Báo cáo khoa học: The HNF1b transcription factor has several domains involved in nephrogenesis and partially rescues Pax8/lim1-induced kidney malformations docx

The 26 aa segment between the POU S and POU H domains of the human HNF1a and HNF1b proteins as well as of the human and Xenopus HNF1b protein are aligned bottom with missing amino acids

The HNF1b transcription factor has several domains involved

in nephrogenesis and partially rescues Pax8/lim1-induced

kidney malformations

Guizhi Wu, Silvia Bohn and Gerhart U Ryffel

Institut fu¨r Zellbiologie, Universita¨tsklinikum Essen, Germany

The tissue-specific transcription factors HNF1a and HNF1b

are closely related homeodomain proteins conserved in

vertebrate evolution Heterozygous mutations in human

HNF1b

1 but not in HNF1a genes are associated with kidney

malformations Overexpression of HNF1b in Xenopus

embryos leads to defective pronephros development, while

HNF1a has no effect We have defined the regions

respon-sible for this functional difference between HNF1b and

HNF1a in transfected HeLa cells as well as in injected

Xenopusembryos Using domain swapping experiments, we

located a nuclear localization signal in the POUHdomain of

HNF1b, and showed that the POUSand POUHdomains of

HNF1b mediate a high transactivation potential in

trans-fected cells In injected Xenopus embryos three HNF1b

domains are involved in nephrogenesis These include the

dimerization domain, the 26 amino acid segment specific for

splice variant A as well as the POUHdomain As HNF1b

together with Pax8 and lim1 constitute the earliest regulators

in the pronephric anlage, it is possible that they cooperate during early nephrogenesis We have shown here that HNF1b can overcome the enlargement and the induction of

an ectopic pronephros mediated by overexpression of Pax8 and lim1 However, the phenotype induced by Pax8 and lim1 overexpression and characterized by cyst-like structures and thickening of the pronephric tubules was not altered by HNF1b overexpression Taken together, HNF1b acts ant-agonistically to Pax8 and lim1 in only some processes during nephrogenesis, and a simple antagonistic relationship does not completely describe the functions of these genes We conclude that HNF1b has some distinct morphogenetic properties during nephrogenesis

Keywords: HNF1b; lim1; nephrogenesis; Pax8; pronephros

The tissue-specific transcription factors, HNF1a (TCF1)

and HNF1b (vHNF1, TCF2), are two unique

homeo-domain proteins [1] The POU homeohomeo-domains (POUH) are

divergent from other homeodomain proteins in that they

contain an extra 21 amino acid (aa) loop between helices 2

and 3 [2,3] Both transcription factors are encoded in distinct

genes on separate chromosomes, and are highly conserved

in vertebrates with homologues in fish [4,5], frog [6,7]

and mammals, including humans [8–10] The evolutionary

conservation is also seen in the exon/intron patterning

which remains essentially the same between Xenopus and

mammals [11] Both HNF1 proteins contain a highly

conserved N-terminal dimerization domain, a bipartite

DNA binding region and a more divergent C-terminal

transactivation domain (Fig 1) Based on the crystal

structure of the dimer, the dimerization domain has been

identified as an intertwined four-helix bundle that allows the

formation of homo- or heterodimers of the HNF1 proteins

[12,13] The DNA binding domain is composed of a POU

specific domain (POUS) and the divergent POU homeo-domain (POUH) Recent three-dimensional structural ana-lysis of the HNF1a protein indicates that the POUSdomain interacts with the 21 aa loop of the POUHdomain to create

a stable interface between the two DNA binding domains This feature distinguishes HNF1a from other, more flexible, POUHfactors [14] As the primary structures of HNF1a and HNF1b are very similar within the DNA binding region, it is reasonable to assume that this structure is also present in the HNF1b protein Depending on the splice variant, there is a 26 aa insertion between the POUSand POUHdomain in the HNF1b protein This variant is found

in mammalian and also Xenopus HNF1b proteins (Fig 1), but never in the HNF1a proteins In contrast to these rather conserved domains, the C-terminal transactivation domain

is the most divergent protein area when the HNF1a and HNF1b proteins are compared

It is not resolved whether the differences between the HNF1a and HNF1b proteins that are highly conserved throughout vertebrate evolution reflect distinct functions Consistent with distinct functional roles, the temporal and spatial expression patterns of HNF1a and HNF1b differ significantly During murine embryogenesis, HNF1a is expressed in the yolk sac endoderm at day 8.5 of gestation as well as in the developing liver, kidney, intestine, pancreas and stomach [15–17] In contrast, HNF1b is expressed earlier in the primitive and visceral endoderm Starting at day 4.5 of gestation, the anterior part of the neural tube as

Correspondence to G U Ryffel, Institut fu¨r Zellbiologie

(Tumor-forschung), Universita¨tsklinikum Essen, D-45122 Essen, Germany.

Fax: +49 201723 5905, Tel.: +49 201723 3110,

E-mail: gerhart.ryffel@uni-essen.de

Abbreviations: NLS, nuclear localization signal; POU S , POU specific

domain; POU H , POU homeodomain.

(Received 2 June 2004, revised 22 July 2004, accepted 29 July 2004)

well as the developing kidney, liver, gut and pancreas

express HNF1b [18] Additionally, HNF1b is also expressed

in the primordia for the genitalia and the lung HNF1b

expression persists in these organs in the adult, whereas

HNF1a is never active in these tissues [19,20]

The embryonic expression pattern of the HNF1 proteins

is evolutionarily conserved in vertebrates The expression of

HNF1b occurs prior to HNF1a in Xenopus embryos

[21,22], and only HNF1b is expressed in the developing

brain in Xenopus [7] as well as in zebrafish [5] In agreement

with the differential embryonic expression patterns of the

two HNF1 proteins, inactivation of the corresponding

genes in the mouse has different effects Homozygous

knock-out of the HNF1b gene led to early embryonic

lethality at day 7.5 of gestation with poorly organized

ectoderm and no discernible visceral endoderm [18,23] In

contrast, HNF1a was not required for embryonic

devel-opment, but HNF1a-deficient mice died during postnatal

life due to hepatic, pancreatic and renal dysfunction

[24–28] These results clearly establish different roles for

the two HNF1 genes Whether differential properties of the

two transcription factors are the cause of these differences,

or rather the differential expression patterns, remains to be

seen A functional equivalence of the HNF1a and HNF1b

protein has recently been shown in embryonic stem cells, as the introduction of HNF1a restores the formation and differentiation of a mature visceral endoderm in HNF1b-deficient embryonic stem cells [29] Further support for functional differences can be deduced from human dis-eases Biallelic inactivation of the HNF1a gene has been described as an early step in hepatocellular carcinoma [30] However, HNF1b has not been associated with tumori-genesis to date Heterozygous mutations in both genes lead

to maturity onset diabetes of the young but HNF1b mutations are additionally associated with severe nondia-betic renal defects as well as genital malformations in females [31–34] In this context, we showed the specific role

of HNF1b during development of the first form of vertebrate kidney, the pronephros, using overexpression experiments in Xenopus embryos The expression of HNF1b led specifically to a reduced formation of the pronephros, whereas HNF1a had no effect [35] This indicates that these two transcription factors have different intrinsic biochemical properties Most recently, the renal-specific inactivation of the HNF1b gene in mice [36] and the kidney-specific expression of mutated HNF1b [37] have linked the HNF1b transcriptional network to genes causing polycystic kidney disease

Fig 1 The related human transcription factors, HNF1a and HNF1b HNF1b and HNF1a are represented schematically (top) The domains are indicated and numbers below the domains refer to the amino acid positions Amino acid identity of the domains between HNF1a and HNF1b is shown in bold (homology)

16 The 26 aa segment between the POU S and POU H domains of the human HNF1a and HNF1b proteins as well as of the human and Xenopus HNF1b protein are aligned (bottom) with missing amino acids indicated by ‘ )’ The 26 aa segment deleted in the B splice variant of the HNF1b is indicated (green) Identical amino acids between b and a or human b and Xenopus b sequences are shown and

amino acid changes are indicated by +.

In vertebrates, three distinct types of kidneys

(pro-nephros, mesonephros and metanephros) are formed

progressively during development [38] Similar regulators

are expressed in all three kidneys, and thus, the molecular

processes by which the different kidneys develop appear to

be closely related [39–41] The pronephros is the simplest

vertebrate kidney, and consists of a single nephron with an

external glomus It represents an attractive system to study

molecular events during kidney development, as several

key regulators have been functionally identified by

inject-ing mRNA into Xenopus embryos [41,42] Usinject-ing the

Xenopus system, we have shown that overexpression of

human HNF1b in the developing frog embryo leads to

agenesis of the pronephric tubules and duct The same

phenotype is seen for some human HNF1b mutants

leading to defective renal development, whereas an

enlargement of the pronephros occurs with other mutants

[35,43] An enlargement of the pronephros has also been

observed by the overexpression of the transcription factors,

Pax8 and lim1, and this effect was additive [44]

Further-more, the artificial expression of Pax8 and lim1 in the

Xenopusembryo induced ectopic pronephric structures, a

phenotype never seen in embryos overexpressing HNF1b

Interestingly, HNF1b, Pax8 and lim1 are the earliest

known regulators in the pronephric anlage, implying that

they may cooperate during early events of nephrogenesis

[41,42]

In the present communication, we functionally mapped

the protein domains of HNF1b, specifically participating in

nephrogenesis via injection of chimeric HNF1a and HNF1b

proteins into Xenopus embryos We also explored whether

Pax8- and lim1-mediated effects can be overcome by

simultaneous HNF1b overexpression

Materials and methods

Plasmid constructions

CMVHNF1a expression vectors have been described

previously [35] HNF1aaa and HNF1bbb were generated

by inserting an EcoRI-XbaI fragment encoding 1–321 aa

of the human HNF1a and 1–352 aa of the human

HNF1b, respectively A BamHI site was introduced both

at G69 (a) and G79 (b) without changing the amino acid

sequence The EcoRI-BamHI and BamHI-XbaI

frag-ments were derived from PCR products made with the

following primers HNF1aaa:

TTTCTAAACTGAGCC-3¢ (forward), 5¢-CGCGGATCC

CCGAGTCTCCCCC-3¢ (reverse); 5¢-CGCGGATCCGA

GGACGAGACGG-3¢ (forward), 5¢-GCTCTAGATTA

GCGCACACCGTGGAC-3¢ (reverse); HNF1bbb: 5¢-CG

GAATTCAATGGTGTCCAAGCTCACGT-3¢ (forward),

5¢-CGCGGATCCCTCGTCGCCGGACAA-3¢ (reverse);

5¢-CGCGGATCCGAGGACGGCGACGA-3¢ (forward),

(re-verse) The restriction sites for cloning are underlined

HNF1abb and HNF1baa were generated by exchanging

the EcoRI-BamHI fragments between HNF1bbb and

HNF1aaa HNF1bbbD was generated by replacing the

BamHI-HincII fragment of the HNF1bbb expression vector

with the BamHI-HincII fragment of a PCR product

made with the forward primer, 5¢-CGCGGATCCGA GGACGGCGACGA-3¢, and the reverse primer, 5¢-GCT CTGTTGACTGAATTGTCGGAGGATCTCTCGT-3¢, containing complementary sequences upstream and down-stream to a segment encoding the 26 aa to be deleted HNF1bD was generated by replacing the PvuI fragment encoding 1–251 aa of HNF1b with the corresponding fragment of HNF1bbbD

HNF1aab, HNF1aabins26 and HNF1aaains26 con-structs were generated using the Quickchange site-directed Mutagenesis Kit (Stratagene) and a PCR fragment gener-ated from the HNF1bbb sequence using the following primers: HNF1aab: 5¢-GATGAGCTACCAACCAAGAA GATGCGCCGCA-3¢ (forward), 5¢-GCCGCTCTAGATT AGCGCACTC-3¢ (reverse); HNF1aabins26: 5¢-CGAGA GGTGGCGCAGCAGTTCAACCAGACAGTCCAG-3¢ (forward), 5¢-GCCGCTCTAGATTAGCGCACTC-3¢ (re-verse); HNF1aaains26: 5¢-CGAGAGGTGGCGCAGCA GTTCAACCAGACAGTCCAG-3¢ (forward), 5¢-CTCC CTGCCCTGCATGGGTGAACTCTGGAAAGAGAA AC-3¢ (reverse)

3 HNF1aabH and HNF1aabHS were generated by repla-cing the BamHI-XbaI fragment of HNF1aab with the BamHI-XbaI fragment of a PCR product generated using the primers 5¢-CGCGGATCCGAGGACGAGACGG-3¢ (forward) and 5¢-GCTCTAGATTAGCTATAGGCGTCC ATGG-3¢ (reverse) and 5¢-CGCGGATCCGAGGACGAG ACGG-3¢ (forward) and 5¢-GCTCTAGATTATTGCCGG AATGCCTCCT-3¢ (reverse), respectively HNF1bhomeo was amplified by PCR using the primers 5¢-CGGAA TTCAAAGAAGATGCGCCGCAAC-3¢ (forward) and 5¢-GCTCTAGATTAGCTATAGGCGTCCATGG-3¢ (re-verse) All amplified HNF1 fragments were verified by sequencing, digested with EcoRI and XbaI, then inserted into the GFP-Rc/CMV and pCS2+MT [45] expression vectors

GFP-Rc/CMV was constructed by inserting the

5¢-GGCAAGCTTCTGGCCACCATGAGTAAAGGA-3¢ (forward) and 5¢-CGGAATTCGTTTTGTATAGTTCAT CCATGC-3¢ (reverse) primers to amplify a region of the pCSGFP2 vector [46] into the Rc/CMV expression vector (Invitrogen) The expression clone encoding Xenopus HNF1b was kindly provided by R Vignali, University of Pisa, Italy

4 [47], and the plasmids encoding Xenopus lim1 and Pax8 were kindly supplied by P D Vize, University of Calgary, Canada [44]

Cell culture, transfection and luciferase assay HeLa cells (our lab stock)

Dulbecco’s modified Eagle’s medium supplemented with penicillin (100 UÆmL)1), streptomycin (100 UÆmL)1) and 10% (v/v) heat-inactivated fetal bovine serum The cells were seeded at a density of 3· 105cells per 3.3 cm dish The transfection was performed 24 h after seeding using 1.3 lg

of reporter gene, 0.3 lg of expression vector, and 6 lL of lipofectamine (Invitrogen) The final DNA concentration was equalized by the addition of Rc/CMV vector The transactivation activity was measured after 20 h using the luciferase reporter assay system (Promega) and a Lumat LB

9501 luminometer (Berthold, Wilbad, Germany)

Embryos, microinjection of synthetic mRNA and

immunohistochemistry

In vitrofertilization and culture of Xenopus laevis

were performed as described previously [48] Adult Xenopus

laevis were obtained from Xenopus I, Inc (Dexter, MI,

USA) and the animal experimentation guidelines were

followed (Regierungspra¨sidium Du¨sseldorf, Germany) The

developmental stages are taken from the
Normal Table of

Xenopus laevis [49]

HNF1 chimeric proteins and the GFP encoding expression

vector (pCSGFP2) were linearized with NotI and PvuII,

respectively, then in vitro transcribed with SP6 RNA

polymerase [35] A total of 250 pg of capped mRNA

encoding a chimeric protein together with 100 pg of capped

green fluorescent protein

one blastomere of two-cell stage embryos After 2 days, the

injected side was scored under a stereofluorescence

micro-scope for the presence of GFP At the swimming larval stage

(45), the animals were fixed in MEMFA [0.1M MOPS,

pH 7.4, 2 mM EGTA, 1 mM MgSO4, 3.7% (v/v)

formal-dehyde], subsequently dehydrated in methanol and stored at

)20 C For whole-mount immunostaining, the embryos

were rehydrated in NaCl/Piand blocked with NaCl/Piand

0.1% (v/v) Triton X-100 (PBT)/20% (v/v) goat serum for

1 h at room temperature Incubation with hybridoma

supernatant of the monoclonal antibodies, 3G8 and 4A6

(kindly provided by E A Jones, University of Warwick, UK

9[50]), was performed overnight at 4C after a 1 : 2 dilution

in PBT/20% (v/v) goat serum After washing five times with

PBT at 20–25C

10;11 , incubation with a 1 : 1000 diluted cyanine

Cy3

10;11 -conjugated rat anti-(mouse) Ig (Jackson

Immuno-Research, West Grove

Embryos were washed five times with PBT at room

temperature, then analyzed by fluorescence microscopy

Statistical analysis

The difference between the injected and the noninjected

sides was evaluated by measuring the whole area using the

lateral view with the widest diameter from the dorsal to the

ventral side of the immunostained pronephros The area

included the pronephric tubules and the anterior part of the

pronephric duct The measurements were made using the

computer program KAPPA IMAGE BASE METEO

(opto-elec-tronics GmbH, Gleichen, Germany), and the noninjected

side was used as a reference for each animal No size

difference was set as 100 The values representing kidney

size obtained from each mutant were compared to values

obtained from GFP control-injected embryos Significant

differences were scored using the Mann–Whitney test to

calculate P-values

Results

The conserved 26 aa segment of HNF1b affects

the transactivation potential

We first explored whether the 26 aa segment specifically

deleted in the splice variant HNF1b-B (Fig 1) could interfere

with nephrogenesis The splice variant B (HNF1bD) was

constructed by deleting the 26 aa segment as shown in

Fig 2A A second construct was created from a truncated HNF1b protein (HNF1bbb) that corresponds to the human Y352insA HNF1b mutation, that we have shown

in previous experiments to be sufficient to induce agenesis of the pronephros in Xenopus [43] By deletion of the 26 aa segment from HNF1bbb we constructed a truncated protein lacking the 26 aa segment (HNF1bbbD, Fig 2A) As a third type we generated a HNF1a variant containing the 26 aa segment from HNF1b that is normally not present in HNF1a As the full-length HNF1a protein has no effect on renal development [35], we assumed that the truncated HNF1a protein (HNF1aaa) lacking the transactivation domain would not have an effect either By adding the 26 aa segment to this truncated version of HNF1a we produced the HNF1aaains26 construct (Fig 2A)

The subcellular localization of these constructs was first assayed in transfected HeLa cells Previous experiments have shown that HNF1a is localized primarily in the nucleus but also to a certain extent in the cytoplasm [51] Localization of HNF1b, however, is exclusively nuclear [43] To define the subcellular distribution of these various proteins, we expressed GFP fusion proteins of these constructs in HeLa cells All HNF1b-derived constructs (HNF1b, HNF1bD, HNF1bbb and HNF1bbbD) were localized exclusively in the nucleus (Fig 2B) In contrast, the HNF1a-derived constructs (HNF1aaa and HNF1aaains26) were present in both the nucleus and the cytoplasm (Fig 2B), as observed previously for full-length HNF1a [51]

Additionally, the transactivation potential of these HNF1 derivatives were investigated Expression vectors encoding these proteins were cotransfected into HeLa cells lacking endogenous HNF1 proteins together with a luciferase reporter plasmid containing an HNF1 inducible promoter Deletion of the 26 aa sequence present in HNF1b reduced the transactivation potential 30% compared to the

full-length HNF1b transcription factor (Fig 2C) As observed previously [43], the truncated HNF1b protein lacking the transactivation domain retained substantial transactivation potential (compare HNF1bbb with HNF1b in Fig 2C) Typically, HNF1bbb was less active at 10–30 ng expression vector, but as active as the full-length protein when 150–

300 ng expression vector were transfected The truncated HNF1b construct missing the sequence encoding the 26 aa segment (HNF1bbbD) transactivated similarly to HNF1bbb (Fig 2C) In contrast, the truncated HNF1a protein (HNF1aaa) had only a residual activity even when 300 ng expression vector were transfected (Fig 2C) This is consis-tent with the initial description of the HNF1a transcription factor and the definition of the C-terminal activation domain

of HNF1a [52,53] The insertion of the b-specific 26 aa segment into the truncated HNF1a construct (HNF1aaa) abolished residual activity This indicates that the 26 aa segment plays some role in the transactivation potential

The conserved 26 aa segment of HNF1b interferes with pronephros development inXenopus laevis

The morphogenetic potential of the various HNF1 con-structs were examined in the developing Xenopus embryo by injecting mRNA encoding these proteins into one blasto-mere of the two-cell stage embryo As initial experiments revealed that the GFP-HNF1 fusion proteins fluoresced too

weakly for the identification of the injected side (data not

shown), GFP mRNA was coinjected with RNA for the

myc-tagged version of the constructs (Fig 3A) as

per-formed previously [35] Injected embryos were raised to free

swimming tadpoles (stage 45) and processed to visualize the

pronephros using a mixture of monoclonal antibodies for

the pronephric tubules and duct [50] Only embryos that

were otherwise phenotypically normal were scored for

effects on pronephric development Examples of dorsal

views of such larvae are given in Fig 3C The pronephric

size was measured in the lateral view (Fig 3B) of a whole

series of larvae, and the quantification of these phenotypic

changes together with the statistical analysis for significance

are summarized in Fig 3D

As found previously [35,43], full length HNF1b led to a

significant reduction of the size of the pronephros (Fig 3D),

and this effect was even more pronounced for the truncated

HNF1b protein (HNF1bbb, Fig 3D) As expected, the

truncated HNF1a protein (HNF1aaa) did not interfere with

pronephros development (Fig 3D) The HNF1b protein

lacking the 26 aa segment (HNF1bD) had no effect on pronephric size, implying a crucial role of this 26 aa segment

in nephrogenesis (Fig 3D) However, the truncated HNF1b protein lacking this 26 aa segment (HNF1bbbD) led to a reduction of the pronephric size (Fig 3D), indicating additional nephrogenic segments in this truncated protein The insertion of the 26 aa segment into the HNF1a protein (HNF1aaains26) led to a reduction of pronephric size (Fig 3D), illustrating that the nephrogenic potential of the

26 aa segment is transferable A dramatic lethality at the injected side was observed when the truncated HNF1b construct lacking the 26 aa segment (HNF1bbbD) was overexpressed (Fig 3E,F) More than 90% of the injected embryos died during gastrulation Even when the amount of HNF1bbbD mRNA was halved, 70% of the embryos still died during gastrulation The majority of the surviving tadpoles were distorted (Fig 3H–J) compared to control animals (Fig 3G)

of healthy larvae were available for immunostaining and the examination of the pronephros-specific effects

Neverthe-Fig 2 Subcellular localization and transactivation potential of HNF1 constructs with deletion or insertion of the 26 aa segment (A) The domains encoded by each HNF1 construct are shown diagrammatically together with their designation The black box indicates the 26 aa segment deleted in HNF1b splice variant B (B) Immunofluorescence of HeLa cells expressing GFP fusion proteins of the various constructs shown in A Bar, 10 lm (C) Increasing amounts of GFP-HNF1b expression constructs (shown in A) were cotransfected with a HNF1-dependent luciferase reporter gene into HeLa cells The fold-activation induced by each of the HNF1 expression constructs is shown Error bars indicate standard deviation of the mean of at least six replicates.

less, this group was sufficient for significant analysis This abnormal development was not observed with any of the other constructs

To control the efficiency of protein production, we tested the amount of HNF1 proteins made by Western blots As exemplified in Fig 3K, very similar levels were found in the injected embryos The truncated HNF1a protein (HNF1aaa) was as abundant as the truncated HNF1b protein (HNF1bbb) demonstrating that both proteins are equally expressed Thus, the presence of HNF1aaa has in contrast to HNF1bbb no effect on pronephric development

Function of the dimerization domain of HNF1b

As overexpression of the truncated HNF1b derivative lacking the 26 aa segment (HNF1bbbD) also reduced the pronephric size (Fig 3D), we postulated that other seg-ments present in this molecule may interfere with nephro-genesis To explore the function of the dimerization domain

of the HNF1b protein, we constructed chimeras of the HNF1a and HNF1b proteins as shown in Fig 4A The molecular and cellular properties of these chimeric con-structs were assayed in transfected HeLa cells as well as in developing Xenopus embryos

The construct encoding the HNF1b-derived POUSand POUHdomains fused to the HNF1a dimerization domain (HNF1abb) was localized exclusively in the nucleus of transfected HeLa cells In contrast, the construct encoding the HNF1a-derived POUSand POUHdomains fused to the HNF1b dimerization domain (HNF1baa) was localized both in the nucleus and the cytoplasm (Fig 4B) These data indicate that the POUSand POUH, but not the

HNF1b-Fig 3 Pronephric phenotype in Xenopus larvae after expression of HNF1 proteins lacking or containing the HNF1b-specific 26 aa segment (A) Control neurula expressing GFP on the injected side (B) Lateral view of a larvae (stage 45) expressing full-length HNF1b protein (C) Dorsal view of larvae (stage 45) expressing the HNF1 protein desig-nated Whole-mount immunostaining for the pronephric tubules and duct using a Cy3-conjugated secondary antibody is shown as red fluorescence The injected side is marked by an arrow Bar, 300 lm (D) Statistical analysis of pronephric size in injected vs noninjected sides after expression of various HNF1 proteins Boxes include 75% of the values, and the vertical line represents the group median, and whiskers represent the outer quartile The P-value calculated using the Mann–Whitney test and the animal number scored per group shown at the far right The reference indicates the GFP-injected control animals (E, F) Example of embryo exhibiting defects at gastrulation illumin-ated at normal light (E) or under green fluorescence (F) Note cell death was only observed in the injected (GFP positive) region Bar,

300 lm (G) Stage 44 control embryo injected with GFP alone Bar,

1 mm (H–J) Developmental defects of different degrees were observed

in tadpoles expressing the truncated HNF1b protein lacking the 26 aa segment (HNF1bbbD) Animals shown in panel H and I could not be scored for pronephric morphology (K) Western blot of protein extracts derived from neurulae stage embryos injected with RNA encoding HNF1aaa, HNF1bbb, HNF1aab or GFP

myc-tag specific antibody GE10 [35] Each sample was an aliquot representing one embryo of a pool of 60 injected embryos At later stages, the amount of HNF1 proteins was too low to be quantified.

derived dimerization domain, determine the exclusively

nuclear localization

Transfection of the chimeric HNF1 constructs together

with an HNF1 dependent luciferase reporter plasmid was

used to measure the transactivation activity in HeLa cells

Only the construct encoding the POUSand POUHof the

HNF1b protein (HNF1abb) resulted in transactivation of

the reporter gene similar to that mediated by the truncated

HNF1b construct (HNF1bbb, Fig 4C) The presence of

the HNF1b-derived dimerization domain in the chimeric

protein (HNF1baa) failed to increase the transactivation of

the reporter compared to the truncated HNF1a protein (HNF1aaa, Fig 4C)

The influence of the chimeric constructs on kidney development was tested in overexpression experiments in Xenopus embryos Injection of mRNA encoding chimeric proteins with either the HNF1b-specific dimerization domain (HNF1baa) or the b-specific DNA binding domains (HNF1abb) of the HNF1b protein led to a reduction in pronephric size (Fig 4D) This indicates that the dimerization domain as well as the DNA binding domain of HNF1b interfere with pronephric development

HNF1aaa

HNF1baa

HNF1abb

HNF1bbb

1

1

1 1

351

321 80

70

A

0 50 100 150 200 relative pronephros size

B

0

2

4

6

8

10

12

ng expression vector

HNF1aaa

HNF1bbb HNF1abb

HNF1baa

C

localizationtransactiv

ation

-N N/C N/C N

-11.1 4.4 4.4 11.2

D

phenotype

reference (95)

4.6 e-24 (226)

0.129 (147)

0.001 (114)

5.7 e-8 (151)

GFP

Fig 4 Function of the dimerization domain of HNF1b (A) The domains included in the HNF1 constructs are shown diagrammatically HNF1b is shown in purple and HNF1a in blue The black box indicates the 26 aa segment deleted from the HNF1b splice variant B (B) Molecular and cellular properties of HNF1 constructs were assayed in transfected cells as well as in developing embryos On the left, N and N/C refer to nuclear and nuclear plus cytoplasmic localization, respectively In the middle, the fold induction of the HNF1-dependent luciferase reporter after trans-fection of the HNF1 constructs into HeLa cells is shown On the right, statistical analysis of pronephric size in injected vs noninjected sides after expression of various HNF1 proteins Boxes include 75% of the values, and the vertical line represents the group median, and whiskers represent the outer quartile The P-value calculated using the Mann–Whitney test and the animal number scored per group shown at the far right The reference indicates the GFP-injected control animals (C) Increasing amounts of HNF1 expression constructs were cotransfected together with a HNF1-dependent luciferase reporter into HeLa cells Mean of fold-activation of the reporter is represented by points, and error bars represent standard deviation of at least six replicates (D) Whole-mount immunostaining for pronephric tubules and duct in Xenopus larvae overexpressing the HNF1 protein indicated on one side The injected side is marked by an arrow Bar, 300 lm.

However, the quantification shows a clear distinction in the

extent of the effect (Fig 4B), as the construct containing

only the dimerization domain of HNF1b (HNF1baa) was

considerably less efficient than the construct containing the

POUSand POUHdomains of HNF1b (HNF1abb)

The homeodomain of HNF1b is essential for nuclear

localization and interferes with pronephric development

To explore the function of the HNF1b homeodomain

(POU ) in more detail, chimeric constructs were created

containing various parts of the HNF1b homeodomain region The chimeric gene constructs generated are shown diagrammatically in Fig 5A Functional performance as measured by subcellular localization, transactivation activ-ity and effect on kidney development is summarized in Fig 5B All chimeric constructs containing the HNF1b homeodomain were found exclusively in the nuclear com-partment, implying that this domain contributes to nuclear localization (Fig 5B) With regard to the transactivation potential, we observed that all chimeric constructs (Fig 5C) were less active than the truncated HNF1b protein

HNF1aabins26

HNF1aab

HNF1aabH

HNF1aabHS

0 50 100 150 200

relative pronephros size

A

GFP

localizationtansactivition

-N

N

N

N

-3.2

5.5

4.3

2.0

-HNF1aab

HNF1aabins26

1

1

351

351 176

183

196 229

HNF1aabH

1

319 196

229

HNF1aabHS

1

311 196

229

D

B

0 50 100 150 200 250 300

HNF1aab HNF1aabH

HNF1aabins26 HNF1aabHS

ng expression vector

C

phenotype

reference (95) 8.5e-9 (148)

2.7 e-14 (45)

2.3e-14 (42)

0.182 (85)

0.641 (81)

Fig 5 The homeodomain of HNF1b is essential for nuclear localization and interferes with pronephric development (A) The domains included in the HNF1 constructs are shown diagrammatically HNF1b is shown in purple and HNF1a in blue The black box indicates the 26 aa segment of the HNF1b splice variant B (B) Molecular and cellular properties of HNF1 constructs were assayed in transfected cells as well as in developing embryos.

19 See Fig 4 legend for details (C) Increasing amounts of HNF1 expression constructs were cotransfected together with a HNF1-dependent luciferase reporter into HeLa cells Mean of fold-activation of the reporter is represented by points, and error bars represent standard deviation of at least six replicates (D) Whole-mount immunostaining for pronephric tubules and duct in Xenopus larvae overexpressing the HNF1 protein indicated on one side The injected side is marked by an arrow Bar, 300 lm.

(HNF1bbb) in transfection assays (Fig 4C) While the

constructs containing the HNF1b homeodomain but

lack-ing the 26 aa segment (HNF1aab) gave approximately a

fivefold transactivation of the reporter plasmid, the

corres-ponding construct containing the 26 aa segment

(HNF1aab-ins26) gave a threefold activation Successive truncation at

the C-terminal end of the homeodomain in the constructs

lacking the 26 aa segment (HNF1aabH, HNF1aabHS) led

to a further decrease in the transactivation, but was still

twofold above base level (Fig 5C)

To identify whether the homeodomain influences

kidney development in Xenopus embryos, mRNA from

the chimeric constructs were injected into one cell at the

two-cell stage, and the pronephric size was measured

(Fig 5B,D) Three chimeric constructs with the

HNF1b-specific homeodomain (HNF1aab, HNF1aabins26 and

HNF1aabH) led to a reduction of the pronephric size with

the constructs lacking the 26 aa segment (HNF1aab and

HNF1aabH) being most effective (Fig 5B) In contrast, the

construct lacking eight amino acid at the C-terminal part of

the homedomain (HNF1aabHS) had no effect on

proneph-ric size indicating the critical C-terminal border A construct

producing the homeodomain alone (HNF1bHomeo) had

no effect on pronephric size (Fig 5B), indicating that the

HNF1 backbone is required to allow the protein function

that interferes with kidney development We observed that

both chimeric constructs lacking the 26 aa segment

(HNF1aab and HNF1aabH) had an adverse effect on

normal development as found for the truncated HNF1b

lacking the 26 aa segment (HNF1bbbD, Fig 3E–J) In fact,

most surviving animals were distorted allowing only a

minority to be analyzed at stage 45 This adverse effect

on embryogenesis was absent in the construct with the

C-terminal truncation (HNF1aabHS) that has also lost its

effect on nephrogenesis

Partial rescue of Pax8/lim1-mediated pronephros

malformation by HNF1b injection

It has been reported that overexpression of the transcription

factors, Pax8 and lim1, in Xenopus embryos led to the

development of an abnormally large pronephros, and to the

formation of ectopic pronephric tissue [44] As both these

transcription factors are expressed at the neurula stage

together with HNF1b in the pronephric anlage, we

won-dered whether simultaneous overexpression of HNF1b

could overcome the effects of Pax8 and lim1

Overexpres-sion of Pax8 or lim1 by themselves led only to marginal

effects, but synergized to have a pronounced effect [44] We

coinjected RNA encoding Pax8 and lim1 into one

blasto-mere of the two-cell stage embryo together with GFP

mRNA as a tracer Injected embryos were raised to the

swimming tadpole stage, and processed to visualize

pro-nephric tubules and duct Overexpression of Pax8 together

with lim1 led to an enlargement of the pronephros as

compared to embryos injected with GFP alone (Fig 6A)

This size difference was shown to be significant using the

Mann–Whitney test (Fig 6G) More importantly, ectopic

pronephric tubules and small cysts close to the main

pronephric body were observed using immunostaining on

the injected side (Fig 6B) Such structures were seen in 16%

of the injected embryos (Table 1), but never observed in

injections with mRNA encoding GFP or any HNF1 derivative Furthermore, 24% of the larvae coinjected with Pax8 and lim1 displayed cyst-like structures or a thickening

of the tubules on the injected side (Table 1) Such abnor-malities were seen in only 4% of larvae injected with the truncated HNF1b protein (HNF1bbb) Our results are similar to those using a different injection protocol reported previously [44] We coinjected mRNAs encoding Pax8 and lim1 together with HNF1b and GFP as a tracer into one cell

of the two-cell stage Xenopus embryos Immunostaining for the pronephric tubules and duct at the tadpole stage showed that these embryos had pronephric structures similar to embryos injected with Pax8 and lim1 alone (Fig 6C,D) The pronephros appeared smaller in some larvae, but the size difference was not significant when compared with larvae injected with Pax8 and lim1 alone (Fig 6G) Furthermore, 17% of the samples were found to have ectopic tubules (Table 1) Cyst-like structures or thickening

of the tubules were also found in 27% of the samples These data imply that the overexpression of Pax8 and lim1 is dominant to the effect of HNF1b It was not possible to inject higher concentrations of HNF1b mRNA, but as the truncated HNF1b protein (HNF1bbb) was more active in reducing the pronephric size (Fig 3C), this construct was coinjected together with Pax8 and lim1 These larvae had slightly smaller pronephroi in the injected side (Fig 6F), suggesting that HNF1bbb coinjection could overcome the effect mediated by Pax8 and lim1 More importantly, no larvae had ectopic tubules (Table 1) However, 28% of the samples were found to have cyst-like structures or thicken-ing of the tubules (Fig 6E,F), similar to the fraction showing this phenotype in Pax8 and lim1 coinjected embryos (Table 1) Therefore, cyst-like structures and thicker tubules mediated by Pax8 and lim1 were not rescued

by HNF1bbb Taken together these results indicate that the Pax8- and lim1-induced phenotype has two separate qualities One is the enlargement and ectopic formation of pronephros which could be antagonized by HNF1b and the other is the induction of cyst-like structures which could not

be antagonized by HNF1b

Discussion The transcription factors, HNF1a and HNF1b, display extensive structural similarities with indistinguishable DNA sequence binding specificity [2] Our data imply that they have acquired distinct functions during evolution as homologous domains of these two factors display disparate properties These include the subcellular localization, the transactivation potential as well as the ability to affect nephrogenesis

HNF1b has a nuclear localization sequence located

in the homeodomain Analyzing the subcellular localization of various chimeric HNF1 proteins, we observed an exclusively nuclear staining

in transfected HeLa cells in all constructs containing the POU homeodomain (POUH) of the HNF1b protein This finding is consistent with our previous data showing nuclear localization of all truncated HNF1b transcription factors retaining the POU domain [43] The occurrence of a

nuclear localization signal (NLS) in the homeodomain of

the HNF1b protein is supported by the presence in the

N-terminal region of the homeodomain (amino acid

229–235, Fig 7) of the amino acid sequence, KKMRRNR,

predicted to be a NLS (PredictNLS Online, http://

cubic.bioc.columbia.edu) The NLS of the HNF1b protein

and the HNF1a protein (KKGRRNR) differ by only one

amino acid (Mfi G) This change may hinder efficient

nuclear translocation of HNF1a in transfected HeLa cells,

and probably results in the nuclear as well as cytoplasmic localization typical for HNF1a

Differential transactivation potential of the HNF1a and HNF1b protein

The C-terminal transactivation domains of HNF1a and HNF1b are only weakly conserved (Fig 1), and in most transactivation assays HNF1a is approximately twofold

Fig 6 Partial rescue of Pax8/lim1-induced kidney malformation by HNF1b (A–F) Lateral views of two representative larvae expressing the proteins listed at the left on one side Larvae are immunostained to visualize the pronephric tubules and duct (A, B) Enlarged pronephroi in Pax8/ lim1 (125 pg mRNA each per embryo) coinjected embryos (C, D) Enlarged pronephroi in embryos coinjected with Pax8 (125 pg mRNA per embryo), lim1 (125 pg mRNA per embryo), and HNF1b (250 pg mRNA per embryo) (E, F) Reduced pronephric size in embryos coinjected with Pax8 (125 pg mRNA per embryo), lim1 (125 pg mRNA per embryo) and truncated HNF1b (HNF1bbb, 250 pg mRNA per embryo) Anterior is

to the left for the injected sides, and to the right for the noninjected sides, and dorsal is up Thickened tubules (T) characterized by a wider diameter and cyst-like structures or bubbles (B) are indicated by arrows Ectopic pronephric tubules are indicated by arrow heads Bar, 200 lm (G) Statistical analysis of pronephric size in injected vs noninjected sides after expression of various HNF1 proteins Boxes include 75% of the values, and the vertical line represents the group median, and whiskers represent the outer quartile The P-value calculated using the Mann–Whitney test and the animal number scored per group shown at the far right The reference indicates the GFP-injected control animals.