Human Molecular Genetics Advance Access originally published online on March 3, 2004
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Human Molecular Genetics, 2004, Vol. 13, No. 8 851-867
DOI: 10.1093/hmg/ddh102
Human Molecular Genetics, Vol. 13, No. 8 © Oxford University Press 2004; all rights reserved
Huntingtin Interacting Protein 1 mutations lead to abnormal hematopoiesis, spinal defects and cataracts
1Department of Internal Medicine, 2Howard Hughes Medical Institute, 3Orthopaedic Research Laboratory, 4Kellogg Eye Center and 5Department of Pathology, University of Michigan Medical School, Ann Arbor, MI 48109, USA
Received December 31, 2003; Accepted February 16, 2004
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Huntingtin Interacting Protein 1 (HIP1) binds clathrin and AP2, is overexpressed in multiple human tumors, and transforms fibroblasts. The function of HIP1 is unknown although it is thought to play a fundamental role in clathrin trafficking. Gene-targeted Hip1–/– mice develop premature testicular degeneration and severe spinal deformities. Yet, although HIP1 is expressed in many tissues including the spleen and bone marrow and was part of a leukemogenic translocation, its role in hematopoiesis has not been examined. In this study we report that three different mutations of murine Hip1 lead to hematopoietic abnormalities reflected by diminished early progenitor frequencies and resistance to 5-FU-induced bone marrow toxicity. Two of the Hip1 mutant lines also display the previously described spinal defects. These observations indicate that, in addition to being required for the survival/proliferation of cancer cells and germline progenitors, HIP1 is also required for the survival/proliferation of diverse types of somatic cells, including hematopoietic progenitors.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
HIP1 was originally isolated by yeast two-hybrid screening as a huntingtin-associated protein (1,2). Since the original isolation of HIP1, it has also been found to be part of a chromosomal translocation with the PDGFßR in leukemia (3). HIP1 and its close relative HIP1-related (HIP1r) (4) share ANTH/ENTH domains, leucine zipper domains and TALIN homology domains and interact with clathrin as well as other clathrin coat components (5–10). The yeast ortholog of Hip1, sla2, is essential for yeast cellular growth, is involved in assembly of the cytoskeleton and is required for endocytosis (11). The similarities in amino acid sequence and predicted domains between HIP1, HIP1r and SLA2P suggest an analogous role for HIP1 and HIP1r in the regulation of cytoskeletal and endocytic processes.
Although by homology and its interaction with clathrin, HIP1 is thought to play a role in endocytosis, its actual biochemical and physiologic role(s) are unknown. For example, expression of HIP1 has been reported to be pro-apoptotic (12,13). In contrast, we have found that HIP1 is overexpressed in multiple primary human tumors (14) and that HIP1 protein is necessary for the survival of mouse germline progenitors (9) and many cell lines (14). Overexpression of full-length HIP1 transforms fibroblasts, and cell lines with HIP1 overexpression have increased levels of multiple growth factor receptors (15,16). This promotion of cell survival, proliferation and receptor signaling maybe due to decreased degradation of growth factor receptors as a result of altered trafficking. However, the physiological requirement for HIP1 in normal somatic cells has received only limited study.
Given that HIP1 is widely expressed, we were interested in determining whether it was required for proliferation and survival in somatic cells beyond the spine (17) and testis (9). To begin to test this we evaluated hematopoiesis in distinct Hip1 mutant mouse lines and consistently found hematopoietic as well as spermatogenic and spine defects. All of the phenotypes must be a result of either increased cell death, altered differentiation or diminished cell proliferation, suggesting that HIP1 is necessary for maintenance of cellularity in multiple tissue types including the bone marrow.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Targeted inactivation of Hip1
Since stable overexpression of HIP1 alters the levels of multiple different receptor tyrosine kinases, we wanted to test if the targeted mutation of Hip1 in the mouse would result in phenotypic effects in multiple different tissues. Because the Hip1 gene has a complex structure (220 kb and contains at least 30 exons), gene targeting has a chance of generating hypomorphic alleles and the phenotypic evaluation of the effects of multiple knockout alleles is valuable. We have therefore analyzed the phenotypes of mice that resulted from the homozygous presence of two new Hip1 mutant alleles and describe hematopoietic and ophthalmic abnormalities that have not been reported previously.
First, a conditional knockout allele and its germline Cre recombined allele were generated. The latter allele has a deletion of exons 3–5 of Hip1 (see Supplementary Material Fig. S1 for construction of targeting vector and Fig. 1 for resultant allele structures). Importantly, the neomycin cassette was deleted along with exons 3–5. From our first targeting event we had several correctly targeted ES cells confirmed by 5', 3' and neo probes (right-hand panels of Fig. 2A and data not shown). The conditional allele was designated Hip1loxp. The recombined allele was designated Hip1
3-5. The latter mutation was used to generate homozygous mice to compare their phenotypes to phenotypes observed in the other Hip1‘null’ mice that we generated (see below).
|
|
The Hip1‘null’ allele that was evaluated in greatest detail for this report was generated serendipitously as follows. While attempting to knock the HIP1/PDGFßR fusion protein cDNA (3) into the Hip1 locus (see Supplementary Material Fig. S2 for construction of targeting vector and Fig. 3 for resultant allele structures), we found that some of our correctly targeted ES cells (Fig. 2A) did not express the HIP1/PDGFßR protein (Fig. 2B). Expression of the HIPl/PDGFßR fusion protein was tested with three different antibodies that detect the fusion protein: anti-HIP1, anti-PDGFßR and the anti-phosphotyrosine antibody, 4G10. The last antibody recognizes the constitutively active HIP1/PDGFßR (18). This analysis was facilitated by our observation that the Hip1 promoter is strongly active in ES cells (see the detection of the endogenous murine HIP1 protein expressed from the non-targeted allele in panel I of Fig. 2B). One of the HIP1/PDGFßR non-expressing ES cells was injected into blastocysts and chimeric mice were generated. Since the targeted allele from the HIP1/PDGFßR targeted cell line failed to express HIP1/PDGFßR at the protein level, it offered the prospect of an independent Hip1‘null’ allele.
|
To determine why a HIP1/PDGRßR fusion protein was not expressed from the targeted allele in some of the targeted ES cell lines (Fig. 2B), the sequence of the genomic locus for the mutant Hip1 allele was analyzed. To do this a genomic library from the blastocyst injected ES cell DNA (represented in lanes 1 of Fig. 2A and B) was created and two clones were isolated that contained the human HIP1/PDGRßR sequences. The human–mouse junction in both clones was sequenced and compared with mouse genomic sequence surrounding exon 2. A 498 bp deletion of mouse genomic sequences was found in the region targeted by the knockin vector (Supplementary Material Fig. S3). This placed the human HIP1/PDGFßR cDNA in murine intron 2 instead of being fused in frame with murine exon 2 (Fig. 3). The rest of the Hip1 allele was intact. This deletion predicts a frameshift mutation in the Hip1 protein product, as murine exon 1 is fused to the human HIP1/PDGFßR cDNA out of frame. Henceforth we refer to the previously described Hip1 allele (9) that replaced exons 2–8 with the neomycin cassette as ‘–’ and the targeted Hip1 allele described in detail here as ‘null’.
Chimeric mice generated from the ES cells with the serendipitous Hip1‘null’ allele as well as the Hip1loxp allele were mated with C57BL/6 females and F1 agouti pups were genotyped by Southern blot analysis. The F1 heterozygotes for the different alleles were then intercrossed to generate F2 animals. The numbers of Hip1null/null mice from this particular intercross were significantly decreased compared with the expected Mendelian ratios (55% hets, 27% wts and 17% homozytgotes; n=395; P<0.001). Genetic and pathologic evaluation of embryos from days 11–18.5 post coitum showed normal Mendelian ratios and normal appearing embryos, indicating that the partial lethality may be perinatal (data not shown). The extent of perinatal lethality increased as Hip1 targeted mice were back-crossed onto a C57/BL6 background. There were no differences noted in growth rates or young adult weights among surviving mice of the F2 generation.
The brains of Hip1null/null mice did not express detectable Hip1 mRNA by northern blot analysis using a pair of probes that spanned the entire cDNA (Fig. 2C). Given that Hip1null/null mice generated from the HIP1/PDGFßR-targeted allele exhibit more severe or penetrant phenotypes than the originally described Hip1–/– mice (9), as described herein, we suggest that the original mice may have expressed an undetected hypomorphic allele. Consistent with the lack of a detectable Hip1 mRNA, there was also no detectable HIP1 protein in the Hip1null/null brains (Fig. 2D; left hand panel) or in multiple other tissues from both male and female Hip1null/null mice (Fig. 4). It is also noteworthy in light of our observed hematopoietic phenotype (see below) that HIP1 protein was present in the whole bone marrow of wild-type mice and not Hip1null/null mice (Fig. 5A inset).
|
|
The F1 heterozygotes for the Hip1loxp allele, as mentioned above, have also been intercrossed to generate F2 animals. So far, the expected number of Hip1loxp/loxp homozygotes has been born and brains from the homozygous mice showed only a slight reduction in the levels of HIP1 protein (right hand panel of Fig. 2D). When these mice were then mated with the transgenic CMVcre deletor mice obtained from Jackson Laboratories [TgN(hCMV-cre)140Sau; stock number 002471], we were fortunate to obtain some Fl heterozygous mice with germline recombination to generate the Hip1
3–5 allele (see Fig. 1 for structure; this allele was identified by Southern blot analysis for loss of the neomycin cassette). F2 mice with homozygous recombined alleles (Hip13–5/3–5) have been generated and have no detectable murine HIP1 protein (right hand panel of Fig. 2D). Our expansion of the Hip13–5 colony has allowed us to accumulate phenotypic data for comparison to the observed phenotypes in the Hip1null/null mice (see below and Table 1). Interestingly, in contrast to the Hip1null/null mice, to date the Hip13–5/3–5 mice have been born in Mendelian ratios (26% wt, 49% hets and 24%; n=119; P=0.8) with no evidence of the predicted partial lethality of the Hip1null/null mice. Potential reasons for this difference include genetic background effects, the possibility that the Hip13–5 allele is hypomorphic or that the Hip1null allele, but not the Hip13–5 allele, affects expression of neighboring genes that are partially necessary for perinatal survival of mice.
|
Hip1 mutant mice exhibit lower numbers of primitive but not more mature hematopoietic progenitors in vitro
In addition to finding HIP1 expressed at the protein level in whole bone marrow, we have found that Hip1 mRNA is expressed at similar levels in both bone marrow and hematopoietic stem cells (HSCs; Supplementary Material Fig. S4). These expression data together with the fact that the human Hip1 allele was identified as part of a oncogenic translocation in leukemia (3) and Hip1 is located at chromosome 7q11.2, a region frequently deleted in hematopoietic malignancies (19), suggested that HIP1 may have a role in hematopoiesis. Because of this, we originally evaluated peripheral blood counts and in vitro hematopoiesis in the original group of Hip1–/– mice and found significant hematopoietic abnormalities in vitro (Fig. 6). We therefore also evaluated hematopoiesis in the new Hip1null/null mice. Neither the Hip1–/– nor the Hip1null/null mice had differences in peripheral blood cell counts relative to control littermates (data not shown). To evaluate whether Hip1 deficiency affected hematopoietic progenitors, bone marrow cells obtained from Hip1–/– mice, HIP1null/null mice as well as heterozygous and wild-type littermates corresponding to each line were cultured in methylcellulose. Since different types of hematopoietic progenitors form distinct colonies in methylcellulose, we were able to assay whether there was a difference between these mice in the frequencies of various types of progenitors (20,21). There was no difference between either type of HIP1-deficient mice and control heterozygous or wild-type littermates in the overall frequency of clonogenic hematopoietic progenitors or in the frequencies of BFU-E (erythroid progenitors) or CFU-GM (myeloid progenitors that give rise to both granulocytes and/or macrophages; Figs 6A and 7A–C). There were also no differences in the frequencies of a number of phenotypically defined lymphoid progenitor populations (including thymocyte subpopulations, pro-B, and pre-B cells) that were compared by flow-cytometry in Hip1 deficient and control littermates (data not shown).
|
|
In contrast, there was a significant difference in the frequency of the most primitive progenitor population that can be detected in the methylcellulose assay, the CFU-GEMM (an earlier progenitor with the potential to make granulocytes, erythrocytes, macrophages and megakaryocytes). In each of four independent experiments, 50% of Hip1–/– bone marrow samples yielded normal numbers of CFU-GEMM, while the remaining 50% exhibited severely reduced numbers of CFU-GEMM relative to littermate controls (P<0.0001; Fig. 6B). These data demonstrate that half of Hip1–/– mice exhibit a defect in the frequency of primitive hematopoietic progenitors, without exhibiting any defect in the frequencies of more mature hematopoietic progenitors or in peripheral blood counts. In comparison, although Hip1null/null bone marrow produced CFU-GEMM colonies of normal size, the frequency of such colonies was severely reduced in 85% of mice analyzed (P<0.0011; Fig. 7D). It is possible that the diminished penetrance of this phenotype in the Hip1–/– mice compared with the Hip1null/null mice was secondary to differing effects of background (both were mixed 129Svj/C57BL6). The other possibility is that the Hip1–/– mice were hypomorphic at the Hip1 locus and the Hip1null/null mice were not. These data demonstrate that HIP1-deficient mice exhibit a defect in the frequency or differentiation of primitive hematopoietic progenitors in vitro, without exhibiting any defect in the frequency of more mature hematopoietic progenitors or in peripheral blood cell counts.
By adding the growth factors thrombopoietin and Flt-3 ligand to the methylcellulose medium, we were able to partially correct the reduced frequency of CFU-GEMM in bone marrow from the same Hip1null/null mice (P=0.l7; Fig. 7E compared with D). One possibility is that the loss of Hip1 expression led to an increased requirement for growth factors to promote the survival of CFU-GEMM. This would be consistent with the observation that stable overexpression of HIP1 increases the levels of growth factor receptors and confers the ability to grow in reduced serum concentrations (16). An alternate possibility is that thrombopoietin and Flt-3 allowed a distinct HIP1-independent progenitor to form CFU-GEMM colonies.
To test whether HIP1 deficiency affects the differentiation of HSCs in culture, HSCs were isolated by flow-cytometry from adult bone marrow as Thy-1.1 1oSca-1+Lineage–c-kit+ cells (22) and sorted into methylcellulose cultures at a density of one cell per well. The frequency of Thy-1.1 1oSca-1+Lineage–c-kit+ cells did not differ between Hip1null/null and wild-type mice. In each of two independent experiments comparing three null mice with three wild-type mice (all littermates), Hip1null/null HSCs exhibited a similar clonogenic capacity to wild-type HSCs, with 74–80% of HSCs forming colonies in methylcellulose. However, Hip1null/null HSCs formed a significantly higher proportion of CFU-GM myeloid colonies (P<0.05) and a significantly lower proportion of more primitive CFU-GEMM colonies (P<0.05; Fig. 7F and G). The CFU-Meg difference was not statistically significant. This suggests that Hip1null/null HSCs were less able to undergo multilineage differentiation in culture upon stimulation by the cytokines present in the methylcellulose medium. Together, the data suggest that growth and differentiation of HSCs, and certain other hematopoietic progenitors are regnlated by HIP1.
HIP1-deficient mice are resistant to myeloablation
As described above, Hip1null/null mice exhibited normal blood cell counts and normal numbers of HSCs. Although Hip1null/null bone marrow cells formed fewer CFU-GEMM colonies in culture, these data did not tell us whether this reflected a reduction in the number of these progenitors in vivo or simply their inability to survive and form normal colonies in culture. Because of this and because HIP1 regulates growth factor receptor signaling, we decided to further test whether HIP1 deficiency alters hematopoiesis in vivo by stressing the hematopoietic system. Treatment with the cytotoxic compound 5-fluorouracil (5-FU) kills many hematopoietic cells, particularly dividing progenitors. Quiescent HSCs are recruited into cycle after 5-FU treatment to replace the blood cells that are killed (22,23). Since proliferating HSCs are more sensitive to 5-FU toxicity, sequential treatment with 5-FU eventually leads to hematopoietic failure due to depletion of the HSC pool and the inability to regenerate lost blood cells.
To stress the mice, we treated Hip1null/null mice and wild-type littermates with sequential weekly doses of 5-FU. Hip1null/null mice were significantly more resistant to 5-FU-induced hematopoietic failure than wildtype mice (Fig. 5A; P<0.003; log rank test). As expected, 5-FU treatment strongly reduced white blood cell (WBC) counts in wild-type mice, and the extent of the reduction increased with successive 5-FU treatments until all of the mice died (Fig. 5B). In the Hip1null/null mice, WBC counts dropped after the first two 5-FU treatments, but then rebounded to normal levels on day 28, 6 days after the third 5-FU treatment. Only three of 10 Hip1null/null mice died. Thus the WBC counts and survival data indicate that Hip1null/null mice were resistant to myeloablation by serial administration of 5-FU. It should be noted that all mice in each experiment were litter-matched males to control for effects of sex and background. This demonstrates that HIP1 deficiency does alter hematopoiesis in vivo in response to stress.
To add evidence in support of the hypothesis that the hematopoietic phenotype is a direct result of HIP1 deficiency, rather than affects that are secondary to the complexity of the ‘null’ allele or presence of the neomycin cassette, we have also challenged Hip13–5/3–5 mice with 5-FU. In these HIP1-deficient mice we found a similar resistance to 5-FU-induced bone marrow failure and death (Supplementary Material Fig. S5).
The mechanism by which HIP1 deficiency makes hematopoietic cells more resistant to serial treatment with 5-FU is not clear. One possibility is that reduced sensitivity of HIP1-deficient HSCs to certain growth factors precluded some of these cells from being recruited into cycle after 5-FU treatment. This would make the HIP1-deficient mice more resistant to serial 5-FU treatment because the reduced proliferation of HIP1-deficient HSCs after 5-FU treatment would make subsequent rounds of 5-FU treatment less toxic to the HSC pool. Owing to the increased lethality of HIP1-deficiency on a C57/BL6 background, bone marrow reconstitution and competitive repopulation experiments have not yet been possible.
Hip1-deficient male mice are infertile
The original Hip1–/– mice showed testicular degeneration (9). The new Hip1null/null mice also showed testicular degeneration with apoptosis at the postmeiotic spermatid stage and most of the males were infertile. The scoring system for testicular degeneration that was described in Table 3 of our previous work on testicular degeneration in the HIP1-deficient background was used for our evaluation of testicular degeneration (9). Consistent with the possibility that the original Hip1 mutant allele was hypomorphic, the degree of testicular degeneration in the Hip1null/null mice was qualitatively increased compared with the Hip1–/– mice.
To quantitate the reduced fertility, five Hip1null/null male mice and male five wild-type male mice were mated with two proven-breeder females each (one male per cage) over a 3-month period. Only 0.8 litters/female were generated in the Hip1null/null cages whereas the wild-type cages generated 2.4 litters/female (P<0.001; Pearson's chi-square test). Of the eight total litters produced from Hip1null/null males, five were derived from just one male. Hence, most Hip1null/null males have profoundly reduced fertility but a minority of Hip1null/null males may have normal fertility. This suggests that there is a low level of incomplete penetrance of this phenotype. In addition, we have not yet observed any fertile Hip13–5/3–5 male mice despite multiple attempts at mating.
Other abnormalities in the HIP1-deficient mice
As early as 4 months of age many of the Hip1null/null and Hip13–5/3–5 mice developed a hunched posture resulting from a severe curvature of the spine (kypholordosis) that could readily be observed on X-rays (Fig. 8). This phenotype has been observed in another Hip1 knockout mouse that was recently published (17) but was not observed in the Hip1–/– mice (9). The mechanism for this phenotype has been hypothesized to be due to a central nervous system defect associated with diminished AMPA receptor trafficking in the Hip1 mutant background. However, it remains unclear exactly how this CNS defect might translate into the hunchback phenotype. In addition, despite a vigilant search, there have been no morphologic abnormalities observed in the CNS associated with Hip1 deficiency in the mice described here or in the previous reports of Hip1 knockout phenotypes (9,17).
|
We have found that this phenotype is more penetrant in females. By 4 months of age, 38% (n=12/33) of the observed Hip1null/null females had the phenotype compared with only 2.7% (n=1/37) of the Hip1null/null males. There were no wild-type mice with this phenotype (n=113). Approximately half of the females (n=17) had been pregnant and were breeding at the time of developing this back deformity, showing an increased trend for development of this phenotype in pregnant (47%, n=8/17) compared with non-pregnant female mice (28%, n=5/18). This phenotype was also associated with severe weight loss as it progressed. By one year of age, 100% of Hip1null/null mice and 0% Hip1+/+ mice in a separate observation group that consisted of nine litter-matched mutant and control paired mice (total n=18) had the kypholordotic spine phenotype.
Histologic examination of the spines or long bones (femurs and tibia) from the affected mice did not show severe osteoporosis, osteoarthritis or neurologic pathology that would account for this phenotype (data not shown). In addition, we have stained the skeletons of Hip1null/null mice and their wild-type littermates at various ages (ranging from 2 weeks to 6 months) with alizarin red (bone) and Alcian blue (cartilage) and, interestingly, found that after digesting away the soft tissue and soaking the skeletons in glycerol as part of the staining procedure, the curvature of the spine was partially relieved. In addition, the bone to cartilage ratio in these mice was not affected by HIP1-deficiency, indicating that there was no developmental or degenerative skeletal or cartilage defect. The radiographic density of the thoracic spine, femurs and skull appeared lower in the nulls compared to the wild types (for examples see arrows in Fig. 8A). Yet microcomputed tomography of femurs, vertebrae and ribs in nine littermate pairs of Hip1null/null (with an obvious huntchback phenotype) and wild-type mice showed no significant difference in bone density between genotypes (24). This indicates that, although the bones of Hip1null/null mice may be slightly osteopenic, they were not osteoporotic. It remains to be determined if this mild decrease in bone density is directly due to the intrinsic loss of HIP1 expression or is due to a secondary effect of the hunchback phenotype. HIP1 is highly expressed in peripheral nerves so the kypholordosis could be secondary to peripheral nerve defects, although it remains unknown whether there are primary defects in peripheral nerve function in HIP1-deficient mice.
A variety of other mutant mice display the kypholordotic phenotype (25–29). However, these studies have shed little light on the mechanism of the spinal abnormalities. There has been speculation that the phenotype represents acceleration of normal process associated with aging such as degenerative osteoarthritis of the intervertebral facet joints, osteoporosis, connective tissue defects, musculoskeletal and/or neurologic defects. In this regard, a recent report showed that activation of the tumor suppressor gene p53 also led to a hunchback phenotype (29). In light of HIP1 acting as an oncoprotein (14,16), it is interesting that the hunchback phenotype in the Hip1null/null mice is consistent with a similar phenotype observed in a mouse with activation of a tumor suppressor gene, such as p53.
Finally, 71% of the Hip1null/null mice had externally visible micro-ophthalmia (Fig. 9A). Interestingly, our Hip1–/– and Hip13–5/3–5 mice have not developed this phenotype. Upon closer examination, all (100%) of the Hip1null/null mice had small eyes when evaluated based on darkfield illumination of the dissected lens (Fig. 9B), histology (Fig. 9C) or eye weight (Fig. 9E). In addition 100% of the Hip1null/null mice had micro-lens with nuclear cataracts and associated cortical abnormalities (Fig. 9B and C). Although the cornea, iris and retina of Hip1null/null mice were also significantly smaller than in wild-type littermates, their histological organization was normal. The eye abnormalities were observed even in mice at 3 weeks of age, suggesting that this phenotype is developmental rather than degenerative (Fig. 9E and data not shown). As expected from this phenotype, the HIP1 protein was expressed in eyes from wild-type mice but not in Hip1null/null mice (Fig. 4). The reduced size of the mutant eyes is consistent with a potential role for HIP1 in the survival or proliferation of cells in multiple regions of the developing eye.
|
A close up view of hematoxylin stained nuclei from the Hip1null/null versus the wild-type lens is shown in Figure 9D. In addition to there being an abnormal presence of nuclei in the cells that form the central portion of the lens, we also found that the nuclei were condensed (see arrows) and there was vacuolization of the cytoplasm. In comparison, the nuclei from the wild-type lens were only found in the epithelium and were characterized by intact, non-pyknotic nuclei. To work towards an understanding of the mechanism of the lens degeneration, TUNEL assays were performed to assess if the nuclei found in the abnormal, pyknotic lens cells (see arrows of Fig. 9D for examples) were apoptotic. These cells were indeed TUNEL-positive (Fig. 9F), suggesting that HIP1 is not only necessary for the appropriate differentiation of the lens epithelial cells into the well-organized, light-transmitting lens fibers, but also for their survival. However, since the homozygous presence of the other alleles of Hip1 does not lead to cataracts, presumably because those alleles are to some degree hypomorphic, it remains possible that this phenotype is related to the alteration of an undefined neighboring gene(s).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Previous studies have demonstrated that a variety of cancers overexpress HIP1 (14) and that HIP1 is capable of transforming fibroblasts (16). The observed pro-growth properties of overexpressed HIP1 probably arise from increased levels of receptor tyrosine kinases (15). One would therefore expect that regulation of HIP1 levels and function would be critical in the maintenance of cellular homeostasis. In fact, analysis of what may have been a Hip1 hypomorphic mouse demonstrated that full-length HIP1 expression is required for the survival of spermatogenic progenitors (9).
To follow this up with additional in vivo data we have generated two new Hip1 mutant alleles that are distinct from the original allele, each with their own advantages and disadvantages. Metzler et al. (17) has constructed a distinct Hip1 mutation, as well. In the latter report they show that HIP1 deficiency leads to a spinal phenotype that we also describe in this report. They describe it as a neurologic phenotype as the mice develop an associated gait abnormality with muscle wasting. On the other hand, we think that the gait abnormality and progressive wasting may be secondary to the thoracolumbar defects compressing on peripheral nerves and note that HIP1-deficient mice, like the mice described here and previously (9,17), do not have any abnormalities in the brain or spinal cord morphology. Since our mice may have a trend towards a worsening of their hunchback phenotype with the physical stress of pregnancy, this also suggests that the vertebral column defects may be the primary abnormality and the gait problems secondary. However, the actual mechanism of how the phenotype develops remains to be elucidated.
What is very interesting and relevant to our report is that Metzler et al. (17) show evidence that Hip1 deficiency may alter the levels of glutamate receptors in neurons cultured from Hip1 knockout mice. They show evidence that there is a relatively diminished level of intracellular glutamate receptors in the HIP1-deficient background (17). This is consistent with Hip1 having a role in trafficking of multiple receptors and adds AMPA receptors to the list of receptors that may be regulated by HIP1 (in addition to the EGFR, FGFR and PDGFßRs) (15).
In contrast to our current observations, the report by Metzler et al. (17) as well as our previous knockout report (9) did not describe the hematopoietic or ophthalmic defects in the HIP1-deficient background. These phenotypes may not have been observed because they are subtler, or the previously characterized alleles may be hypomorphic. There could also be differences in background effects (all of our observations have been on a mixed B6/129 background) or neighboring gene effects that could result from the various mutations.
Despite the differences between the various knockout mice, the collective data indicate that HIP1 expression is necessary for the normal proliferation, survival or differentiation of cells from a number of somatic tissues, including the hematopoietic system. Analyses of the response of Hip1null/null bone marrow to growth factors in culture suggest that these HIP1-deficient cells required higher levels of cytokines that are present or additional cytokines than wild-type cells for appropriate survival and differentiation. Together, with data that implicate HIP1 in the regulation of clathrin-mediated receptor trafficking, these results suggest that a variety of somatic cells, as well as germline cells (9), and cancer cells (16), depend on HIP1 for the efficient transduction of survival and proliferation signals from ligand stimulated receptors.
Within the hematopoietic system, HIP1-deficient primitive but not more mature hematopoietic progenitors were less efficient at forming multilineage (CFU-GEMM) colonies in methylcellulose. However, Hip1null/null mice appeared to have normal numbers of HSCs based on the frequency of Thy-1.1loSca-1+Lineage–c-kit+ cells. Hip1null/null Thy-1.1loSca-1+Lineage–c-kit+ cells also formed normal numbers of colonies in methylcellulose, but these colonies included an increased frequency of CFU-GM colonies and a decreased frequency of CFU-GEMM colonies. Adding additional cytokines to the methylcellulose medium increased the frequency of CFU-GEMM colonies formed by Hip1null/null bone marrow cells. These data suggest that hematopoietic progenitors require HIP1 in order to exhibit normal sensitivities to cytokines and to efficiently undergo multilineage differentiation in culture.
In order to test whether hematopoiesis was altered in vivo by HIP1 deficiency, the mice were serially treated with 5-FU. Although Hip1null/null and Hip13-5/3-5 mice exhibited normal blood cell counts under normal conditions, they differed from wild-type mice in their response to weekly 5-FU treatment. The Hip1 mutant mice exhibited higher WBC and platelet counts at day 21 compared with wild-type mice and most of the Hip1 mutant mice survived the sequential 5-FU treatments. Since serial, weekly 5-FU treatments leads to hematopoietic failure by killing the HSCs that are recruited into cycle as a result of 5-FU treatment, it is likely that Hip1 mutant mice better tolerate 5-FU treatment because their HSCs are not recruited into cycle in the same way as wild-type HSCs. This would also be consistent with Hip1null/null whole bone marrow exhibiting reduced cytokine sensitivity, as cytokine stimulation after myeloablation is presumably what triggers HSC activation. Taken together, the data indicate that HIP1 is required in a stage-specific manner by primitive hematopoietic progenitors in order to respond normally to cytokines in vitro and in vivo.
The selective effect of HIP1 deficiency on multipotent progenitors is reminiscent of the functions of the cyclin-dependent kinase inhibitors p27 (30) and p21 (31), which are also required by hematopoietic progenitors in a stage-specific manner. HIP1, p21 and p27 are unusual in that most proteins that have been found to affect hematopoiesis have widespread effects on both primitive and restricted progenitors. It will be interesting to determine whether HIP1 selectively regulates the signaling of receptors that regulate p21 function in hematopoietic progenitors.
A requirement for HIP1 in cell survival, differentiation or proliferation could also explain the other phenotypes observed in this study. The Hip1null/null mice that exhibited the most severe spinal deformity had associated generalized weight loss and absence of subcutaneous fat (Fig. 8). Moreover, the Hip1null/null mice were consistently micro-ophthalmic and cataracts were observed in 100% of the Hip1null/null mice compared with a frequency of 0.5% in wild-type littermates. Interestingly, the cataract phenotype was not observed in the other Hip1 mutant mouse lines, suggesting that there is either a more complete knockout with this allele, or that there are neighboring genes that are affected by this relatively complex allele. Since Hip1 is such a large gene with at least 30 exons and its first intron is more than 100 kb in size, it is expected that there could be difficulties in generating completely null mutant alleles. In addition, mutations leading to neighboring gene effects resulting in a phenotype related to two or more gene alterations are always a possibility, especially in a gene-rich area such as the area of chromosome 5 that is syntenic to human 7q11.
It is unlikely that the complex HIP1 phenotype is attributable to neighboring gene effects caused by the presence of a residual neomycin cassette in some of the Hip1 mutant alleles. First, the presence of the neomycin cassette in the Hip1loxp/loxp mice does not cause any of the observed phenotypes in the absence of HIP1 deletion. Second, the Hip13–5/3–5 mice exhibited the testis, back and hematopoietic phenotypes, despite lacking a residual neomycin cassette. The incomplete penetrance originally observed in the Hip1–/– mice could be due to the effects of a hypomorphic allele or a background effect.
Two of the phenotypes related to homozygous Hip1 mutations, hematopoietic and spinal defects, together with the chromosomal location of the human HIP1 gene to 7q11, suggests that there may be autosomal recessive mutations of HIP1 in human patients with genetic syndromes that include hematopoietic and skeletal abnormalities. The chromosomal locus of 7q11 and its syntenic 5q region in mice is a gene-rich region and is somatically deleted in acute and chronic leukemias (19). In addition, germline mutations in patients with Shwachman–Diamond syndrome have been mapped to human chromosome 7q11 (32), where HIP1 is located. This syndrome is autosomal recessive, and has hematopoietic and skeletal defects. Germline mutations in an open reading frame designated SBDS (that does not correspond to HIP1) on 7q11 have been described in some of these individuals (33). It will be important to determine if individuals that do not exhibit mutations in SBDS have germline mutations in HIP1.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Antibodies
The 3' polyclonal HIP1 antibody, pcHIP1 and the monoclonal antibodies to HIP1, HIP1/4B10 and HIP1/1B11 have previously been described (14,34). The anti-mHIP1–5' end antibody was created by injecting a glutathione sepharose-purified GST-mHIP1–5' antigen into rabbits using standard procedures. The GST-mHIP1–5' was constructed by subcloning the 5' end of murine HIP1 (included all sequences up to the single EcoRI site in murine HIP1) into pGEX 4T3.
Generation of Hip1 mutant mice
The conditional Hip1 knockout vector (pHIP1CKO) was constructed from two mouse genomic subclones designated EcoB/E and Bam187. These are shown in Supplementary Material Figure S1. Using these two clones, the targeting vector, pHip1CKO, was completed as follows. The most 5' subclone (subclone EcoB/E) was digested with KpnI and BamHI (5' homologous arm) and subcloned into (p98TB/KpnI-BamHI to create ‘step 1’ plasmid (5.2 kb). Subclone Bam187 was then digested with EcoRV and HindIII (3' homologous arm) and subcloned into EcoRV/HindIII-digested ‘step 1’ plasmid to obtain ‘step 2’ plasmid (8.7 kb). Subclone Bam187 was then digested with BamHI, filled in, digested with EcoRV (conditional knockout region) and sub-cloned into HindIII digested and blunted 38LoxpNeo and screened for correct orientation by sequencing. This was called the ‘step 3’ plasmid (9.7 kb). The latter plasmid was digested with XhoI and SfiI, filled in with Klenow. The released insert was sub-cloned into ‘step 2’ plasmid that had been digested with AscI and filled in with Klenow. The correct orientation was confirmed by sequencing to obtain the pHip1CKO (15.5 kb).
To construct the HIP1/PDGFßR targeting vector (Supplementary Material Fig. S2), the original Hip1 knockout vector (pHIP1KO) (9) was sequenced with a reverse primer in the neomycin resistance cassette to choose a sense oligo 5' of the XhoI site in intron 1 (oligo 1). Antisense oligo 2 was designed from the human HIP1 cDNA starting at nucleotide 156. Sense oligo 3 was also derived from the human HIP1 cDNA and started at nucleotide 157 (5'-GCT GTA AAG GAA AAA CAC GCC-3') and reverse oligo 4 was derived from the human HIP1 cDNA starting at position 588. Subclone EcoB/E (Supplementary Material Fig. S2) was then used as a template for PCR using oligos 1 and 2. The 2.5 kb PCR fragment was then digested with XhoI. The pcDNA3H/P template (18) was then used as a template for generation of a 0.4 kb PCR fragment using oligos 3 and 4. A vector designated pGL-1/polyA/loxPneo was then digested with SalI (compatible with XhoI) and Ecl1361 (blunt). A three-fragment ligation was then completed with this sticky-blunt vector and the two PCR fragments described above. The resultant vector was designated pGL-1/step5 with an insert of 5 kb. The 3' end of the H/P cDNA as an insert was then created by digesting pcDNA3H/P with NotI, filling in the overhangs with Klenow and then digesting with HpaI to obtain the 3.6 kb fragment. The pGL1/step5 vector was next digested with HpaI and ligated with the 3.6 kb HpaI/NotI (blunt) H/P 3' end. This was designated pGL-1/step6 with an insert size of 8.6 kb. This vector was then digested with NotI and AscI and filled in with Klenow. This generated an 8.6 kb fragment that contained the H/P cDNA fused at its 5' end with Hip1 murine genomic sequence that included the 5' portion of exon 2 and fused at its 3' end to a polyadenylation signal and neomycin resistance cassette (Hip1_H/P_polyA_loxPneo). The p38LoxPNeo/HIP1 (9) was digested with XhoI and SfiI to excise the already existing neomycin resistance cassette, filled in with Klenow and then ligated with the 8.6 kb (Hipl_H/P_polyA_loxPneo) insert. The final targeting vector (18.9 kb) was designated H/P-KI/LoxPNeo. The H/P-KI/LoxPNeo vector was electroporated into 129SvJ RW1 ES cells, selected for G418 resistance and screened by Southern blotting for correctly targeted clones (Fig. 3A). Generations of chimeric mice and germ line transmission of the mutant alleles were achieved using standard techniques.
Preparation of genomic DNA
Mouse tails were resuspended in Lysis buffer containing 10 mM Tris (pH 7.5–7.9), 5 mM EDTA and 0.4 M NaCl, followed by addition of 10% SDS to the final concentration of 0.2% and proteinase K to 500 µg/ml. The cell lysis samples were shaken at 55°C overnight. The samples were then subjected to organic extraction, twice with phenol-chloroform and twice with chloroform. Cellular DNA was precipitated with 2.5 vols of 100% ethanol and washed with 70% ethanol once. DNA pellets, after drying, were resuspended in TE buffer and DNA concentrations were measured by spectrophotometer.
Southern bolt analysis
For analysis of the Hip1 ‘null’ or ‘loxp’ and ‘3–5’ alleles, 10 µg of genomic DNA was digested with EcoRI or BamHI overnight and run on a 0.8% agarose Tris borate EDTA gel at 80 V for 16 h. The gel, after acidic solution treatment and alkaline neutralization, was blotted onto Hybond-N filter. The DNA bound filter was then blocked in prehybridization buffer containing 5xSSC, 5x Denhardt's solution, 1% SDS and 100 µg/ml denatured salmon sperm DNA for 3 h at 65°C. The hybridization was carried out overnight at 65°C with the neomycin cassette probe, the Hip1CKO 3' probe (created by PCR amplification of sequences 0.3 kb 5' of the BamHI site that delineates the 3' end of the Bam187 subclone; Supplementary Material Fig. S1). The forward primer sequence (RSS3CONDF) was 5'-GAG GGA GCA GGC TCC TCC-3' and the reverse primer sequence (RSS3CONDR) was 5'-TGG ATT CAC CAT GTC GCC-3'), the Hip1CKO 5' probe (generated by digestion and release of a 0.5 kb fragment from the zeocin resistant subclone EcoB/E with EcoRI and XbaI), Hip1 knockin 3' probe (generated by digestion of the zeocin resistant BamH/B subclone in pZero-1 with HindIII and XhoI to release a 1.3 kb fragment) or the Hip1 knockin 5', probe (generated by digestion of the EcoB/E subclone with EcoRI and XhoI to release a 0.5 kb fragment; see Figs 1 and 3 for positions of the probes).
To prepare the probes for labeling the probe fragments were excised from a 1% agarose–TBE gel and DNA was extracted from the gel using QIAEX II agarose gel as directed by the supplier (Qiagen). The purified 5' and 3' probes were then labeled with 32P using a random-primed labeling kit (Roche). After hybridization, the blot was washed stringently, twice at 65°C with 2xSSC for 20 min and twice at 65°C with 1xSSC for 10 min and twice at 65°C with 0.1xSSC for 10 min. The blot was exposed to Kodak Biomax film overnight.
Northern blot analysis
Total RNA was isolated from mouse brain using TRIZOL (Invitrogen). RNA was electrophoresed on 1% agarose gel (Invitrogen) with 6% formaldehyde (Sigma). The gel was stained with ethidium bromide for 30 min, and then destained for 3 h. The presence of sharp bands corresponding to the 18 and 28 s ribosomal RNAs were confirmed by ultraviolet illumination. RNA was transferred overnight to Nytran (Schleicher and Schuell) by capillary action, and the blot was UV cross-linked and prehybridized in buffer containing 5xSSC, 5xDenhardt's solution, 1% SDS (w/v), 100 µg/ml denatured salmon sperm DNA for 3 h at 65°C. 32P labeling of the mHip1 (24396) probe was made by random primed labeling (Roche) with 32P-dCTP (NEN). The blot was then hybridized overnight at 65°C, washed twice in 2xSSC for 20 min, twice in 1xSSC for 10 min, twice in 0.1xSSC for 10 min, and imaged on Kodak Biomax film. The film was exposed for 4 days.
Isolation of bone marrow cells for functional characterization
Bones were dissected and overlying blood vessels, muscle and fascia were thoroughly excised. Cells were flushed from each marrow cavity with Hank's Buffered Salt Solution without calcium or magnesium, supplemented with 2% heat-inactivated calf serum (HBSS+) using a 3 ml syringe and 27G needle. Cells were triturated into single cell suspension and filtered through nylon screen prior to antibody staining.
Flow-cytometric isolation of hematopoietic stem cells
HSCs were isolated as previously described (22). Briefly, whole bone marrow cells from mice that had been mated with C57BL/Ka-Thy1.1 mice and selected to have the Thy1.1 surface marker were incubated with a panel of unconjugated monoclonal antibodies to lineage specific surface markers including B220 (6B2), CD3 (KT31.1), CD5 (53–7.3), CD8 (53–6.7), Gr-1 (8C5) and Ter119. Following dilution, pelleted cells were resuspended in anti-rat IgG specific F(ab)2 fragment conjugated to phycoerythrin (PE). Cells were subsequently stained with directly conjugated antibodies to Sca-1 (Ly6A/E; allophycocyanin (APC), c-kit [2B8; biotinylated (bio)], Thy-1.1 [19EX5; fluorescein-5-isothiocyanate (FITC)] Mac-1 (M1/70; PE) and CD4 (GK1.5; PE). Streptavidin conjugated to PharRed (APC-Cy7) was used to visualize c-kit. HSCs were often pre-enriched by using paramagnetic microbeads (Miltenyi Biotec) to select Sca-1+ or c-kit+ cells. Prior to flow-cytometry, cells were resuspended in 2 µg/ml 7-AAD to discriminate live from dead cells. Only live (7-AAD negative) cells were included in the sorts and analyses. HSCs were isolated as Thy-1.1 1oSca+ lineage–Mac- 1–CD4–c-kit+. All flow-cytometry was performed on a FACS Vantage (Beckton Dickinson).
Lineage analysis of whole bone marrow cells by flow cytometry
Briefly, directly conjugated antibodies to B220 (6B2), Mac-1 (M1/70), Gr-1 (8C5), CD3 (KT3 1.1), IgM, CD4 (GK1.5), Thy-1.1 (19EX5) and Sca-1 (Ly6A/E) were added at appropriate dilution to whole bone marrow cells suspended at 
lx108 cells/ml. Cells were resuspended in 2 µg/ml 7-AAD in HBSS+ prior to flow-cytometry.
Methylcellulose culture
Approximately 750–1000 live whole bone marrow cells or single resorted HSCs were plated per well of 96-well plates. Each well contained 100 µl 1.0% methylcellulose medium with 20% charcoal absorbed fetal bovine serum, 1% BSA, 50 or 150 ng/ml stem cell factor (SCF), 10 ng/ml interleukin-6 (IL-6), 3 U or 9 U/ml erythropoietin (Epo) and 10 ng/ml interleukin-3 (IL-3), plus or minus 10 ng/ml Flt-3 and 10 ng/ml thrombopoietin (Tpo). Cultures were maintained at 37°C at fully humidified conditions in 5% CO2. BFU-E and CFU-GM were scored on days 8–10 and verified on day 12. CFU-GEMM was scored on day 12 and verified on day 14.
Treatment of mice with 5-FU
Littermate mice of varying genotypes in the B6/129 mixed background (12–18 weeks of age) were injected intraperitoneally with 5-FU (Adrucil, Pharmacia and Upjohn Co., Kalamazoo, MI, USA; 150 mg/kg) once a week. The survival of the mice was recorded daily and peripheral blood counts were obtained every 14 days. Mice were bled using a needle stick in the saphenous vein of the hind leg. Approximately 50–100 µ1 of blood were collected from each animal. Blood counts were measured on a HEMAVET Multispecies Hematology Analyzer (CDC Technologies, Oxford, CT, USA).
Preparation and histologic analysis of the eye
Animals were euthanized with isoflurane (Aerrane, Baxter Pharmaceutical, Deerfield, II, USA) and the eyes removed intact with the optical nerve. The eyes were then fixed in Bouin's Fixative (LabChem Inc., Pittsburgh, PA, USA) for 24 h and stored in 70% ethanol. The fixed eyes were weighed (Fig. 5E), embedded in paraffin, sectioned and stained with hematoxylin and eosin (Fig. 5C and D). The TUNEL assay was performed after deparaffinization of the eye sections using the in situ cell death detection kit (Roche), using an alkaline phosphatase-conjugated anti-fluorescein dUTP antibody.
Methods for quantitative real time-PCR (qPCR)
Approximately 2000–10 000 cells were directly sorted into 400 µl Trizol (Ambion, Austin, TX, USA) containing 250 µ/ml glycogen (Roche, Indianapolis, IN, USA). RNA was extracted according to the manufacturer's instructions. The extracted RNA (30 µl volume) was treated for 20 min at 37°C with 2 µl RNase-free DNase-1 (2 U/µl; Ambion) in the presence of 2 µl RNase-inhibitor (10 U/µl; Invitrogen). The RNA was then purified using an RNeasy Mini Kit (Qiagen, Valencia, CA, USA) according to the manufacturer's instructions and washed three times with RNase-free water. The RNA was used for making cDNA by reverse transcription with random hexamer. The cDNA was extracted with phenol-chloroform and precipitated with 20 µg glycogen. After dissolving the cDNA with RNase-free water, cDNA equivalent to 200 cells was used for each PCR reaction. Primers were designed to generate short amplicons (100–150 bp). The PCR reactions were performed using a LightCycler (Roche Diagnostic Corporation) according to the manufacturer's instructions. The RNA content of samples compared by qPCR was normalized based on the amplification of hypoxanthine phosphoribosyl transferase (HPRT). In addition to confirming the specificity of the qPCR reactions by examining the melting curves of the products, qPCR products were separated in 2% agarose gels to confirm the presence of a single band of the expected size. To estimate the difference in the expression levels of individual RNAs between samples, we assumed that one cycle difference in the timing of amplification by qPCR was equivalent to a 1.9-fold difference in expression level (90% amplification efficiency), a typical value (35).
The primers used for each qPCR assay were as follows. HPRT forward primer sequence was 5'-CCTCATGGACTGATTATGGACA-3' and reverse primer sequence was 5'-ATGTAATTCCAGCAGGTCAGCAA-3'; HIP1 forward primer sequence was 5'-CGGACTCAAGAGCAACAGGATG-3' and reverse primer sequence was 5'-AGCCATTTCGCTTCTGACTGGG; HIP1r forward primer sequence was 5'-GCTTACCGTGGAGATGTTTGACTAC-3' and reverse primer sequence was TCCTGAATGACCTGGATGAGCG-3'; huntingtin forward primer sequence was 5'-ATGGGCACACATCTCTGGAAAC-3' and reverse primer sequence was 5'-TTCAGCAGGGATACGGTTGACC-3'.
SUPPLEMENTARY MATERIAL
Supplementary Material is available at HMG Online.
|
ACKNOWLEDGEMENTS |
|---|
We would like to thank members of the Ross Laboratory for critical review of the manuscript. This work was supported by a postdoctoral fellowship grant from the Huntington's Disease Society of America (D.R.), the grants R0l CA82363-01A1 and RO1 CA098730-02, ASH and the Damon Runyon Foundation (T.S.R.).
|
FOOTNOTES |
|---|
* To whom correspondence should be addressed. Tel: +1 7346155509; Email: tsross{at}umich.edu
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
- Kalchman, M.A., Koide, H.B., McCutcheon, K., Graham, R.K., Nichol, K., Nishiyama, K., Kazemi-Esfarjani, P., Lynn, F.C., Wellington, C., Metzler, M. et al. (1997) HIP1, a human homologue of S. cerevisiae Sla2p, interacts with membrane-associated huntingtin in the brain. Nat. Genet., 16, 44–53.[CrossRef][ISI][Medline]
- Wanker, E.E., Rovira, C., Scherzinger, E., Hasenbank, R., Walter, S