JOURNAL OF PATHOLOGY, VOL.

1 7 9 283-287 (1996)

LOCALIZATION OF ERYTHROPOIETIN GENE

EXPRESSION IN PROXIMAL RENAL TUBULAR
CELLS DETECTED BY DIGOXIGENIN-LABELLED
OLIGONUCLEOTIDE PROBES
JONATHAN H. SHANKS*, CLAIRE M. HILL*, TERENCE R. J. LAPPIN? AND A. PETER MAXWELLS

Departments of *Pathology and THaematology, The Queen's University of Belfast, Royal Group of Hospitals Trust and
$Regional Nephrology Unit, Belfast City Hospital, Belfast, Northern Ireland, U.K.

SUMMARY
Erythropoietin (EPO) is the main humoral stimulus of erythropoiesis. In adult mammals, the kidney releases EPO in response to
hypoxic stress. Conflicting data have suggested either renal tubular or peritubular cell origins of EPO synthesis in viva Zn situ
hybridization studies were performed to define further the kidney cell type@) capable of increasing EPO gene expression during hypoxic
stimulation. EPO gene expression was stimulated in mice exposed to acute hypobaric hypoxia. Kidneys from hypoxic and control
normoxic mice were obtained. Six digoxigenin-labelled oligonucleotide probes complementary to EPO exon sequences were utilized for
in situ hybridizationfor EPO messenger RNA. Positive hybridization signals were identified in some proximal tubular cells, confined to
the inner third of the renal cortex of hypoxic mouse kidney.
KEY WoRDS-erythropoietin; gene expression; in situ hybridization; digoxigenin; hypoxia; renal tubular cell

INTRODUCTION EPO, both labelled with 32P,in an anaemic sheep model

Erythropoietin (EPO), a glycoprotein hormone, is the
main regulator of red cell synthesis. Plasma EPO levels
increase in response to anaemia or hypoxemia and
stimulate erythropoiesis. In adult mammals the major
source of EPO is the kidney, but there is conflicting
experimental evidence concerning the identity of the
kidney cells responsible for the production of EPO.
Earlier immunohistochemical studies to identify EPO-
producing cells were unsuccessful because the antisera
employed had been raised against impure EPO prepar-
ations and the antibodies thus lacked specificity.
Immunohistochemical methods, even with currently
available highly specific antibodies, cannot discriminate
between endogenous production of EPO by cells versus
cellular uptake of EPO protein.
In situ hybridization detects messenger RNA (mRNA)
directly and has been used to identify the cellular origin
of EPO. Lacombe et at.' and Koury et aL2 found EPO
mRNA in interstitial/peritubular cells of the renal cortex
from mice rendered anaemic by bleeding or by treatment
with phenylhydrazine and irradiation. In both studies,
35S-labelled probes derived from EPO genomic
sequences were used. Kurtz et aL3 used a 35S-labelled
cDNA probe to detect EPO mRNA in the kidneys of
rats treated with 0.1 per cent carbon monoxide and
Bachmann et d4used a digoxigenin-labelled probe in an
anaemic rat model; both groups reported that EPO gene
expression was confined to peritubular cells. More
recently, Darby et ~ 1 have . ~ employed synthetic oligo-
nucleotide probes and a riboprobe (cRNA) for ovine
Addressee for correspondence: Dr J. H. Shanks, Department of
Histopathology, Royal Preston Hospital, Preston PR2 4HG, U.K.
CCC 0022-341 7/96/070283-05
01996 by John Wiley & Sons, Ltd.
and found that EPO gene expression was confined to
peritubular cells. In contrast, three groups6-8 have
reported the detection of EPO mRNA in tubular cells of
the renal cortex. Smith et aL6 used cobalt to stimulate
EPO synthesis in rats and detected EPO gene expression
in tubular cells using a 35S-labelledgenomic EPO probe.
Rich et aL7 also detected EPO mRNA using a 3sS
genomic probe in mice previously subjected to bleeding.
Our group (Maxwell et d 8 )similarly found EPO
mRNA in tubular cells using a 32P-labelled oligo-
nucleotide probe in the kidneys of mice subjected to
hypobaric hypoxia.
In our previous study8 frozen sections had been used,
resulting in sub-optimal morphology; in addition, the
use of 32P, a high energy p-emitter, made it difficult to
determine the exact cellular location of the signal. In the
present study we chose digoxigenin-labelled oligo-
nucleotides. This was an attempt to provide better
cellular localization of the gene product, which was also
helped by using paraffin sections.
MATERIALS AND METHODS
Animals
Male C57/BL6 mice approximately 3 months old and
weighing 20-25 g were used.
Oligonucleotide probes
Six 25-mer oligonucleotide probes complementary to
exon sequences of the murine EPO gene (Fig. 1) were
synthesized by R & D Systems Europe Ltd., Abingdon,
Received 3 August 1995
Accepted 8 February 1996
284 J. H. SHANKS ET A L .

A SEQUENCES COMPLEMENTARYTO SIX OLIGONUCLEOTIDES n

OF PROBE COCKTAIL FOR EPO mRNA (each 25 - mer) :- zrc

Oligo I II 111 IV v VI

Exon I II Ill IV v

B OLIGONUCLEOTIDE SEQUENCES :- ZE3-I

Probe Base numbers Sequence flcq

I 1242-1266 CTGGGAGGCCCAGAGGAATCAGTAG
II 1292-1316 CCTCTCCAGAACTCGACTGTCGCAG
Ill 1817-1841 CAGTCTGGGACCTTCTGCACAACCC
IV 231 1-2335 GAAGGGTCTCTGGTGGCTGGGAGGA
V 3173-3197 TTCGGAGTGGAGCAGGTGGGGTGGT
VI 3277-3301 CCTGTCCCCTCTCCTGCAGACCTCT
Fig. I-(A) Schematic diagram of the mouse EPO gene. The five exons are shaded. Open boxes represent the 5' and 3'
untranslated regions. (B) The oligonucleotide probe sequences

Oxfordshire. The probes were 5'-end-labelled with
digoxigenin and purified by HPLC.
In situ hybridization
Eighteen mice were used in total. In each experiment,
two mice were subjected to hypobaric hypoxia (0.5
atmospheres) in a chamber for 6 h, for comparison with
one normoxic control mouse. Kidneys were promptly
fixed in 4 per cent buffered paraformaldehyde for 24 h at
4°C and 4 p m thick parafin sections were prepared.
They were treated with proteinase K (33pgfml) at 37°C
for 30 min. Following prehybridization, the sections
were hybridized in a l'ormamide solution with a probe
mixture consisting of all six 25-mer oligonucleotides
5'-end-labelled with digoxigenin. For direct immuno-
histochemical detection, anti-digoxigenin alkaline
phosphatase (Boehringer-Mannheim, Lewes, Sussex)
was prediluted 1:lOO in a solution of 3 per cent (w/v)
bovine serum albumin in Tris-buffered saline. Incu-
bation was at room temperature for 30 min. Following
three wash steps, colour substrate for alkaline
phosphatase consisting of nitroblue tetrazolium chloride
(NBT) and S-bromo-4-chloro-3-indolyl phosphate
(BCIP) was added. A brown-black coloured reaction
product was allowed to develop in a light-proof humidi-
fied chamber for 4-6 h. The sections were mounted
in aqueous mountant and photographed immediately
where appropriate.
Negative controls were employed as follows: (i)
kidney sections from normoxic mice; (ii) hypoxic kidney
sections with the omission of probe in the hybridization
step; (iii) kidney sections digested with RNAse prior to
hybridization; (iv) kidney sections incubated without
antibody in the detection step. No other tissues con-
tained a high abundance of EPO m R N A , so a positive
tissue control was unavailable. As a positive procedural
control, ribosomal (r) RNA was detected with a mouse/
human rRNA oligonucleotide probe (a gift from Dr
J. H. Pringle, University of Leicester), employing the
same protocol as for EPO mRNA in situ hybridization.
RESULTS
Signal was detected after 6 h of hypoxia and was
confined predominantly to the inner third of the renal
cortex (Fig. 2). Positive cytoplasmic staining was present
as a diffuse light brown precipitate in a sub-population
of tubular epithelial cells (Fig. 3). No signal was evident
in the cell nuclei or in the epithelial brush borders. Those
tubules which contained positively staining cells had a
tall brush border and had the morphological character-
istics of proximal tubules. Interestingly, not all of the
inner cortical proximal tubules contained positive cells.
Distal tubular cells and peritubular cells had no signifi-
cant positive staining. Although there was some initial
variability in signal intensity between different test
 CELLULAR LOCALIZATION OF ERYTHROPOIETIN 285

Fig. 2-Junction of the inner renal cortex and medulla in ex-hypoxic mouse kidney following hybridization
with the EPO mRNA probe cocktail. Hybridization signal is present in the cortical tubules (BCIP/NBT
following anti-digoxigenin AP)
Fig. 3-Higher-power view of the inner renal cortex in ex-hypoxic mouse kidney following in situ
hybridization with the EPO probe cocktail. A sub-population of proximal tubular cells shows cytoplasmic
staining (BCIP/NBT following anti-digoxigenin AP)
Fig. &Negative controls. (A) Junction of the inner renal cortex and medulla in ex-hypoxic mouse kidney
following hybridization without the erythropoietin mRNA probe cocktail (BCIPINBT following anti-
digoxigenin AP). (B) Higher-power view of the inner renal cortex from nomoxic mouse kidney following
in situ hybridization with the erythropoietin probe cocktail (BCIP/NBT following anti-digoxigenin AP)
286 J. H. SHANKS ET A L
mice, further optimization of the in situ hybridization
technique, particularly proteinase K digestion, led to
consistent and reproducible experimental results.
No specific pattern of hybridization was detected
using the kidney sections from the negative controls
(Figs 4a and 4b). The absence of signal in sections
incubated with no antibody, but with the inclusion of
colour substrate in the detection step, indicates that
endogenous alkaline phosphatase was inactive. The
positive procedural control, utilizing the rRNA oligo-
nucleotide probe, resulted in stronger cytoplasmic stain-
ing of tubular cells, as expected, due to their abundant
mitochondria (data not shown).
DISCUSSION
The present results suggest that EPO mRNA can be
localized to a sub-population of proximal tubular cells in
the renal cortex of the ex-hypoxic mouse kidney. These
cells appear to be the main location of EPO mRNA in
this animal model, although it is possible that low-level
gene expression is present in peritubular cells below the
threshold for detection by this in situ hybridization
protocol.
The absence of renal proximal tubular brush border
staining is consistent with a lack of endogenous alkaline
phosphatase activity which could have confounded the
interpretation. If this enzyme had been active in the
sections, it would have been expected to be localized
mainly to the brush borders of the proximal tubular
cells.9 Exposure to heating and formamide during
processing of tissue sections would be expected to
inactivate all enzymatic activity. l o The localization of
EPO to the renal tubular cells accords with some,6-8 but
not other investigation^.'-^ The localization of EPO
mRNA to the tumour cells in renal carcinomas which
secrete the hormone'' could be taken as evidence to
support the tubular origin for EPO. However, it is well
known that deregulation of gene expression can result in
ectopic hormone production in a variety of tumour
types.
It is possible that more than one type of cell within the
kidney is capable of EPO production, analogous to the
situation in the liver.12 l4 Maxwell et ~ 1 .have
' ~ reported
localization of the EPO-producing cells in the liver to
both hepatocytes and non-parenchymal cells using
SV 40 large T antigen as a transgenic marker in mice.
The non-parenchymal cells involved in EPO gene
transcription within the liver were reported to be Ito
cells, which have a number of functions including fat
storage. l 5
Different probe types and probe-labelling methods
have been cited as possible reasons for the differing
results in the work reported. It is also possible that
variations in the rate or mechanism of application of the
hypoxic stimulus could play some part in determining
which cells predominate in expression of the EPO
gene.16 Some studies have used agents such as phenyl-
hydrazine to induce haemolysis, thereby causing an
anaemic hypoxic stimulus. The possibility that the
haemolysis so induced has nephrotoxic effects which
could influence results has not been systematically
investigated.
The use of transgenic mice to localize gene expres-
sion17 is a methodological approach fundamentally
distinct from in situ hybridization. Transgenic animals
can also be used to investigate the sensitivity of in situ
hybridization protocols.I8,l9 Some of these techniques
have been applied to localization of EPO gene expres-
sion.12,20,21 Maxwell et ~ 1reported
. ~ on ~ transgenic mice
bearing a construct containing the SV 40 large T antigen
and the 5' untranslated region of the mouse EPO gene.
Expression was inducible by the anaemia caused by
phenylhydrazine administration and was localized, using
immunohistochemistry for SV 40 large T antigen, to a
population of cells in the interstitium of the renal cortex
and outer medulla. Electron microscopy was performed
and these cells were identified as 'interstitial fibroblast-
like cells'. It was concluded that renal fibroblast-like
cells produce erythropoietin.22
The most recent transgenic study designed to localize
EPO-producing cells in the kidney has been reported
by Loya et a1.,21 using a lac Z reporter gene in the
transgenic construct. Hypoxia-inducible expression of
lac 2 was confined to the proximal tubules but was not
seen in any other renal cells, including peritubular cells.
It was also seen to a lesser extent in hepatocytes, but
only under hypoxic conditions.
The various models which have been used to stimulate
EPO gene expression differ both in the speed of onset
and in the duration of the hypoxic insult. The present
study suggests that a sub-population of proximal tubu-
lar cells is responsible for restricted EPO gene expression
following the onset of hypobaric hypoxia. The data
derived from application of a non-radioactive in situ
hybridization method support other reports of a renal
tubular cell origin of EPO synthesis.6-8.2'The peritubu-
lar or interstitial population of cells recruited to produce
EPO may require a more prolonged period of hypoxia
prior to induction of EPO gene expression. The differ-
ences in cellular location of EPO production reported
thus far might have a parallel in the embryological
switch of EPO production from fetal liver tissue to
infant and adult renal tissue.
Multiple cell types are capable of producing EPO, but
the regulation of EPO gene expression appears to be
dependent on the cellular developmental stage and its
relation to the timing and possibly the type of inciting
stimulus. Further characterization of the cis- and tuans-
acting factors involved in the sensing of oxygenation will
be required to understand why EPO gene expression has
been identified in a number of renal and hepatic cell
types.
REFERENCES
Lacombe C, Da Silva J-L, Bruneval P, el ( I / . Perituhular cells are the site of
erythropoietin synthesis in the murine kidney. J C'lm lnvest 1988; 81:
620-623.
Koury ST, Bondurant MC, Koury MJ. Localization of erythropoietin
synthesizing cells in murine kidney by in situ hybridization. Blood 1988; 71:
524-521.
Kurtz A, Eckardt K-U, Neumann R, Kaissling B, Le Hir M , Bauer C. Site
of erythropoietin formation. In: Baldamus CA, Scigalla P, Wieczorek L,
Koch K M , eds. Erythropoietin: From Molecular Structure to Clinical
Application. Contributions to Nephrology. Basel: Karger, 1989; 76: 1 4 2 3 .
 CELLULAR LOCALIZATION OF ERYTHROPOIETIN
4. Bachman S, Le Hir M , Eckhardt K-U. Co-localisation of erythropoietin
mRNA and ecto-5'-nucleosidase immunoreactivity in peritubular cells of rat
renal cortex indicates that fibroblasts produce erythropoietin. J Hi.stoehm
Cj~or~hrm 1993; 41: 335-341.
5. Darby lA, Evans BA, Fu P, Lim GB, Moritz KM, Wintour EM. Erythro-
poietin gene expression in fetal and adult sheep kidney. Br J Huematoll995;
8 9 266-270.
6. Smith DH, Maples PB, Beru N, Goldwasser E. Identification of
erythropoietin-producing cells in mammalian kidney by in .rim hybridir-
ation. Blood 1986; 6 8 170A.
7. Rich IN. Not G, Vogt C, Lemke C. Erythropoietin is produced in the
tubules of the renal cortex (Abstract). Blood 1990; I(Supp1): 76.
8. Maxwell AP, Lappin TRJ, Johnston CF. Bridges JM, McGeown MG.
Erythropoietin production in kidney tubular cells. Br J Ifaemutol 1990; 74
535 539.
9. Verpooten GF, Nuyts GD, Hoylaerts MF, PI a/. lmmunoassay in urine of a
specific marker for proximal tubular S3 segment. C h Chem 1992; 38:
642 647.
10. Farquharsou M, Harvie R, Kennedy A, McNicol AM. Detection of mRNA
by in siru hybridisation and in Northern blot analysis using oligonucleotide
probes labelled with alkaline phosphatase. J Clin Pathol1992; 4 5 999-1002.
I I . Da Silva J-L, Lacombe C, Bruneval P, PI crl. Tumour cells are the site of
erythropoietin synthesis in human renal cancers associated with poly-
cythemia. Blood 1990; 75: 577-582.
12. Koury ST, Bondurant MC, Koury MJ, Semenra GL. Localization of cells
producing erythropoietin in murine liver by in .S(IU hybridization. Blood
1991; 77: 2497-2503.
13. Maxwell PH, Ferguson DJP, Osmond MK, et NI. Expression of a homolo-
gously recombined erythropoietin-SV40 T antigen fusion gene in mouse
liver: evidence for erythropoietin production by Ito cells. Blood 1994; 84
1823-1 830.
287
14 Schuster SJ, Koury ST, Bohrer M, Salceda S, Caro J . Cellular sites of
extrarenal and renal erythropoietin production in anaemic rats. BY J
Huematol 1992, 81: 153-159.
15 Brouwer A, Wisse E, Knook DL. Sinusoidal endothelial cells and perisinu-
soidal fat-storing cells. In: Arias IM, Jakoby WB, Popper H, Schacter D.
Shafritz DA, eds. The Liver: Biology and Pathobiology. 2nd edn. New
York: Raven Press, 1988; 665-682.
16 Eckhardt K-U, Koury ST, Tan CC, ef a/.Distribution of erythropoietin
producing cells in rat kidneys during hypoxic hypoxia. K i d n q Int 1993: 43:
815-823.
17 Jaenisch R. Transgenic animals. Science 1988; 240 1468-1474.
18 Lo CW. Localization of low abundance DNA sequences in tissue sections
by in situ hybridisation. J Cell Sci 1989; 81: 143-162.
19 Fleming KA, Evans M, Ryley C, Franklin D, Lovell-Badge RH, Morey AL.
Optimization of non-isotopic in situ hybridization on formalin-fixed
paraffin-embedded material using digoxigenin-labelled probes and
transgenic tissues. J PaIhol 1992; 167: 9- 17.
20 Porter DL, Goldberg MA. Regulation of erythropoietin production. Exp
Henratol 1993; 21: 399404.
21 Loya F, Yang Y, Lin H, Goldwasser E, Albitar M. Transgenic mice
carrying the erythropoietin gene promoter linked to IurZ express the
reporter in proximal convoluted tubule cells after hypoxia Blood 1994 84:
1831-1836.
22 Maxwell PH, Osmond MK, Pugh CW, et ul. Identification of the renal
erythropoietin-prodncing cells using transgenic mice. Kidney Znr 1993; 44.
1149-1 162.