD deficiency’ and
searching for associated genes by genome-wide association studies (GWAS) or whole-exome
or whole-genome sequencing may be useful. It may be that patients with ‘apparent partial
3bHSD deficiency’ are simply a subgroup of patients with PCOS. GWAS studies of PCOS have
identified many candidate genes [100,101], but specific causal roles have yet to be proven for
any of these.

Genes without Diseases


Question 11: Genetic Deficiencies Are Not Known for Several Steroidogenic Enzymes and
Cofactors  Are These Hypothetical Disorders Compatible with Human Life?
Importance
Mouse knockouts and human genetic diseases have been essential indicators of the functions
of many genes. For example, finding that mutations of StAR cause lipoid CAH proved the
essential role of StAR two years before a mouse knockout was built. While it is possible that
failing to find mutations in some genes may mean that such mutations are incompatible with
human life, it is also possible that their mutation(s) may cause unexpected phenotypes, such
that the involvement of a particular gene is not obvious. For example, mouse knockouts of
POR cause embryonic lethality, but human mutations cause DSD with and without the
Antley–Bixler skeletal malformation syndrome [60]. Thus it is important to pursue animal
knockouts of as-yet unstudied genes and to search broadly when confusing or novel
phenotypes are encountered.

Trends in Endocrinology & Metabolism, November 2017, Vol. 28, No. 11 789
Background Outstanding Questions
Most steroidogenic disorders affect enzymes. Deficient enzyme activity typically leads to How does each molecule of StAR,
accumulation of the steroidal precursor and deficiency of the steroidal product; these abnormal while attached to the OMM, move
large numbers of cholesterol mole-
steroid concentrations are readily measured and lead to familiar clinical manifestations. Devel- cules to the IMM? What is the molec-
opment of steroid assays in the 1950s and 1960s led to the identification of deficient activities of ular itinerary of a cholesterol molecule
21-hydroxylase, 11-hydroxylase, 3bHSD, and 17-hydroxylase; aldosterone synthase and as it moves from cellular stores to the
IMM?
aromatase came a little later, but the logic was the same. Lipoid CAH is a key disease in this
history: the classical hormonal logic indicated that it was a disorder in the enzyme that converts
How is the 17,20 lyase activity of
cholesterol to pregnenolone, but when P450scc was cloned and studied, no mutations were P450c17 regulated? Does P450c17
found. I proposed that mutations in CYP11A1 would never be found because the placenta uses phosphorylation contribute to this
P450scc to make the progesterone required to suppress uterine contractility and permit term activity in normal physiology and/or
in hyperandrogenic states?
gestation [102]. I was spectacularly wrong! CYP11A1 mutations were found in 2001, and by
2017 over 30 cases had been reported [3]. Thus logic and steroid measurements may not
Is the backdoor pathway of androgen
accurately predict a genetic cause. synthesis an essential component of
normal androgen production? What
If complete absence of P450scc activity is compatible with human fetal development, why are the identities of all the enzymes
in this pathway? What are its
should not deficiencies of ferredoxin and ferredoxin reductase (Figure 7) also be compatible
disorders?
with life? To date, no human mutations and no mouse knockouts have been reported for these
factors, although a ferredoxin knockout has been reported in zebrafish [103]. Similarly, no What role(s) do newly described adre-
human mutations or animal knockouts have been reported for HSD3B1 or SRD5A1. A related nal androgens play in normal physiol-
problem is the frequent discordance between the findings in a mouse knockout and the ogy and hyperandrogenic states?

corresponding human disease. As noted in Question 7, POR is a prime example. A single report
What regulates fetal adrenal growth
describes Cyp17a1 knockout mice, which suffered preimplantation embryonic lethality [104], and involution? Do fetal adrenal ste-
whereas the human disease is well characterized and there is no known decrease in hetero- roids play any essential roles? What
zygote fertility. Perhaps the mouse embryonic lethality was caused by off-site effects, but no triggers the onset of adrenarche? Do
follow-up knockout studies have been done. Thus the results of mouse knockout studies must DHEA and/or DHEAS play any essen-
tial role in adults?
be viewed with caution unless reproduced by a second group with different vectors. For
example, mouse knockout of the peripheral diazepam receptor (PBR; subsequently renamed Do genetic polymorphisms in POR
the mitochondrial transporter protein, TSPO) was claimed to be embryonic lethal, but this was influence variations in steroidogenesis
only mentioned in passing in a 1997 review article, and the data were never reported. Nearly 20 and/or drug metabolism?
years later, multiple groups using different mouse knockout strategies proved that knockout of
Is the low incidence of 21OHD found in
PBR/TSPO had no detectable effect on mouse growth, reproductive capacity, or longevity [5–
African-Americans in Texas also found
7,105,106]. Thus a single unverified claim misled the steroidogenic community for 20 years! in other populations ancestrally
derived from or residing in sub-
Potential Approaches Saharan Africa? If so, what is the
genetic mechanism for this apparent
Careful, fully reported mouse knockout studies should be performed and reported for any factor
difference with populations of Euro-
that is thought to play a role in steroidogenesis. However, although necessary, mouse studies pean and Asian ancestry?
are not sufficient (see above). Thus, unusual patients such as those with DSD of unknown
cause need to be investigated with modern genetic technologies – whole-exome sequencing, Is the subcellular distribution of 3bHSD
competitive genome hybridization, and possibly whole-genome sequencing. In many cases the regulated? If so, how, and does it influ-
ence patterns of steroidogenesis in
data analysis is more challenging than the genetic technology. Mutations thought to be
different cell types and/or in disease?
causative must be studied by reproducing the mutant protein in an appropriate expression
system in vitro, followed by appropriate assays of the function of the protein. POR is the perfect There are many steroidogenic factors
example because the ‘standard’ facile assays with cytochrome c (a non-physiologic substrate) for which monogenic disorders have
give very different results than assays with cytochrome P450 enzymes [62]. Thus novel in vitro not been described or created in ani-
mal systems. Why do these hypotheti-
and in vivo model systems and mechanistic data are needed; clinical inferences are simply not cal diseases appear not to exist? What
sufficient in the 21st century. would be their phenotypes? What
would their global and conditional
Concluding Remarks knockouts look like in animal systems?

The study of steroidogenesis is more exciting than ever, and provides many research ques-
tions, as summarized in the Outstanding Questions. Some of these are amenable to clinical
investigation, some are genetic, others concern cell biology, and still others require detailed

790 Trends in Endocrinology & Metabolism, November 2017, Vol. 28, No. 11
biochemistry. Thus the study of steroidogenesis is not for the casual part-time investigator, or
for the faint of heart, because it requires the full array of contemporary biological, biochemical,
and medical technologies.
Imagination is more important than knowledge.
A. Einstein.

References
1. Miller, W.L. and Auchus, R.J. (2011) The molecular biology, 21. Auchus, R.J. (2004) The backdoor pathway to dihydrotestos-
biochemistry, and physiology of human steroidogenesis and terone. Trends Endocrinol. Metab. 15, 432–438
its disorders. Endocr. Rev. 32, 81–151 22. Kamrath, C. et al. (2012) Increased activation of the alternative
2. Miller, W.L. and Bose, H. (2011) Early steps in steroidogenesis: ‘backdoor’ pathway in patients with 21-hydroxylase deficiency:
intracellular cholesterol trafficking. J. Lipid Res. 52, 2111–2135 evidence from urinary steroid hormone analysis. J. Clin. Endo-
3. Miller, W.L. (2017) Disorders in the initial steps of steroid hor- crinol. Metab. 93, E367–E375
mone synthesis. J. Steroid Biochem. Mol. Biol. 165, 18–37 23. Flück, C.E. et al. (2011) Why boys will be boys. Two pathways of
4. Stocco, D.M. and Clark, B.J. (1996) Regulation of the acute fetal testicular androgen biosynthesis are needed for male sex-
production of steroids in steroidogenic cells. Endocr. Rev. 17, ual differentiation. Am. J. Hum. Genet. 89, 201–218
221–244 24. Biason-Lauber, A. et al. (2013) Of marsupials and men: ‘back-
5. Morohaku, K. et al. (2014) Translocator protein/peripheral ben- door’ dihydrotestosterone synthesis in male sexual differentia-
zodiazepine receptor is not required for steroid hormone bio- tion. Mol. Cell. Endocrinol. 371, 124–132
synthesis. Endocrinology 155, 89–97 25. Dhayat, N.A. et al. (2017) Androgen biosynthesis during mini-
6. Tu, L.N. et al. (2014) Peripheral benzodiazepine receptor/trans- puberty favors the backdoor pathway over the classic pathway:
locator protein global knock-out mice are viable with no effects on insights into enzyme activities and steroid fluxes in healthy
steroid hormone biosynthesis. J. Biol. Chem. 289, 27444–27454 infants during the first year of life from the urinary steroid metab-
olome. J. Steroid Biochem. Mol. Biol. 165, 312–322
7. Banati, R. et al. (2014) Positron emission tomography and
functional characterization of a complete PBR/TSPO knockout. 26. Marti, N. et al. (2017) Genes and proteins of the alternative
Nat. Commun. 5, 5452 steroid backdoor pathway for dihydrotestosterone synthesis
are expressed in the human ovary and seem enhanced
8. Stocco, D.M. et al. (2017) A brief history of the search for the
in the polycystic ovary syndrome. Mol. Cell. Endocrinol. 441,
protein(s) involved in the acute regulation of steroidogenesis.
116–123
Mol. Cell. Endocrinol. 441, 7–16
27. Rege, J. et al. (2013) Liquid chromatography–tandem mass
9. Prasad, M. et al. (2015) Mitochondria-associated endoplasmic
spectrometry analysis of human adrenal vein 19-carbon steroids
reticulum membrane (MAM) regulates steroidogenic activity via
before and after ACTH stimulation. J. Clin. Endocrinol. Metab.
steroidogenic acute regulatory protein (StAR)–voltage-depen-
98, 1182–1188
dent anion channel 2 (VDAC2) interaction. J. Biol. Chem. 290,
2604–2616 28. Storbeck, K.H. et al. (2013) 11b-Hydroxydihydrotestosterone
and 11-ketodihydrotestosterone, novel C19 steroids with
10. Prasad, M. et al. (2017) Mitochondrial metabolic regulation by
androgenic activity: a putative role in castration resistant pros-
GRP78. Sci. Adv. 3, e1602038
tate cancer? Mol. Cell. Endocrinol. 377, 135
11. Iaea, D.B. et al. (2017) Role of STARD4 in sterol transport
29. Swart, A.C. et al. (2013) 11b-Hydroxyandrostenedione, the
between the endocytic recycling compartment and the plasma
product of androstenedione metabolism in the adrenal, is
membrane. Mol. Biol. Cell 28, 1111–1122
metabolized in LNCaP cells by 5a-reductase yielding 11b-
12. Miller, W.L. and Tee, M.K. (2015) The post-translational regula- hydroxy-5a-androstanedione. J. Steroid Biochem. Mol. Biol.
tion of 17,20 lyase. Mol. Cell. Endocrinol. 408, 99–106 138, 132–142
13. Peng, H.M. et al. (2016) Cytochrome b5 activates the 17,20- 30. Swart, A.C. and Storbeck, K.H. (2015) 11b-hydroxyandroste-
lyase activity of human cytochrome P450 17A1 by increasing nedione: downstream metabolism by 11bHSD, 17bHSD, and
the coupling of NADPH consumption to androgen production. SRD5A produces novel substrates in familiar pathways. Mol.
Biochemistry 55, 4356–4365 Cell. Endocrinol. 408, 114–123
14. Pandey, A.V. et al. (2003) Protein phosphatase 2A and phos- 31. Turcu, A.F. et al. (2016) Adrenal-derived 11-oxygenated 19-
phoprotein SET regulate androgen production by P450c17. J. carbon steroids are the dominant androgens in classic 21-
Biol. Chem. 278, 2837–2844 hydroxylase deficiency. Eur. J. Endocrinol. 174, 601–609
15. Holien, J.K. et al. (2017) A homodimer model can resolve the 32. Imamichi, Y. et al. (2016) 11-Ketotestosterone is a major andro-
conundrum as to how cytochrome P450 oxidoreductase and gen produced in human gonads. J. Clin. Endocrinol. Metab.
cytochrome b5 compete for the same binding site on cyto- 101, 3582–3591
chrome P450c17. Curr. Protein Pept. Sci. 18, 517–519
33. Pretorius, E. et al. (2016) 11-Ketotestosterone and 11-ketodi-
16. Pandey, A.V. and Miller, W.L. (2005) Regulation of 17,20 lyase hydrotestosterone in castration resistant prostate cancer:
activity by cytochrome b5 and by serine phosphorylation of potent androgens which can no longer be ignored. PLoS
P450c17. J. Biol. Chem. 280, 13265–13271 One 11, e0159867
17. Tee, M.K. and Miller, W.L. (2013) Phosphorylation of human 34. du Toit, T. et al. (2017) Profiling adrenal 11b-hydroxyandroste-
cytochrome P450c17 by p38a selectively increases 17,20 lyase nedione metabolites in prostate cancer cells, tissue and plasma:
activity and androgen synthesis. J. Biol. Chem. 288, 23903–23913 UPC2-MS/MS quantification of 11b-hydroxytestosterone,
18. Bremer, A.A. and Miller, W.L. (2008) The serine phosphorylation 11keto-testosterone and 11keto-dihydrotestosterone. J. Ste-
hypothesis of polycystic ovary syndrome: a unifying mechanism roid Biochem. Mol. Biol. 166, 54–67
for hyperandrogenemia and insulin resistance. Fertil. Steril. 89, 35. Turcu, A.F. and Auchus, R.J. (2017) Clinical significance of 11-
1039–1048 oxygenated androgens. Curr Opin Endocrinol Diabetes Obes
19. Sondhi, V. et al. (2016) Impaired 17,20-lyase activity in male 24, 252–259
mice lacking cytochrome b5 in Leydig cells. Mol. Endocrinol. 30, 36. Yazawa, T. et al. (2008) Cyp11b1 is induced in the murine gonad
469–478 by luteinizing hormone/human chorionic gonadotropin and
20. Paris, D.B.B.P. et al. (2005) Birth of pouch young after artificial involved in the production of 11-ketotestosterone, a major fish
insemination in the Tammar wallaby (Macropus eugenii). Biol. androgen: conservation and evolution of the androgen meta-
Reprod. 72, 451–459 bolic pathway. Endocrinology 149, 1786–1792

Trends in Endocrinology & Metabolism, November 2017, Vol. 28, No. 11 791
37. Campana, C. et al. (2016) Development of a novel cell based 61. Huang, N. et al. (2008) Genetics of P450 oxidoreductase:
androgen screening model. J. Steroid Biochem. Mol. Biol. 156, sequence variation in 842 individuals of four ethnicities and
17–22 201 activities of 15 missense mutants. Proc. Natl. Acad. Sci.
38. Turcu, A.F. et al. (2017) 11-Oxygenated androgens are biomarkers U. S. A. 105, 1733–1738
of adrenal volume and testicular addrenal rest tumors in 21-hydrox- 62. Miller, W.L. et al. (2011) Consequences of POR mutations and
ylase deficiency. J. Clin. Endocrinol. Metab. 102, 2701–2710 polymorphisms. Mol. Cell. Endocrinol. 336, 174–179
39. Krone, N. et al. (2010) Gas chromatography/mass spectrometry 63. Pandey, A.V. et al. (2007) Modulation of human CYP19A1
(GC/MS) remains a pre-eminent discovery tool in clinical steroid activity by mutant NADPH P450 oxidoreductase. Mol. Endocri-
investigations even in the era of fast liquid chromatography nol. 21, 2579–2595
tandem mass spectrometry (LC/MS/MS). J. Steroid Biochem. 64. Dhir, V. et al. (2007) Differential inhibition of CYP17A1 and
Mol. Biol. 121, 496–504 CYP21A2 activities by the P450 oxidoreductase mutant
40. Wooding, K.M. and Auchus, R.J. (2013) Mass spectrometry A287P. Mol. Endocrinol. 21, 1958–1968
theory and application to adrenal diseases. Mol. Cell. Endocri- 65. Lv, J. et al. (2016) Effects of the selected cytochrome P450
nol. 371, 201–207 oxidoreductase genetic polymorphisms on cytochrome P450
41. Auchus, R.J. (2014) Steroid assays and endocrinology: best 2B6 activity as measured by bupropion hydroxylation. Pharma-
practices for basic scientists. Endocrinology 155, 2049–2051 cogenet. Genomics 26, 80–87
42. Mesiano, S. and Jaffe, R.B. (1997) Developmental and func- 66. Elens, L. et al. (2013) Impact of POR*28 on the clinical phar-
tional biology of the primate fetal adrenal cortex. Endocr. Rev. macokinetics of CYP3A phenotyping probes midazolam and
18, 378–403 erythromycin. Pharmacogenet. Genomics 23, 148–155
43. Goto, M. et al. (2006) In humans, early cortisol biosynthesis 67. Tan, S.L. et al. (2013) Cytochrome P450 oxidoreductase genetic
provides a mechanism to safeguard female sexual develop- polymorphisms A503 V and rs2868177 do not significantly
ment. J. Clin. Invest. 116, 953–960 affect warfarin stable dosage in Han-Chinese patients with
44. Asby, D.J. et al. (2009) The adrenal cortex and sexual differenti- mechanical heart valve replacement. Eur. J. Clin. Pharmacol.
ation during early human development. Rev. Endocr. Metab. 69, 1769–1775
Disord. 10, 43–49 68. Gomes, L.G. et al. (2008) The common P450 oxidoreductase
45. McMahon, S.K. et al. (2010) Neonatal complete generalized variant A503V is not a modifier gene for 21-hydroxylase defi-
glucocorticoid resistance and growth hormone deficiency ciency. J. Clin. Endocrinol. Metab. 93, 2913–2916
caused by a novel homozygous mutation in helix 12 of the ligand 69. Xiao, X. et al. (2015) Functional POR A503V is associated with the
binding domain of the glucocorticoid receptor gene (NR3C1). J. risk of bladder cancer in a Chinese population. Sci. Rep. 5, 11751
Clin. Endocrinol. Metab. 95, 297–302 70. Li, H. et al. (2016) Associations of cytochrome P450 oxidore-
46. Ben-David, S. et al. (2007) Parturition itself is the basis for fetal ductase genetic polymorphisms with smoking cessation in a
adrenal involution. J. Clin. Endocrinol. Metab. 92, 93–97 Chinese population. Hum. Genet. 135, 1389–1397
47. Agrawal, V. et al. (2015) Potential role of increased oxygen 71. Frank, D. et al. (2007) Evaluation of probe drugs and pharma-
tension in perinatal patterns of adrenal steroidogenesis. Pediatr. cokinetic metrics for CYP2D6 phenotyping. Eur. J. Clin. Phar-
Res. 77, 298–309 macol. 63, 321–333
48. Abbott, D.H. and Bird, I.M. (2009) Nonhuman primates as 72. Liu, Y.T. et al. (2007) Drugs as CYP3A probes, inducers and
models for human adrenal androgen production: function and inhibitors. Drug Metab. Rev. 39, 699–721
dysfunction. Rev. Endocr. Metab. Disord. 10, 33–42 73. Agrawal, V. et al. (2010) Substrate-specific modulation of
49. Bernstein, R.M. et al. (2012) Adrenal androgen production in CYP3A4 activity by genetic variants of cytochrome P450 oxido-
catarrhine primates and the evolution of adrenarche. Am. J. reductase (POR). Pharmacogenet. Genomics 20, 611–618
Phys. Anthropol. 147, 389–400 74. Tomalik-Scharte, D. et al. (2010) Impaired hepatic drug and
50. Maninger, N. et al. (2009) Neurobiological and neuropsychiatric steroid metabolism in congenital adrenal hyperplasia due to
effects of dehydroepiandrosterone (DHEA) and DHEA sulfate P450 oxidoreductase deficiency. Eur. J. Endocrinol. 163,
(DHEAS). Front. Neuroendocrinol. 30, 65–91 919–924
51. Ibáñez, L. et al. (2009) Clinical spectrum of premature pubarche: 75. Zhang, X. et al. (2011) Identification of cytochrome P450 oxido-
links to metabolic syndrome and ovarian hyperandrogenism. reductase gene variants that are significantly associated with the
Rev. Endocr. Metab. Disord. 10, 63–76 interindividual variations in warfarin maintenance dose. Drug
52. Nakamura, Y. et al. (2009) Adrenal changes associated with Metab. Dispos. 39, 1433–1439
adrenarche. Rev. Endocr. Metab. Disord. 10, 19–26 76. Tee, M.K. et al. (2011) Transcriptional regulation of the human
53. Hui, X.G. et al. (2009) Development of the human adrenal zona P450 oxidoreductase gene: hormonal regulation and influence
reticularis: morphometric and immunohistochemical studies of promoter polymorphisms. Mol. Endocrinol. 25, 715–731
from birth to adolescence. J. Endocrinol. 203, 241–252 77. Compagnone, N.A. and Mellon, S.H. (2000) Neurosteroids:
54. Huang, N. et al. (2005) Regulation of cytochrome b5 gene biosynthesis and function of these novel neuromodulators.
expression by Sp3, GATA-6, and NF-1 in human adrenal Front. Neuroendocrinol. 21, 1–58
NCI-H295A cells. Mol. Endocrinol. 19, 2020–2034 78. Slominski, A.T. et al. (2014) The role of CYP11A1 in the produc-
55. Missaghian, E. et al. (2009) Role of DNA methylation in the tion of vitamin D metabolites and their role in the regulation of
tissue-specific expression of the CYP17A1 gene for steroido- epidermal functions. J. Steroid Biochem. Mol. Biol. 144, 28–39
genesis in rodents. J. Endocrinol. 202, 99–109 79. Slominski, A.T. et al. (2015) On the role of skin in the regulation of
56. Pandey, A.V. and Flück, C.E. (2013) NADPH P450 oxidoreduc- local and systemic steroidogenic activities. Steroids 103, 72–88
tase: structure, function, and pathology of diseases. Pharmacol. 80. Nikolakis, G. et al. (2016) Skin steroidogenesis in health and
Ther. 138, 229–254 disease. Rev. Endocr. Metab. Disord. 17, 247–258
57. Nebert, D.W. et al. (2013) Human cytochromes P450 in health 81. Gomes, L.G. et al. (2009) Extra-adrenal 21-hydroxylation by
and disease. Philos. Trans. R. Soc. B 368, 20120431 CYP2C19 and CYP3A4: effect on 21-hydroxylase deficiency.
58. Zanger, U.M. and Schwab, M. (2013) Cytochrome P450 J. Clin. Endocrinol. Metab. 94, 89–95
enzymes in drug metabolism: regulation of gene expression, 82. Mellon, S.H. (2007) Neurosteroid regulation of central nervous
enzyme activities, and impact of genetic variation. Pharmacol. system development. Pharmacol. Ther. 116, 107–124
Ther. 138, 103–141 83. Kumar, S. et al. (2009) The role of GABAA receptors in the acute
59. Miller, W.L. (2005) Regulation of steroidogenesis by electron and chronic effects of ethanol: a decade of progress. Psycho-
transfer. Endocrinology 146, 2544–2550 pharmacology 205, 529–564
60. Flück, C.E. et al. (2004) Mutant P450 oxidoreductase causes 84. Porcu, P. et al. (2016) Neurosteroidogenesis today: novel tar-
disordered steroidogenesis with and without Antley–Bixler syn- gets for neuroactive steroid synthesis and action and their rele-
drome. Nat. Genet. 36, 228–230 vance for translational research. J. Neuroendocrinol. 28, 12351

792 Trends in Endocrinology & Metabolism, November 2017, Vol. 28, No. 11
85. Therrell, B.L., Jr et al. (1998) Results of screening 1.9 million 97. Rajapaksha, M. et al. (2016) An outer mitochondrial translocase,
Texas newborns for 21-hydroxylase-deficient congenital adre- Tom22, is crucial for inner mitochondrial steroidogenic
nal hyperplasia. Pediatrics 101, 583–590 regulation in adrenal and gonadal tissues. Mol. Cell. Biol. 36,
86. Gidlöf, S. et al. (2014) Nationwide neonatal screening for 1032–1047
congenital adrenal hyperplasia in Sweden. JAMA Pediatr. 98. Storbeck, K.H. et al. (2013) Cytochrome b5 modulates multiple
168, 567–574 reactions in steroidogenesis bydiverse mechanisms. J. Steroid
87. Tsuji, A. et al. (2015) Newborn screening for congenital adrenal Biochem. Mol. Biol. 151, 66–73
hyperplasia in Tokyo, Japan from 1989 to 2013: a retrospective 99. Lutfallah, C. et al. (2002) Newly proposed hormonal criteria via
population-based study. BMC Pediatr. 15, 209 genotypic proof for type II 3b-hydroxysteroid dehydrogenase
88. Silveira, E.L. et al. (2009) Molecular analysis of CYP21A2 deficiency. J. Clin. Endocrinol. Metab. 87, 2611–2622
can optimize the follow-up of positive results in newborn 100. Shi, Y. et al. (2012) Genome-wide association study identifies
screening for congenital adrenal hyperplasia. Clin. Genet. eight new risk loci for polycystic ovary syndrome. Nat. Genet.
76, 503–510 44, 1020–1025
89. Coeli-Lacchini, F.B. et al. (2013) A rational, non-radioactive 101. Louwers, Y.V. et al. (2013) Cross-ethnic meta-analysis of
strategy for the molecular diagnosis of congenital adrenal genetic variants for polycystic ovary syndrome. J. Clin. Endo-
hyperplasia due to 21-hydroxylase deficiency. Gene 526, crinol. Metab. 98, E2006–2012
239–245 102. Miller, W.L. (1998) Why nobody has P450scc (20,22 desmolase)
90. Nascimento, M.L. et al. (2014) Ten-year evaluation of a Neonatal deficiency. J. Clin. Endocrinol. Metab. 83, 1399–1400
Screening Program for congenital adrenal hyperplasia. Arq. 103. Griffin, A. et al. (2016) Ferredoxin 1b (Fdx1b) is the essential
Bras. Endocrinol. Metabol. 58, 765–771 mitochondrial redox partner for cortisol biosynthesis in zebra-
91. de Carvalho, D.F. et al. (2016) Molecular CYP21A2 diagnosis fish. Endocrinology 157, 1122–1134
in 480 Brazilian patients with congenital adrenal hyperplasia 104. Bair, S.R. and Mellon, S.H. (2004) Deletion of the mouse
before newborn screening introduction. Eur. J. Endocrinol. P450c17 gene causes embryonic lethality. Mol. Cell. Biol. 24,
175, 107–116 5383–5390
92. Therrell, B.L. et al. (2015) Current status of newborn screening 105. Selvaraj, V. et al. (2016) Crucial role reported for TSPO in viability
worldwide: 2015. Semin. Perinatol. 39, 171–187 and steroidogenesis is a misconception. Commentary: condi-
93. Bryc, K. et al. (2015) The genetic ancestry of African Americans, tional steroidogenic cell-targeted deletionof TSPO unveils a
Latinos, and European Americans across the United States. crucial role in viability and hormone-dependent steroid forma-
Am. J. Hum. Genet. 96, 37–53 tion. Front. Endocrinol. 7, 91
94. Tishkoff, S.A. et al. (2009) The genetic structure and history of 106. Wang, H. et al. (2016) Global deletion of TSPO does not affect
Africans and African Americans. Science 324, 1035–1044 the viability and gene expression profile. PLoS One 11,
95. 1000 Genomes Project Consortium (2012) An integrated map e0167307
of genetic variation from 1,092 human genomes. Nature 491, 107. Miller, W.L. (2012) The syndrome of 17,20 lyase deficiency. J.
56–65 Clin. Endocrinol. Metab. 97, 59–67
96. Cherradi, N. et al. (1997) Submitochondrial distribution of 108. Gupta, M.K. et al. (2003) 5a-Reduced C21 steroids are sub-
three key steroidogenic proteins (steroidogenic acute regulatory strates for human cytochrome P450c17. Arch. Biochem. Bio-
protein and cytochrome P450scc and 3b-hydroxysteroid phys. 418, 151–160
dehydrogenase isomerase enzymes) upon stimulation by intra- 109. Endo, T. et al. (2009) Truncated form of tenascin-X, XB-S,
cellular calcium in adrenal glomerulosa cells. J. Biol. Chem. 272, interacts with mitotic motor kinesin Eg5. Mol. Cell. Biochem.
7899–7907 320, 53–66

Trends in Endocrinology & Metabolism, November 2017, Vol. 28, No. 11 793

You might also like