STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT The work leading to the invention was supported at least in part by National Science Foundation Grants OPP- 9420712 and OPP-9815381. Therefore, the U. S. government may have certain rights in the invention.

BACKGROUND OF THE INVENTION Hematopoiesis is the developmental process of blood cell formation and encompasses the production, turnover, and replenishment of all blood cell lineages. Specialized cells, termed"hematopoietic stem cells" (HSCs), give rise to multiple cell lineages that perform a wide array of functions from infectious defense (lymphocytes and myeloid cells) to transport of molecular oxygen (erythrocytes).

The dominant blood cell type is normally the erythrocyte, which constitutes up to 45% of the total volume of blood.

Although intensely studied, red cell development from individual progenitors, termed"erythropoiesis", remains poorly understood.

Red blood cells differentiate from HSCs under the control of lineage-specific factors and of cell-specific gene expression. During erythropoiesis, cytokines such as stem cell ligand (SCL), interleukin-3 (IL-3), interleukin- 6 (IL-6), and granulocyte-macrophage colony-stimulating factor (GM-CSF), direct HSCs toward the erythroid/myeloid

differentiation program, beginning with development of the BFU-E (blast forming unit-erythroid) (1). The BFU-Es are activated by the lineage-restricted factor erythropoietin (Epo) to differentiate into colony forming units of erythroid lineage (CFU-E). Erythropoietin, in synergy with IL-3, is required for commitment and differentiation of CFU-Es and for maintenance of functionally mature erythrocytes 2). Binding of the Epo ligand to its cognate receptor, EpoR, on the CFU-E initiates a signal transduction cascade in which Jak 2 kinase and STAT5 (signal transducer and activator of transcription 5) activate downstream targets such as erythroid-specific transcription factors (e. g., GATA-1, erythroid Kruppel- like factor (EKLF), etc.) (3; 4). Once these downstream targets are activated, proerythroblasts, also termed "colony forming units-erythroid" (CFU-E), continue to differentiate into mature red blood cells. Structural genes encoding Band 3, Band 4.1, ankyrin and spectrin, are activated to provide the cytoskeletal framework to shape the erythrocyte for movement through the capillary network. Downstream transcription factors (e. g., EKLF) activate production of the oxygen transporter, hemoglobin, and red cell enzymes such as carbonic anhydrase.

The study of blood cell development has been greatly facilitated by the use of model vertebrate genetic systems, in particular the mouse and the zebrafish (Danio rerio) (5). The murine system permits functional analysis of hematopoietic genes by means of"gene knock-out"technology (6). However, this method is biased because it can only provide insight into the functions of genes that have already been discovered. Systematic, large-scale mutagenesis of the zebrafish, on the other hand, has provided a random approach to the discovery of novel hematopoietic genes (7), but the production and screening

of mutants and the isolation of mutated genes by positional cloning is a laborious and expensive process. Therefore, there remains a need for a more comprehensive, yet simpler and cost-effective strategy for screening for novel gene products involved in blood cell formation and other processes, including erythropoesis, lymphopoiesis, myelopoiesis, and blood clotting, which need is served by the present invention.

SUMMARY OF THE INVENTION The inventors have discovered that a unique family of Antarctic fish, the hemoglobin-lacking or"white-blooded" icefish (15 species in the family, Channicthyidae, suborder Notothenioidei) fail to express hemoglobin, due to deletion of most of the juvenile and adult globin gene complexes.

Absent hemoglobin, the icefish also no longer produce red blood cells, most likely due to mutations in genes required for erythropoiesis. Interestingly, most other families of the Notothenioid suborder have retained the ability to produce hemoglobin and mature erythrocytes.

Thus, one aspect of the present invention encompasses a method for uncovering unknown hematopoiesis-related genes, particularly erythropoiesis-related genes, by screening for and identifying genes that are expressed by red-blooded animals but not by white-blooded animals. A preferred embodiment screens for genes expressed by red-blooded Notothenioid fish but not white-blooded Notothenioid icefish.

The erythropoiesis-related gene screening method employs representational difference analysis (RDA) (12) to clone presumptive erythroid genes that are expressed by red-blooded fish, preferably red-blooded Notothenioids of the Antarctic, such as the rock-cods (used here as the"tester"genome), but not by white-blooded icefish (used here as the"driver" genome). When applied to cDNA libraries, this subtractive

hybridization methodology enriches for genes expressed solely by the tester genome. That is, the driver cDNA eliminates genes expressed by both genomes to produce a cDNA population enriched for the unique cDNAs of the tester genome. To enrich for red blood cell or erythropoiesis-related cDNA probes,"representations"of total cDNA from the red-cell- forming organ, the head kidney, of the red-blooded Antarctic yellow-belly rock-cod, Notothenia coriiceps (the tester), were hybridized to total cDNA representations from the head kidney of the white-blooded Antarctic icefish, Chaenocephalus aceratus (the driver).

Specifically, using polymerase chain reaction (PCR)- based representational difference analysis as a subtractive tool, the inventors obtained 316 partial cDNA clones, or probes. After identification and elimination of cDNA probes corresponding to several known genes (e. g., a-and (3-globin), 179"background-cleansed"RDA clones remained, whose sequences have been or are being determined. The preliminary sequences of the RDA clones were assembled using the program DNASTAR to yield 45 distinct contigs (i. e., overlapping, contiguous sections of a cDNA, gene, or segment of genome) that are expressed by N. coriiceps but not by the icefish, C. aceratus. The success of the present method for screening for erythropoesis-related genes has been shown by the recovery of many clones encoding the a and ß chains of hemoglobin, expressed only by N. coriiceps and not by C. aceratus. The subtraction products obtained also included others of as yet undetermined or"unknown"function. Using the representational difference products as probes for Northern blot analysis and in situ hybridization, the patterns of expression of both known and unknown genes in fish hematopoietic tissues, including blood, head kidney, and spleen, have been characterized. Furthermore, the genomes of red-and white-blooded Notothenioids have been probed for

other known erythropoietic genes, using zebrafish (Danio rerio) hematopoietic-specific markers. Genes essential for early, but not late, development of the erythroid lineage are present in the genome of the icefish C. aceratus.

Microscopic analysis has helped to identify putative proerythroblasts in the head kidney and spleen of several icefish species. Together, these results suggest that erythrocytes fail to form in icefish due to mutational blockage of the later stages of differentiation. Analysis of the subtraction products (i. e., unique to the Antarctic rock- cod head kidney, absent in the icefish head kidney) has begun to reveal novel genes, at least some of whose activities appear to be required for terminal differentiation of erythrocytes.

Thus, the invention encompasses several novel RDA clones, their corresponding full-length cDNAs, and their cognate genes that appear to have a role in blood cell formation and differentiation, particularly in erythropoesis.

The invention encompasses the identification of these gene sequences and their encoded proteins, and the application of these RDA, cDNA, and gene reagents to diagnosis and treatment of erythroid diseases (e. g., sickle cell anemia, clinical anemias resulting from chemotherapy, thalassemias, and the like). The screening method of the invention is useful for discovering gene products such as erythroid transcription factors, cytokines, cytokine receptors, and signaling enzymes.

Additionally, the invention encompasses methods for discovering genes and gene products relating to lymphopoiesis, myelopoiesis and the like, through cloning of cDNAs from the primarily white blood cell-containing hemapoietic tissues of the erythrocyte-lacking icefish.

BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWING Figure 1 shows an overview schematic of the representational difference analysis procedure (RDA) applied to subtract the red-blooded N. coriiceps head kidney cDNA (tester) by hybridization to the white-blooded C. aceratus head kidney cDNA (driver), according to the method of the invention; Figure 2 shows a Northern slot blot produced using total RNAs from various tissues of either N. coriiceps (N. c.) or C. aceratus (C. a.) ; Figure 3 depicts a Northern slot blot analysis of the tissue-specific expression patterns of five representational difference products (RDA15, RDA23, RDA34, RDA263, RDA289); Figure 4 shows the results of in situ hybridization of the a-globin RNA transcripts to blood cells of red-and white-blooded Antarctic fish species; Figure 5 shows the results of in situ hybridization of RDA15 clone; Figure 6 shows the results of in situ hybridization of the RDA197 clone; Figure 7 shows the results of in situ hybridization of the RDA23 clone; Figure 8 illustrates the morphologies of the cells of spleen tissue prints; Figure 9 shows a Southern hybridization of the murine erythropoietin receptor cDNA to genomic DNAs of different fish species; Figure 10 shows a Southern hybridization of zebrafish Band 3 cDNA to genomic DNAs of different fish species;

Figure 11 depicts a Southern hybridization of zebrafish a-spectrin cDNA to genomic DNAs of different fish species; Figure 12A depicts a schematic representation of the cDNA 23-8 clone and its coding region, while Figure 12B is a Northern slot blot showing cDNA 23-8's hybridization pattern in different organs of N. coriiceps and C. aceratus; Figure 13 is a schematic depiction of cDNA clone 23- 10/4 and its coding region; Figure 14A is a schematic depiction of the coding region of clone 15-17/13, while Figure 14B shows its hybridization pattern in N. coriiceps and C. aceratus; Figure 15A is a schematic depiction of the cDNA 162- 3 clone and its coding region, and Figure 15B is a Northern slot blot showing its hybridization pattern in different organs of N. coriiceps and C. aceratus; and Figure 16A is a schematic depiction of the cDNA 295- 5 clone and its coding region, and Figure 16B is a Northern slot blot showing its hybridization pattern in different organs of N. coriiceps and C. aceratus.

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to the discovery of hematopoiesis-related gene by means of representational difference analysis (RDA) to compare the genomes of a red-blooded species and a closely related hemoglobin- lacking, white-blooded species of the same suborder (i. e., one devoid of erythrocytes), thereby subtracting out common DNA sequences and selecting for those uniquely involved in a hematopoiesis pathway of interest. A preferred embodiment exploits a novel, alternative vertebrate model for erythropoietic gene discovery, based

on a unique group of Antarctic fish, the hemoglobin- lacking icefish.

The fifteen species of the Antarctic icefish family Channichthyidae, belonging to the suborder Notothenioidei, appear unique among vertebrates in their inability to produce red blood cells (erythrocytes).

Their blood contains normal numbers of white blood cells (lymphoid and myeloid cells) and, hence, is transparent (often referred to as"white"in appearance), in contrast to the opaque, crimson blood of other closely related Antarctic Notothenioids: e. g., Artedidraconidae (spiny plunderfish), Bathydraconidae (dragonfish), Harpagiseiidae (plunderfish), and Nototheniidae (rock- cods).

Icefish transport oxygen to their tissues solely in physical solution. In the cold (approximately-1.86°C to +1°C), stable, and oxygen-rich environment experienced by these organisms, reduction of the hematocrit to near zero is likely to have been selectively advantageous because it greatly reduces the energy cost associated with circulation of a highly viscous, corpuscular blood fluid (8-10). Zhao et al. (11) have shown previously that the icefish fail to synthesize hemoglobin due to deletion of most of the adult globin gene complex from their genomes.

The present invention is based on the inventors' hypothesis that the erythroid cell lineage fails to develop in the icefish due to mutations (e. g., deletions, point mutations) that eliminate expression of many of the genes required for red cell development and differentiation. Thus, the genomes of icefish provide a unique molecular reagent for discovery of genes and proteins involved in blood cell formation and development. Examples of potential discoveries include

genes that code for erythroid transcription factors, cytokines, cytokine receptors, clotting components, and signaling enzymes.

The invention also encompasses the use of representational difference analysis (RDA) to isolate unknown, red cell-specific genes from a red-blooded, Notothenioid fish species, by subtracting cDNA from its major hematopoietic organ, the head kidney, against cDNA from the head kidney of a white-blooded icefish species.

RDA was originally developed to analyze the differences between two complex genomes (12), and was subsequently modified to detect differential gene expression via cDNA subtraction (13). In RDA, a"tester"genome that potentially expresses genes of interest is hybridized to an excess of a"driver"genome that is not expected to, or does not, express these genes. In the invention's screening method, the tester DNA is cDNA from the head kidney of the red-blooded Antarctic rock-cod, Notothenia coriiceps, and the driver DNA is head kidney cDNA from the hemoglobin-lacking icefish, Chaenocephalus aceratus.

Preliminary experiments have detected some cDNAs encoding novel proteins that, given their observed expression patterns in hematopoietic tissues, are likely to function in erythropoiesis.

With respect to erythrocytic gene discovery, the present invention provides for isolation and identification of unique, red-cell specific genes involved in erythropoiesis, by cDNA-based RDA using head kidney cDNAs from N. coriiceps and from C. aceratus as tester and driver, respectively. The screening method is guided in part by the tissue-specific expression patterns of the subtraction products (i. e., cDNA clones unique to the red-blooded fish), as discerned by Northern slot blot

analysis and by in situ hybridization of the subtraction products to blood smears and hematopoietic tissue prints from the two species. Genomic Southern blot analysis using the N coriiceps RDA probes and erythropoietic cDNA probes from other vertebrates, has also provided information about the retention or loss of erythropoietic and other hemapoietic genes in C. aceratus. As a result of these efforts, several full-length cDNAs corresponding to several of the RDA products have been or are being obtained. Finally, the invention provides for determining the erythropoietic function of each protein encoded by every full-length clone found in the blood- forming organ of red-blooded fish but not in that of white-blooded icefish.

Specifically, an embodiment of the invention directed to screening for and isolation of unique genes for red blood cell formation, uses polymerase chain reaction (PCR)-based, representational difference analysis (RDA) to subtract cDNAs common to the head kidneys of red-and white-blooded Notothenioid fish.

Figure 1 shows an overall schematic of the RDA protocol employed, briefly summarized as follows. Tester DNA was a PCR-amplified and enzymatically digested cDNA representation from N. coriiceps head kidney. Particular adapters were ligated to the tester representation (Step 1-A) prior to subtractive hybridization, which adapters were designed to allow, in a later step, exponential PCR amplification of double-stranded (ds) cDNAs unique to the tester population. The adapters also include known tags or signature sequences (e. g., restriction endonuclease sites), allowing one to ascertain that it was the tester DNA that was selectively PCR-amplified. Driver DNA was a PCR-amplified and enzymatically digested cDNA

representation from C. aceratus head kidney. Subtractive hybridization (Step 1B) was performed by mixing the tester DNA with excess driver DNA and subjecting the mixture to appropriate conditions for dissociating or melting the DNA molecules into single strands and then to conditions for re-annealing the DNA molecules (Step 1-B).

The resulting DNA pool includes double-stranded (ds) tester DNA, single-stranded (ss) tester DNA, ds-driver DNA, ss-driver DNA, and hybrid molecules of ss-tester DNA hybridized to ss-driver DNA (i. e., representing DNA sequences common to both tester and driver populations).

The DNA mixture was subjected to conditions for filling in the single-stranded or sticky ends of the double- stranded DNA molecules, and then to PCR amplification.

This procedure resulted in the exponential amplification of unique ds-tester cDNA, linear amplification of tester/driver hybrids, and no amplification of ss-DNA.

Afterwards, single-stranded DNA species were removed by treatment with Mung Bean nuclease, and the tester-derived double-stranded DNA difference products were amplified by PCR (step 1-D), prior to digestion with at least one restriction endonuclease. The choice of restriction endonuclease will generally be guided by the particular PCT protocol used. For instance, in Example I below, DPNII, a restriction enzyme whose DNA-binding sequence encompasses four bases (termed"a four base-cutter") was used because it made frequent cuts, thereby producing many fragments of about 400-500 bp in size. Other restriction enzymes can be used (e. g., isoschizomers cutting at the same site as DPNII), as will be appreciated by one of ordinary skill in the art. Three rounds of this subtraction procedure were performed, producing an enriched"difference product"of DNA

fragments roughly 200-500 base pairs (bp) in size. The enriched difference products were then each cloned into a plasmid vector, individual clones were isolated and grown, and their DNA inserts (i. e., fragments resulting from the subtraction) were analyzed to. determine their respective sequences and other characteristics, including any functionality. (A more detailed example of a representational difference analysis in accordance with the invention may be found in the"Materials and Methods" section.) Additionally, the invention encompasses methods for discovering and applying genes and their encoded proteins, which are involved in the formation of white- blood cells. Since icefish blood contains only white- blood cells (i. e., no contaminating erthrocytes), cDNA libraries from their blood should be highly enriched for genes whose expression regulates lymphopoiesis (B and T cell formation) and myelopoeisis (myeloid cell development). Enrichment for unknown lymphopoietic and myelopoietic genes can be accomplished by subtracting a blood cDNA library from C. aceratus with an ensemble of genes that are already known to participate in white- blood cell formation. For instance, isolation of either peripheral, mononuclear lymphocytes (for T or B cell lineages) or myeloid cells (for monocyte/macrophage lines, neutrophils, and basophils), can be done through centrifugation of C. aceratus blood on Percoll and/or Ficoll density gradients and removal of the appropriate, corresponding density gradient band. The RNA of the particular cell type is used to make corresponding cDNA clones, which are then subjected to"normalization" procedures, as known in the art, to subtract out common clones, while"unique"clones are retained, resulting in

clones enriched for lymphopoiesis-or myelopoiesis- related genes. Alternatively, one could use a variation of the representational difference analysis subtractive cloning, using an icefish (e. g., C. aceratus) genome as the tester DNA for unique lymphopoiesis or myelopoiesis genes, while a red-blooded fish genome (e. g., any of the red-blooded fish of Notothenioidei suborder: spiny plunderfish, dragonfish, plunderfish, or rock-cods) serves as the driver DNA for subtracting out common clones.

Potential applications of newly identified, rare, lymphogenesis-related genes or clones include, for example, probes for diagnosis and therapeutics for treatment of leukemias (lymphomas, myelomas, etc.).

Moreover, icefish produce large numbers of thrombocytes, the fish equivalent of the platelet-forming megakaryocytes of mammals. Therefore, the gene screening and discovery approach of the invention can also be directed to discovering, from icefish, novel factors for controlling blood clotting in animals, including mammals, particularly humans. Discovery of thrombopoiesis-related cDNA and genes would require isolation of thrombocytes from a white-blooded icefish, by, e. g., centrifugation of its hemapoietic cells on a density gradient such as a Ficoll or Percoll and taking the cells from the appropriate band. Those cells (i. e., thrombocytes') RNA is prepared according to known techniques, and at least one thrombocytic cDNA library is created from the RNA, by known techniques. Normalization or substractive analysis of the icefish thrombocytic cDNA library is performed to screen out common clones. The remaining cDNA clones are screened against a suite of known thrombopoiesis genes

and analyzed for rare, previously unidentified genes involved in blood-clotting.

The novel hematopoiesis-related genes and encoded proteins thus far discovered and others currently undergoing identification by the method of the invention, are likely to offer new routes for pharmacological treatment of blood diseases.

Another species that might also serve as a good hematopoiesis gene-screening system may be immature, Anguilliform eels, which have been suggested to lack red- blood cells at early developmental stages.

The invention is further described with reference to the following, non-limiting example of using RDA to uncover new erythrocytic genes.

Example 1: PCR RDA-based screening for erythropoietic genes Using a RDA screening as described above and later in greater detail in the"Materials and Methods"section, three rounds of subtraction were performed to enhance recovery of N. coriiceps cDNAs that were not present in the head kidney of the hemoglobin-lacking icefish, C. aceratus.

Proof of the effectiveness of the present subtractive- analysis screening method in selecting for certain hemapoietic genes was established by the isolation of partial a-globin and P-globin cDNAs, which represented approximately 30% of the total, N. coriiceps-derived RDA clones (see Table I). These globins are absent from C. aceratus. Three mitochondrial cDNAs, encoding portions of NADH dehydrogenase subunits 4 and 5 and cytochrome c oxidase, were also abundant RDA products (10%, 2.5%, and

2%, respectively). To eliminate these five known, "background"genes, all 316 RDA clones were spotted onto nylon membranes and probed with the corresponding RDA fragments.

Table I.

Dot Blot Screening of Several Representational Difference Analysis Products Gene Probed No. of clones Total Percentage positive Clones of Total a-globin 38 316 12. 0% (3-globin 52 316 16.5% RDA65 6 316 1.9% RDA5 8 316 2.5% RDA2 33 316 10.5% The remaining,"background-cleansed"clones (179 total) were sequenced manually by the dideoxynucleotide chain-termination method, and the sequences were assembled to yield 45 contigs using DNASTAR. The deduced protein sequence from each contig was compared to the protein sequences of the SWISS-PROT database and to the EpoDB database of erythrocytic proteins using the BLAST-X algorithm. The contigs were divided into three groups of sequences: known, uncertain, and unknown, based on similarity to proteins in the two databases. Twenty-two percent (22%) of the 45 contigs showed strong similarity to known proteins. Examples included erythrocyte-specific ß- tubulin, a-amylase, 12-lipoxygenase, F-actin capping protein, proteosome component C13, the DEZ receptor of developing osteoblasts, and S-adenosylhomocysteinase. A second group of RDA clones was categorized as"uncertain".

These proteins (corresponding to 18 contigs or 40% of the background-cleansed isolates) gave probability scores (0.003-1.0) to known proteins that reduce confidence in the matches. Considering that the databases contain primarily mammalian protein sequences, the lower probability scores for these teleost clones do not eliminate them as potential "knowns". The uncertain matches included the GAG polyprotein, the eosinophil peroxidase precursor, the myeloperoxidase precursor, hematopoietic lineage cell- specific protein, andp-tryptase. The third group contained 17 contigs (38% of cleansed isolates) whose deduced protein sequences failed to match any proteins in the two databases. Some of these clones may have been derived from 3'-untranslated sequences of the subtracted mRNA population. Nevertheless, the isolation of full-length cDNAs, whose coding sequences corresponded to some of the unknown RDA probes, demonstrated that unique, hematopoietic genes were recovered by the subtractive strategy of the invention.

Northern Analysis of the Expression of the RDA Products To validate the subtracted clones, Northern slot blots containing total RNAs from hematopoietic (spleen, head kidney, and whole blood) and non-hematopoietic (brain, heart, liver, trunk kidney, and gills) tissues of N. coriiceps and C. aceratus were probed with cDNAs representing the 45 RDA contigs. Figure 2 shows a Northern slot blot produced using total RNAs from various tissues of either N. coriiceps (N. c.) or C. aceratus (C. a.). The blot was hybridized to the N. coriiceps 600 bp a-globin cDNA and washed at high stringency. Hybridization signals were detected after exposure to X-Ray film (Kodak) for two hours. Figure 2 shows that a-globin was strongly expressed by the hematopoietic tissues of the red-blooded rock-cod,

as anticipated, and was completely absent from tissues of the hemoglobin-lacking icefish. Hence, these results demonstrate the effectiveness of the screening method of the invention in uncovering erythropoiesis-related genes.

Figure 3 depicts a Northern slot blot analysis of the tissue-specific expression patterns of five representational difference products (RDA15, RDA23, RDA34, RDA263, RDA289). Northern slot blots were produced using N. coriiceps (N. c.) or C. aceratus (C. a.) total RNAs from various non-hematopoietic tissues brain (B), heart (H), liver (L), trunk kidney (TK), and gill (G) as well as hematopoietic tissues spleen (SP), head kidney (HK), and whole blood (BL). Figure 3 thus illustrates that five "unknown"RDAs were expressed differentially by the tissues of the two fish. In general, the RDA clones were expressed preferentially in the hematopoietic tissues of N. coriiceps. Furthermore, most of the RDA products present in the head kidney of N. coriiceps were absent from the head kidney of C. aceratus, consistent with the latter's erythrocyte-lacking phenotype. Interestingly, some RDA clones (RDA23, RDA289) expressed by the hematopoietic tissues of N. coriiceps were also found to be expressed by the spleen, liver, and/or trunk kidney of C. aceratus. These RDA clones may represent critical hematopoietic genes whose functions, in the absence of head kidney erythropoiesis, have been transferred to other sites in the icefish.

In Situ Hybridization of RDA Probes The cell-specific expression of the RDA clones was analyzed by in situ hybridization of RNAs transcribed from the probes to blood smears and tissue prints of N. coriiceps and C. aceratus. To validate the procedure, pilot studies utilizing the a-globin cDNA from N. coriiceps (14) were performed.

Figure 4 shows the results of in situ hybridization of the a-globin RNA transcripts to blood cells of red-and white-blooded Antarctic fish species. Blood smears were prepared from the red-blooded N. coriiceps and the white- blooded C. aceratus as discussed. RNA probes, either antisense or sense, were hybridized to slides in tandem.

Cells were counter-stained with Methylene green for contrast. Panels A and B represent N. coriiceps while Panels C and D represent C. aceratus. Note that the antisense transcript of a-globin stained intensely the cytoplasm of the red blood cells of N. coriiceps, as expected, whereas the sense transcript did not stain these cells. Neither antisense nor sense transcripts stained the blood cells of C. aceratus. Thus, rock cod erythrocytes contain a-globin mRNA, whereas the blood cells of icefish do not. (Blood cells were photographed using a SPOT-32 digital camera system attached to a Nikon 800 series microscope. All pictures were taken at 400X magnification.) Figure 5 shows the results of in situ hybridization of RDA15 clone. Blood smears for N. coriiceps or C. aceratus were hybridized with antisense or sense RNA probes transcribed from RDA15. Slides were counter-stained with methylene green. Panels A and B represent N. coriiceps while panels C and D represent C. aceratus. The positive signal obtained in N. coriiceps red blood cells with the

antisense RNA probe demonstrates the presence of the corresponding mRNA. The sense control showed no staining, and neither sense nor antisense RNAs gave a signal for the icefish blood cells. Photomicrographs of the slides were taken at 400X magnification on a Nikon 800 microscope.

Figure 6 shows the results of in situ hybridization of the RDA197 clone. Blood smears for N. coriiceps or C. aceratus were hybridized with antisense RNA probes transcribed from RDA197. Slides were counter- stained with methylene green. Panel A represents N. coriiceps while panel B represents C. aceratus. The mRNA corresponding to the RDA197 clone is present in blood cells of N. coriiceps and absent in those of C. aceratus.

Photomicrographs of the slides were taken at 400X magnification on a Nikon 800 microscope.

Therefore, Figures 5 and 6 demonstrate that the RDA15 and RDA197 antisense probes detected mRNAs in erythrocytes of N. coriiceps, that the sense probes gave no reaction, and that the white cells of C. aceratus blood lacked the message.

Figure 7 shows the results of in situ hybridization of the RDA23 clone. Head kidney tissue prints for N. coriiceps or C. aceratus were hybridized with sense or antisense RNA probes transcribed from RDA23. Slides were counter-stained with methylene green. Panel A represents N. coriiceps head kidney while panel B is C. aceratus head kidney. (Photomicrographs of the slides were taken at 400X magnification on a Nikon 800 microscope.) When RDA23 was hybridized to head kidney prints from N. coriiceps and C. aceratus (see Figure 7), the antisense probe produced a strong signal in nascent N. coriiceps erythrocytes, whereas cells resembling proerythroblasts from C. aceratus head kidney gave a faint color reaction. The latter result suggests that the RDA23 mRNA accumulates in late-stage

rock-cod erythrocytes and that the icefish may possess red- cell precursors that are blocked in terminal differentiation.

Library Screening of Unknown RDA Products The invention also encompasses screening a Notothenia coriiceps spleen cDNA library for full-length clones corresponding to the"unknown"RDA products isolated by the erythrocytic gene screening method. The first two clones chosen were RDA23 and RDA15.

Upon complete sequencing, the RDA23 product was found to be a chimeric clone composed of two subtraction products. Thus, two distinct clones, termed"23-8"and "23-10", were recovered from the N. coriiceps splenic cDNA library using the RDA23 product as a probe.

A schematic representation of the cDNA 23-8 clone is shown in Figure 12A, while Figure 12B is a Northern slot blot showing cDNA 23-8's comparative tissue distribution in N. coriiceps and C. aceratus. The 23-8 cDNA probe detected high levels of mRNA in the spleen, head kidney, and trunk kidney (i. e., hemapoietic organs) of N. cqriiceps, but detected significant mRNA only in the spleen and not the head kidney or blood of C. aceratus. The cDNA 23-8 clone, roughly 5 kb in length, is currently being sequenced to completion. The thus-far deduced sequence of clone 23-8 appears to encode a protein that, when compared to the protein sequences of the SWISS-PROT database, includes a conserved, approximately 100 amino-acid B30.2 domain and an adenylosuccinate synthetase domain. Interestingly, the B30.2 domain is also present in nuclear transcription factors such as the RET finger protein (a potential zinc- finger transcription factor) and pyrin, and in trans- membrane receptor or adhesion proteins like the recently described erythroid membrane-associated protein (ERMAP), a

single-pass membrane receptor possibly involved in differentiation of murine proerythroblasts (21). ERMAP appears to mediate interactions between developing red cells and the stroma of the bone marrow, and it may be involved in a tyrosine kinase signalling pathway or pathways. Thus, clone 23-8 may encode a novel membrane receptor or transcription factor involved in erythroid maturation, thereby making it a potential drug target for stimulating blood formation in anemic patients.

Clone 23-10 (also identified as cDNA 23-10/4 in Figure 13) is approximately 1 kb long and has the following sequence (SEQ ID NO: 1): GAATTCGCGG CCGCTGCCTC CTCCTGCAGC TCCTGCTTTT CAGATTCAAT 50 AGTCTCCTAC TACTGAAAAG CTCAAAACGA CTTAAATAAC ATGTCGGACG 100 AAGGTAAACT CTTCATCGGT GGTCTCAGCT TCGACACCAA CGAAGAATCT 150 CTGTCTGCGG CCTTCTCCAA GTATGGAACC ATCGAGAAGG TGGACGTGAT 200 CAGAGACAGA GAGACCGGGA ATTCCCGCGG TTTCGGCTTC GTGAAGTACG 250 ACAACGCCGA GGATGCCAAA GACGCAATGG ACGGCATGAA CGGAACCTCC 300 CTCGATGGCC GGTCAATCCG CGTCGACGAG GCAGGTGAAG GCGGTGGTCC 350 TAGGGGTGGC GGCGGATCAA GAGGCGGGCG CGGAAGAGGC GGATATGGCG 400 GTGAGAGAAG CTATGGCGGA GGAGACAGAA GCTATGGCGG CGGCGGCGGC 450 GGCGRAGGAG ACAGAAGCTA TGGCGGCGGC GGAGGAGACA GAAGCTATGG 500 CGGCGGAGGA GGAGAAAGAA GCTATGGCGA AAGATCAAGC TACGGATCCG 550 ACAGTCGTTC CGGCGGCGGC GGTGGCTACA GATCCGGCGG AGGCAGCGGC 600 GGATACTCAC GAGGTGGTGG AGGAGGAGGA GGCGGCGGCG GCTACAGAGA 650 CAACAGAGGC AGTCGCGATG GCGGATACGA ATAAAGCATT CTTCCCGCTG 700 GCTTAAAAAC AGTGGTTTAA ATATAATTGT GTTGATATGG ACTAAGTAAA 750 AGTTGTGTTT TCCCAGCAAG CTCCATTCAC GGTGACAGAG CGCCATAATA 800 CTTTGACTTT GATTTGATTT TGATAATAAA CAGCCTCAGG CCTTGTGAGT 850 CACAAATAGG TGAGTGTTTT TTTTATTGTT ATCCGGTAGC CATTGCAAGC 900 AGATGAACTC GTCGCCTAGT GACACACAAG TGACTCCRGA TAAACTCTTA 950 ACATGCTCTT TGATCCAGTT TCAACATTTT CTACTTACCT CTGCTTTTTG 1000 AGGGAGAATT CTCATAATGT CTGATCGAGA GGATAGATAG TCAGAGAGCA 1050 AAACAGTCAC CCTGAGAATA TTCTGTGAAT CCATTGGGCT AAATCCTTCG 1100 GTCAATTTCA CGGAAATATA AGATTTTTTA TTTTACCCCA CACCATGTAA 1150 CTAGTTAATG TATTTTGCCT GGGATATTTG TTGCAGGGTA AAATTGTCAA 1200 TTTCCAGTTA AAGTATGTCT AAAGAAATCC ACAATCAATC CTAAATCCTG 1250 CAATGATATC AACTCTGCTG TTCCCAGTCA GTGTAGAGTA AGTCCACACT 1300 GACTCTGTAT GTGATCATAA TAAAGGTTTA CAGCGGCCGC GAATTCTTTT 1350 GCTT (SEQ ID NO: 1)

The clone 23-10 cDNA, i. e., SEQ ID NO: 1,, can be used as a nucleic acid probe, with the probe also optionally including up to another 100 on the 5'end of SEQ ID NO: 1 (preferably in a range of about 50-100 bases), or up to about 1 kb on the 3'end of SEQ ID NO: 1 (preferably in a range of about 50-1000 bases), or both.

Computer modeling programs and comparative analysis with known sequences have helped to preliminarily identify a putative amino acid sequence encoded by cDNA 23-10.

Clone 23-10 represents the first finding, in an Antarctic fish, of what appears to be a gene encoding a cold- inducible RNA-binding protein (CIRP) that is orthologous to one found in Xenopus, XCIRP (22), with which it shares about 53% similarity. The putative N. coriiceps CIRP protein also shares about 51% similarity to the human form and 50% similarity to the mouse form. Thus, clone 23-10 appears to encode a ribosomal protein helpful in stabilizing RNA binding to ribosomes, particularly at temperatures significantly below 37°C.

Library screening using RDA15 yielded two cDNA clones, 15-9 and 15-17 (also designated cDNA 15-17/13, schematically represented in Figure 14A). Clone 15-17 (approximately 2 kb in length) was partially sequenced (SEQ ID NO: 2), as follows, and found to match to a major histocompatibility complex (MHC) class I gene: GAATTCGCGGCCGCTTGTGTTTCTGTGGATAAAGAACCTTTCACAGAGAAAGCCTCGTGA GATCAAG ATGAAGATGTTGCTTTTTCTGCTTTTCCTGGGAACAGTAGGCCCTCACTGCACATCTGCA CGGACTC ACTCTCTGATGAATTTCTACACTGCGTCCTCTGGAGTCCCAAACTTCCCAGAGTTTGTGG CTGTTGG GTTGGTTGATGACGTTGAGATGGTGCATTACGACAGCAACACCAAGAGAGCAGAGCCCAA ACAGGAC TGGATGAAGAACATCATAGACGAGGATCCTAATTACCGGGAGGGGCAGACTCAGATCTTT CAGGGTA ACCAGCAGAACTTCAAAGTCAACATTGAAATTGTAAAGCAGCGCCTCAACCAAACTGGAG GTGTCCA CATTGTCCAGCTGATGTACGGCTGTGATTGGGATGATGAGACTGGTGAGGGCAAAGGTTA TTGTAAC CATGGTTTTGATGGAGAAGACTTCATAGCACTGGACCTGGAGACAGAGACATGGACCGCT CCAAGAC GAGAGGCTGTCCTCACCAAACAGAAGTGGGATAAGAACAAAGCTCAGATGGCCCAGGACA AGAACTA CTTCACCCAGAGATGTCCTGAGTATCTGAAGAAGTATGTGAAGCTTGGGAGGAGCTCTCT GATGAGA ACAGAGATCCCCTCAATGTCTCTGCTCCAGAAGTCCCCCTCCTCTCCAGTCAGCTGCCAC GCTACAG GTTTCTACCCGGACAGAGCCGCCATGTCCTGGAGCAGAGACGGAGAGGAGCTTCAGGACG ACAACAT GGACCACGGAGAGATCCTGCCCAACCACGACGGATCCTTCCAGATAGCTTGGACCTGGAC GTCTCCT CATTCCCCGCTGAAGACTGGGGCAAGTACCGATGTGTGTTCCAGCTCTCTGGGTGAAGGA GGACCAC

GTCATCGAACTGGGCTCAGCTGTGATTCAGGACCAACAGAGGGAATCCCCTCCTCTTCAT TCATCAT CGGTGCTGAACGGCTGGTCTTCTCCTCCTCCTCCTCCTGGGATGGATTCCTGGTTACAGG AGAGGAC GACAGAAGAACACTCCTTTCGGCTTCA (SEQ ID NO: 2) Figure 14B shows that clone 15-17/13 cDNA hybridizes to mRNA in essentially all N. coriiceps tissues, especially in its blood. (Its expression in red blood cells may possibly serve as a mechanism of self-recognition.) In contrast, the blood and trunk kidney of erythrocyte-lacking C. aceratus give no or only a faint mRNA signal when probed with cDNA 15-17. This expression pattern again proves the success of the method of the invention and resulting RDA products in uncovering red-blood cell specific genes and gene products.

Clone 15-9 was found to encode P-microglobulin, and appears to be a spurious isolate.

Also isolated by the present screening method was clone 162-3, an approximately 1-kb cDNA which, based on preliminary data, appears to encode the Septin-like Protein Sint 1 (see Figure 15A). Its occurrence is high in N. coriiceps head kidney and spleen, and even stronger in liver, spleen, trunk kidney, and head kidney of C. aceratus, as shown in the Northern slot blot in Figure 15B. The septins play critical roles in cell division, probably by serving as a scaffold for assembly of signalling proteins at the cleavage furrow of dividing cells. Furthermore, fusions of septin-like genes to the MLL protein gene, through chromosomal translocation, have been implicated as mutations that may contribute to development of acute myeloid leukemia. This 162-10 clone may represent a novel septin variant, and may be useful as a nucleic acid probe consisting essentially of SEQ ID NO: 3 (i. e., SEQ ID NO: 3 optionally including up to 50-100 bases on either the 5'end, up to 1000 bases on the 3'end, or both), and as a diagnostic tool for acute myeloid leukemia.

The 162-3 product may also serve as a novel target for chemotherapeutic agents directed against various cancers.

Also isolated and sequenced has been clone 295-5, a 1- kb cDNA (SEQ ID NO: 3): GAATTCGCGGCCGCTAGGAGACAGACCTACCGGCGGTATAACGCACACCATGGCAGAGCA GGAACCC ACACCCGAGCAGCTGGAGGCGCTTGCAGCAGCCAATGACGAGCCGGAGAACCCTCTCAAC TACAAAC CCCCCGCCGCCAAGAGCCTACAGGAGATCCAGGAGCTGGACCAGGATGACGAGAGCCTGC GCAAGTA CAAGGAGACCCTGCTGGGCAACGTGGCCTGTGTGGCAGACCCCTCGGCCCCCAACGTGCA GGTGATA GGGATGACTCTGAAGTGTGAAACGGCCCCTCACCCTCTGGTGCTCGACCTGACGGGAGAC CTGACTA AATTCAAGAAGAGCCCATTTGTTCTGAAGGAAGGGGTGGAATATAGAATACAGATCAACT TCAAGGT CAACAAGGATATTGTGTCCGGCATGAAGTACACTCAGCAGTCATTCCGGGGAGGAATTAA GGTTGAG AAGTCTGACTACATGGTGGGCAGCTATGGCCCGAGACCCTCTGAGGAGTACACGTTCGTC ACCCCCA <BR> <BR> TGGAGGACGCCCCCAAAGGGATGATCGCCCGCGGCACCTACACCATCAAGTCCAAGTTCA CCGACGA CGACAAGCACGACCACCTCTCCTGGGAGTGGTGCATCGCGATCAAGAAAGACTGGACCGA ATGATTC GCCCTCCACGACTTCCTCCGCTTGGTTCCTCCTTTCCTTTCTTGCTGTCATTCCATTTTA TCTGTCT GCATTACGTGTTGATTTGAGATGTCGCTCTCTCCGTCCATCGACAACCCCCCCCCCCCCG CTCTTTA CAGGGTTTCTAAATTCCAATCACCTCCTTTCCATCCTAGCGCTTTCATTCATTTCCACCT TTCTGCT AAGCTATTCCAAAGGAGAAAGCAACCGATGCGCACGATCCAGATTTCTAGTTGGTTATTA AAGAAAA AATTAGTA (SEQ ID NO: 3) Clone 295-5 has been identified as encoding a protein homologous to the ubiquitously expressed Rho-GDI Dissociation Inhibitor 1 (see Figure 16A). However, as shown in Figure 16B, the distribution of Rho-GDI 1 appears to be specific for the blood-related organs (similar to the distribution of Rho-GDI Dissociation Inhibitor 2), giving a strong mRNA signal in N. coriiceps spleen, head kidney and blood) and no or low mRNA signals in those same organs in C. aceratus. This family of Rho-GDI factors interacts with G proteins (e. g., Rho) to control the rate of GTP/GDP exchange. Hence, the clone 295-5-encoded protein is likely to perform an important role in G-protein-coupled signalling pathways involved in blood-cell differentiation.

Determination of the actual protein sequence encoded by this cDNA is underway.

Currently, subtractive analysis products RDA263 and RDA289 are being used as probes to screen the spleen cDNA library, and sequencing of their cDNAs is in progress.

Morphological Analysis of Hematopoietic Tissues from Antarctic Fish

If icefish are blocked in an early stage of the erythropoietic program, then one would expect to recover by RDA subtraction, one or more red-cell genes encoding structural proteins. However, RDA products encoding Band 3, Band 4.1, and spectrins were conspicuous in their absence. Perhaps the icefish C. aceratus has retained "proerythroblast-like"cells in its hematopoietic tissues that are, nonetheless, blocked in terminal differentiation.

To address this question, spleen prints from both red- blooded Antarctic fish species (G. gibberifrons and N. coriiceps) and white-blooded icefish (C. aceratus, C. rastrospinosus, and P. georgianus) were examined for the presence of cells that were similar morphologically to proerythroblasts. Using Wright/Giemsa staining, proerythroblasts can be distinguished by their large, purple-pink nuclei and their small volume of deep blue cytoplasm, characteristics that are not shared by myelocytes. Figure 8 demonstrates that spleens of the three icefish species contain cells that are similar to presumptive proerythroblasts of the two red-blooded species. Thus, the morphological data are consistent with the possibility that icefish contain cells of the erythroid lineage that are unable to differentiate into mature red cells. (Specifically, for Figure 8, spleen tissue prints were stained with Wright/Giemsa and photographed at 400X magnification (panels A through D) or 600X magnification with immersion oil (panel E). White-blooded fish species were presented by: (A) Chionodraco rastrospinosus; (B) Chaenocephalus aceratus; and (C) Pseudochaenichthys georgianus. Red-blooded fish species were represented by: (D) Gobionotothen gibberifrons; (E) Notothenia coriiceps.

(Abbreviations used for cell types: ProE = proerythroblasts, lymph = lymphocytes.)

Southern Analysis of Erythropoietic Genes It has been demonstrated previously that most of the adult a/p-globin gene complex has been deleted from several species of the hemoglobin-lacking icefish (11; 14). Thus, it appeared plausible that other genes required for red cell development may also have been deleted from icefish genomes. Representational difference analysis, performed as in Example I, did not yield common erythrocytic genes, such as Band 3, Band 4.1 and others typically expressed during terminal erythrocyte differentiation. To determine whether these genes remain in icefish genomes, testicular DNAs from C. aceratus and N. coriiceps were probed with vertebrate cDNAs that encode various hematopoietic proteins. Figures 9-11 show the hybridization patterns for the murine EpoR cDNA, and for zebrafish Band 3, and a- spectrin cDNAs.

An essential component of the erythrocyte developmental program is the erythropoietin receptor (EpoR). Erythropoietin is a hormone required for the maintenance and proliferation of red cells (2). Its receptor is expressed at high levels in blast-forming unit erythroid (BFU-E) colonies, and expression declines as the BFU-E cells mature. Figure 9 shows a Southern hybridization of the murine erythropoietin receptor cDNA to genomic DNAs of different fish species. A 1.8 kb cDNA fragment for EpoR was hybridized to genomic DNAs from zebrafish, C. aceratus and N. coriiceps. Genomic DNAs were digested with PstI (lane P), HindIII (lane H), and BamHI (lane B). Molecular weight markers (BBstEII) are labeled at right. The results in Figure 9 show that the murine EpoR probe hybridized strongly to genomic DNAs from C. aceratus and N. coriiceps, yielding similar fragment

patterns. Thus, the EpoR gene, whether active or inactive, is present in the genome of a white-blooded icefish.

Band 3 is an essential multipass membrane protein of the red cell that acts as an anion transporter for removal of the waste product bicarbonate. Figure 10 shows a Southern hybridization of zebrafish Band 3 cDNA to genomic DNAs of different fish species. A 600-bp cDNA fragment for Band 3 was hybridized to genomic DNAs from zebrafish, C. aceratus and N. coriiceps. Genomic DNAs were digested with PstI (lane P), HindIII (lane H), and BamHI (lane B).

Molecular weight markers (kBstEII) are labeled at right.

Figure 10 demonstrates that the icefish genome hybridizes to a zebrafish Band 3 DNA probe, but at reduced intensity relative to that observed for N. coriiceps. This suggests that the icefish genome may contain a truncated, potentially inactive Band 3 gene.

The icefish genome was examined for the gene encoding a- spectrin. This structural protein forms a heterodimer with P-a-spectrin and polymerizes to form long thin rods found beneath the red cell membrane. Figure 11 is a Southern hybridization of zebrafish a-spectrin cDNA to genomic DNAs of different fish species. A 1.0-kb cDNA. fragment for a-spectrin was hybridized to genomic DNAs from zebrafish, C. aceratus, and N. coriiceps. Genomic DNAs were digested with PstI (lane P), HindIII (lane H), and BamHI (lane B).

(Molecular weight markers (kHindIII) are labeled at right.) Figure 11 demonstrates that sequences complementary to the zebrafish a-spectrin probe are present in red-and white- blooded Notothenioid fish.

Materials and Methods Specific details relating to a preferred embodiment of the screening method of the invention are herein described.

Collection of Fish Samples Fish were collected from Antarctic waters near Low and Brabant Islands in the Palmer Archipelago by bottom trawling from the R/V Polar Duke or the ARSV Laurence M.

Gould. The fish were transported alive to Palmer Station, Antarctica, where they were maintained in seawater aquaria at about-1°C to +1°C. Species included three icefish (Chaenocephalus aceratus, Pseudochaenichthys georgianus, Chionodraco rastrospinosus) and two rock-cods (Notothenia coriiceps, Gobionotothen gibberifrons). Zebrafish (Danio rerio) were maintained at 25°C in freshwater tanks at Northeastern University under a 12: 12 h light/dark cycle.

Tissue Sampling and Slide Preparation Whole blood (5-10 ml) was collected from live fish by caudal venipuncture using heparinized syringes. Blood cells from the icefish species were pelleted by centrifugation at 1000 rpm in an IEC clinical centrifuge at room temperature. The cell pellets were resuspended in Notothenioid Ringer's solution (260 mM NaCl, 5 mM KC1,2.5 mM MgCl2,2.5 mM CaCl2,2 mM NaHCO3,2 mM NaH2PO4,5 mM glucose) for preparation of blood smears. Whole blood from red-blooded species was used undiluted for smear preparation. Solid hematopoietic tissues (head kidney or spleen) from the five Antarctic species were dissected, and tissue prints were made by pressing tissues gently on glass slides. Microscope slides were fixed for 2-5 min in 100% methanol (room temperature) and then air-dried. Slides were stored in boxes and transported to Northeasten University from Palmer Station at ambient temperature.

Other specimens of head kidney, spleen, and blood were snap-frozen in liquid nitrogen and stored at-70°C.

Preparation of mRNA and Synthesis of cDNA Cytoplasmic RNA from hematopoietic (head kidney, spleen, and blood cells) and non-hematopoietic (brain, gill, heart, liver, trunk kidney) tissues was isolated using a modification of the acid guanidinium isothiocynate/phenol/chloroform method (14) and then stored in ethanol at B70°C. Poly (A) + mRNA was purified from the head kidneys of N. coriiceps or C. aceratus via affinity chromatography on oligo (dT)-cellulose. Double- stranded cDNA was synthesized using poly (A) + mRNA; first- strand synthesis was primed with oligo-dTl5 and second- strand synthesis was performed by standard protocols (15).

Representational Difference Analysis Representational difference analysis, as described by Hubank and Schatz (13), was performed on red-blooded N. coriiceps and white-blooded C. aceratus fish. To create representations of the head kidney cDNAs of the two fish, double-stranded cDNAs from N. coriiceps (tester) and C. aceratus (driver) (2 pg each) were digested by the enzyme DpnII (New England BioLabs). The resulting fragments were isolated by ethanol precipitation in the presence of carrier glycogen (2 gag). Oligonucleotide primers, R-Bgl-12 (5'-GATCTGCGGTGA-3') (SEQ ID NO: 4) and R-Bgl-24 (5'-AGCACTCTCCAGCCTCTCACCGCA-3') (SEQ ID NO: 5), were annealed to create double-stranded adapters. The adapters were ligated to the DpnII-digested DNA fragments from each species by incubation at 50°C for 1 minute followed by cooling to 10°C over about 1 hour. T4 DNA ligase (40 U) was added to each sample, and the samples were incubated overnight at 16°C. Ligations were diluted in TE (pH 8.0) to a final volume of 200 , 1. The ligation products were amplified in 1X PCR buffer (66 mM Tris-HCl (pH 8.8), 4 mM MgCl2,16 mM (NH2S033g/ml bovine serum albumin)

containing the four deoxynucleotides (dATP, dCTP, dGTP, dTTP [3 mM each]) and 2 ig of the R-Bgl-24 primer in a total reaction volume of 200 pl. Prior to amplification, the samples were treated at 72°C for 3 min to melt the 12- mer adapter, 5 U of Klentaq DNA Polymerase (Clontech) was added to each, and the samples were incubated for 5 min at 72°C to fill in the ends. PCR was performed for 20 cycles according to the following schedule: 1) denaturation steps, 95°C, 1 min; 2) annealing and extension steps, 72°C, 3 min; 3) a terminal extension, 72°C, 10 min; and 4) final soak at 4°C. For each fish, four PCR reactions were combined to give a total volume of 800 pl, and the samples were extracted once with phenol/chloroform/isoamyl alcohol (25: 24: 1) followed by one extraction with chloroform/isoamyl alcohol (24: 1). Representations were then precipitated and resuspended to a concentration of 0.5 mg/ml in TE.

To create the driver DNA population, amplified cDNA representations (300 g) from C. aceratus were digested with DpnII for 4 h, purified by phenol/chloroform extraction and precipitation, and resuspended to a final concentration of 0.5 mg/ml in TE. To create the tester DNA population, 20 pg of N. coriiceps cDNA representations were digested with DpnII, electrophoresed in a 1.2% agarose gel in 1X TAE, and purified using the Qiaex II gel extraction kit (Qiagen). Two p. g of purified tester were ligated to J- oligo adapters (J-Bgl-12,5'GATCTGTTCATG 3' (SEQ ID NO: 6) and J-Bgl-24 5'ACCGACGTCGACTATCCATGAAGA 3') (SEQ ID NO: 7)), under the conditions described for the R- Bgl adapters (Figure 1, step 1-A). After overnight ligation, the tester was diluted with TE to a final concentration of 10 ng/pl.

For the first subtractive hybridization, 0.4 g of the tester N. coriiceps DNA was combined with 40.0 pg of driver C. aceratus DNA (i. e., the latter is present in an excess over the amount of tester DNA provided; in this case, 100- fold, though other amounts could be used). The pooled DNAs were extracted with phenol/chloroform, and the DNA collected by ethanol precipitation. The pellet, containing both driver and tester DNAs, was resuspended in 4 pLi EE 3X buffer (30 mM EPPS, pH 8.0 (Sigma), 3 mM EDTA). The driver/tester DNA solution was heated to 37°C for 5 min, overlaid with mineral oil, denatured by incubation for 5 min at 98°C, and cooled to 67°C. One 1 of 5 M NaCl was added and the samples were incubated at 67°C for 20 h (Figure 1, step 1-B). After removal of mineral oil, 8 1 TE (pH 8.0) and 5 g/1 yeast transfer RNA (tRNA) were added to the DNA solution. An additional 25 1 of TE (pH 8.0) was added, and the solution was pipetted vigorously and then diluted to 400 gl with TE (pH 8.0). Twenty gl of this sample were diluted 1: 10 with 10X PCR buffer and 10X nucleotide mix, and the solution was heated at 72°C for 3 min to remove the 12-mer. Five U of Klentaq polymerase were added, the solution was heated at 72°C for 5 min, then 1 gl of J-Bgl-24 primer (2 mg/ml) was added. This sample, which corresponded to the first-round subtraction products, included double stranded (ds) tester DNA, single-stranded (ss) tester DNA, tester/driver hybrids, double-stranded (ds) driver DNA, and single-stranded (ss) driver DNA, with all double-stranded DNA molecules having their single- stranded ends filled in, so as to blunt-ended : The sample was amplified by PCR for 10 cycles according to the following schedule: 1) denaturation steps, 95°C, 1 min; 2) annealing and extension steps, 72°C, 3 min; 3) a terminal extension, 72°C, 10 min; and 4) cooling to room temperature

(Figure 1, step 1-C). The PCR samples were then extracted with phenol/chloroform and resuspended in 40 1 of 0.2X TE (pH 8.0). Half of this amplified sample was incubated with Mung Bean nuclease in 1X digestion buffer (New England Biolabs) for 35 min at 30°C to digest single-stranded DNA.

The reaction was stopped by addition of 50 mM Tris-HC1 (pH 8.9) and incubation at about 98°C for 5 min. After treatment with Mung bean nuclease, the samples were incubated with 2 ju. l J-Bgl-24 primer (1 mg/ml) for about 1 minute at about 95°C and cooled to approximately 80°C. Five units of Klentaq polymerase mix were added and PCR was performed for 18 cycles according to the following schedule: 1) denaturation steps,-95°C, 1 min; 2) annealing and extension steps, ~70°C, 3 min; 3) a terminal extension, -72°C, 10 min; and 4) final soak at 4°C (Figure 1, step 1- D). The PCR samples were extracted with phenol/chloroform and resuspended to a concentration of approximately 0.5 mg/ml in 100 pl TE (pH 8.0). This procedure gave the first Difference Product, termed"DPI" (i. e., the PCR strategy resulted in exponential amplification of double-stranded tester cDNA that should be enriched for DNA fragments of genes expressed by the tester N. coriiceps head kidney but not by the driver C. aceratus head kidney).

Difference Product I (DPI) was digested with DpnII as before to create representations. Primers N-Bgl-12 (5'GATCTTCCCTCG 3' (SEQ ID NO: 8)) and N-Bgl-24 (5'AGGCAACTGTGCTATCCGAGGGAA 3' (SEQ ID NO: 9)) were annealed (50°C for 1 min, then incubation at 10°C for 1 h), and ligated to the digested DPI products (50 ng/p. 1) overnight at 14°C. The ligation was diluted to 1.25 ng/pLl and 50 ng of N-ligated DPI were hybridized at 67°C with 40 , driver DNA as previously described, followed by amplification (under the same cycle parameters as for DPI),

using N-Bgl-24 as primer to create Difference Product II (DPII). DPII was digested with DpnII and annealed with J- Bgl-12/24 adapters (see protocol above). The J-ligated DPII representations were diluted to 10 pg/pl, 100 pg of tester was hybridized to 40 Fg driver, and a final PCR amplification was performed for 22 cycles (see cycle parameters for DPI) to yield Difference Product III (DPIII).

Difference Profuct II (DPIII) was digested with DpnII, and the DNA fragments were purified by Qiaex II (Qiagen) gel extraction and ligated into the BamHI site of pBluescript KS (+) II plasmid (Stratagene). Transformation into electrocompetent JM109 (25 pF, 1.5 mV in a BioRad Electroporator) was performed, and transformants were selected on LB agar plates (100 ng/pl ampicillin, IPTG and X-Gal). White colonies were selected and grown in 1X LB plus antibiotic, and minipreps were performed using the boiling lysis protocol (16).

To eliminate common clones such as a-globin, the 316 clones obtained by RDA were spotted onto Magnagraph uncharged nylon membrane (Micron Separation, Inc) and screened using previously identified, but unwanted, RDA clones. The following clones were used in screening: a- globin and P-a-globin cDNAs from N. coriiceps_ (previously cloned in the lab), cDNA clone RDA2 (encoding NADH subunit 4) and RDA10 cDNA clone (encoding cytochrome C oxidase subunit II). Blots were hybridized in Church-Gilbert solution [7% SDS, 1% BSA, sodium phosphate buffer (pH 7.4)] at 65°C overnight. Blots were washed with 0.2X SSC, 0.1% SDS at hybridization temperature. The blots were placed onto x-ray film overnight and analyzed (Table 1). After elimination of the"background" (i. e., known genes), the remaining clones were then sequenced manually by the

dideoxynucleotide chain termination method using T7 primers, USB Sequenase 2.0 (United States Biochemicals), and 35S-labeled dATP. The sequencing products were electrophoresed on 6% polyacrylamide gels. The sequences of the clones were compared to the gene sequences of the GenBank database (NCBI) using Blast 2.0 (17) and to those of the Erythropoiesis Database (EpoDB; University of Pennsylvania) (18). The complete sequences of selected difference product clones (amplicons) were obtained by automated DNA sequencing on an ABI Model 370 Sequencer (University of Maine DNA Sequencing Facility). Sequence overlaps were established using Seqman (DNAStar), yielding 45 distinct contigs.

Northern Slot Blots DNA fragments resulting from the representational difference analysis (RDA) were excised from their pBS (KS+) vector by digestion with PstI and XbaI, and the fragments were electrophoresed on 0.9% low-melting agarose gels and stored in the gel slice with water to a concentration of 5 ng/pl at-20EC. Samples (150 ng) were labeled with 32p dCTP (ICN) using a NEBlot kit (New England Biolabs). Five Rg of cytoplasmic RNA from various tissues of the tester fish, red-blooded N. coriiceps and the driver fish, white- blooded C. aceratus, were denatured in 50% formamide, 2 M formaldehyde, 1X SSC at 65°C for 5 min. RNA samples were applied to an uncharged nylon membrane (Magnagraph, Micron Separation, Inc.) using a Bio-Dot SF microfiltration apparatus (BioRad) and then rinsed extensively with 10X SSC (room temperature). RNAs were cross-linked to the membranes by exposure to UV irradiation for 45 s using a Spectra-Linker (Fisher). The blots were hybridized at 65°C overnight with labeled probes in either 50% formamide, 5X SSC, 5X Denhardt's solution, 0.2% SDS or PerfectHyb hybridization buffer (Sigma). Blots were washed to high stringency in 0.2X SSC, 0.1% SDS at 65°C. Membranes were stripped by incubation in 0.1% SDS, 0.1X SSC at 100°C for 10 min and reused a maximum of 4 times.

In Situ Hybridization To synthesize RNAs for in situ hybridization studies, RDA clones in pBS were linearized by digestion either with XbaI or with SmaI. Synthetic RNAs were labeled with digoxigenin-UTP (DIG-UTP) by transcribing the templates with either T7 or T3 RNA polymerases, to yield, respectively, antisense (experimental) and sense (control) probes following the DIG RNA Labeling protocol (Boerhinger Mannheim) (19). Briefly, linearized plasmid DNA was

extracted once with phenol followed by chloroform. After ethanol precipitation, the DNA (1 pg) was mixed with 1X transcription buffer, 1X nucleotide mix, 20 Units RNasin, 40 Units RNA polymerase (T7 or T3) and DEPC-treated water to give a final volume of 20 (J. 1. The samples were incubated at 37°C for 2 h, and DIG-labeled RNAs were precipitated by addition of 0.4 M LiCl and ethanol. The cRNA probes were resuspended in 100 Rl of DEPC-treated water and stored at-20°C until use.

For in situ hybridization (20), microscope slides containing blood smears, head kidney prints, and spleen prints from N. coriiceps or C. aceratus were placed into a humidified chamber, overlaid with 500 Rl of hybridization buffer (50% Formamide, 5X SSC, 0.1% Tween 20,5 mg/ml yeast tRNA, 50 pg/ml heparin), and incubated at 37°C for 1 h.

After removal of the pre-hybridization solution, slides were overlaid with 500 pu ouf fresh hybridization buffer containing probe RNA (200 ng/ml) and incubated at 37°C overnight in the humidified chamber. Slides were washed twice with 2X SSC 37°C, three times with 60% formamide, 0.2X SSC at 37°C, twice with 2X SSC at room temperature, and then washed with 100 mM Tris-HC1 (pH 7.5), 150 mM NaCl for 5 min at room temperature. The slides were overlaid with a saturated solution of blocking reagent (Boehringer Mannheim) prepared in 100 mM Tris-HC1 (pH 7.5) plus 150 mM NaCl and incubated at room temperature for 30 min. Anti- DIG Fab fragments were diluted 1: 200 in blocking reagent and placed onto the slides for 2 hours at room temperature. To remove unbound antibody, slides were washed twice for 5 min each in 100 mM Tris-HC1 (pH 7.5) plus 150 mM NaCl, then once for 10 min in detection buffer (Tris-HCl (pH 9.5), 100 mM NaCl, 50 mM MgCl2). The slides were placed into detection buffer with the color reagents BCIP (0.18 mg/ml),

and NBT (0.34 mg/ml) plus levamisole, an inhibitor of alkaline phosphatase (0.24 mg/ml), and allowed to react overnight at room temperature. The color reaction was stopped by addition of TE (pH 8.0) for 5 min, and slides were rinsed with distilled water prior to counterstaining with 1% Methylene Green.

Cytology For conventional cytology, fixed blood smears or tissue prints were incubated in Accustain Wright stain for 15 s, washed for 1 minute in distilled water, incubated in Accustain Giemsa for 1.5 min, and washed for 3 min in distilled water according to manufacturer's protocol (Sigma Diagnostics).

Microscopy All slides were examined using a Nikon E800 differential-interference contrast microscope equipped with 10X, 20X, and 40X dry objectives and with 60X and 100X oil objectives. Micrographs were recorded at various magnifications using a Digital SPOT32 camera (Diagnostic Instruments, Inc). Images were prepared using Adobe Photoshop 4.0 or 5.0.

Library Screening A N. coriiceps spleen cDNA library in lambda gtl0 was screened to obtain full-length clones corresponding to the RDA products. Bacteriophage were plated on C600Hf1 at 20,000 pfu/100 mm NZY plate, and the plates were incubated overnight at 37°C. Replicas containing plaque DNA were prepared on nylon filters (Osmonics, Inc.) by conventional methods (16). Filter-bound phage DNA was denatured in 1.5 M NaCl, 0.5 M NaOH for 2 min, renatured in 1.5 M NaCl, 0.5 M Tris-HC1 (pH 8.0) for 5 min, and washed for 30 s in 2X

SSC. DNA was fixed to the filters by UV photocross-linking for 45 s using a Spectra-Linker (Fisher). Filters were prehybridized in 6X SSC, 0.5% SDS, 50 g/ml heparin, 0.1% sodium diphosphate for at least 1 h at 60°C. After removal of the prehybridization solution, radiolabeled probes, prepared from RDA clones as before (see Northern Slot Blots), were added, and the filters were incubated at 60°C overnight with agitation. Filters were washed with 3X SSC, 0.5% SDS, 1 mM EDTA at 60°C at least 2 times for 15 min/wash and then exposed to X-Ray film (Fuji Film).

Positive plaques chosen for secondary screening were picked and eluted into SM buffer. All phage isolates were purified to the tertiary level. Clone DNA was prepared from miniprep cultures using the Qiagen Lambda Mini Kit.

Phage DNA inserts were sequenced by automated DNA sequencing on an ABI Model 370 Sequencer (University of Maine DNA Sequencing Facility).

Genomic Southern Blots Genomic DNA of high molecular weight was isolated from testes tissues of N. coriiceps or C. aceratus, or from whole zebrafish, by the proteinase K protocol (16) and stored in distilled H20 or TE (pH 8.0). Aliquots of DNA (10 p. g) were digested separately by addition of 40 U of PstI, HindIII, BamHI or EcoRI (New England Biolabs) at 37°C for at least 3 h, and the digests were electrophoresed on 0.8% agarose gels (Fisher) in 1X TBE overnight at 30 Volts.

DNA fragments were denatured in the gels by soaking in 1.5 M NaCl, 0.5 N NaOH twice for 30 min each, then renatured twice in 1.5 M NaCl, 0.5 M Tris-HC1 (pH 8.0) (30 min per treatment). The gels were rinsed briefly in 6X SSC, and the DNA fragments were transferred to Hybond XL nylon membrane by capillary blotting overnight in 6X SSC. DNA was fixed to the membranes by UV photocross-linking for 45

s using a Spectra-Linker (Fisher) and stored in the dark until use.

To determine the status of known hematopoietic genes in the genomes of red-and white-blooded Antarctic fish, Southern blots of genomic DNAs were hybridized to cDNA probes from other vertebrates. Zebrafish hematopoietic gene probes were kindly provided by Dr. Leonard Zon (Children's Hospital and Harvard Medical School). These included cDNAs encoding a-spectrin and Band 3. The murine erythropoietin receptor (mEpoR) cDNA was a gift from Dr.

Alan D'Andrea and Dr. Daniel Tenen. Following pre- hybridization, blots were hybridized to 32P-labeled probes overnight in 1X PerfectHyb Plus (Sigma) at 60°C (zebrafish cDNAs, and murine EpoR). Blots probed with zebrafish cDNAs were washed twice to moderate stringency (0.5X SSC, 0.1% SDS at hybridization temperature) followed by higher stringency washes (0.2X SSC, 0.1% SDS) if necessary. Blots probed with mEpoR cDNA were washed twice at moderate stringency (0.5X SSC, 0.1% SDS). Membranes were exposed to Kodak X-Ray film for 24-72 h. Membranes were stripped by incubation in 0.1% SDS, 0.1X SSC at 100°C for 10 min and reused a maximum of 3 times.

It is understood that one of ordinary. skill in the art will be able to effect minor variations to the embodiments of the methods herein described, without departing from the spirit and scope of the invention as set forth in the claims.

References 1. Weissman, I. L. (2000) Cell 100,157-168 2. Migliaccio, A. R., Vannucchi, A. M., and Migliaccio, G.

(1996) International Journal of Hematology 64,1- 29 3. Mortiz, K. M., Lim, G. B., and Wintour, E. M. (1997) American Journal of Physiology 273, R1829-R1844 4. Ward, A. C., Touw, I., and Yoshimura, A. (2000) Blood 95,19-29 5. Orkin, S. H. and Zon, L. I. (1997) Annual Review of Genetics 31,33-60 6. Pandolfi, P. P. (1998) Seminars in Hematology 35,136- 148 7. Ransom, D. G., Haffter, P., Odenthal, J., Brownlie, A., Vogelsang, E., Kelsh, R. N., Brand, M., van Eeden, F. J. M., Furutani-Seiki, M., Granato, M., Hammerschmidt, M., Heisenberg, C.-P., Jiang, Y.- J., Kane, D. A., Mullins, M. C., and Nusslein- Volhard, C. (1997) Development 123,311-319 8. di Prisco, G., D'Avino, R., Caruso, C., Tamburini, M., Camardella, L., Rutigliano, B., Carratore, V., and Romano, M. (1991) in Biology of Antarctic Fish (di Prisco, G., Maresca, B., and Tota, B., eds) pp.

263-281, Springer-Verlag, Berlin 9. Eastman, J. T. (1993) Antarctic Fish Biology: Evolution in a Unique Environment, Academic Press, San Diego 10. Macdonald, J. A., Montgomery, J. C., and Wells, R. M. G.

(1987) Advances in Marine Biology 24,321-388 11. Zhao, Y., Ratnayake-Lecamwasam, M., Parker, S. K., Cocca, E., Camardella, L., di Prisco, G., and Detrich, H. W. I. (1998) Journal of Biological Chemistry 24,14745-14752

12. Lisitsyn, N., Lisitsyn, Ni., and Wigler, M. (1993) Science 259,946-951 13. Hubank, M. and Schatz, D. G. (1994) Nucleic Acids Research 22,5640-5648 14. Cocca, E., Ratnayake-Lecamwasam, M., Parker, S. K., Camardella, L., Ciaramella, M., di Prisco, G., and Detrich, H. W. III. (1995) Proceedings from the National Academy of Sciences 92,1817-1821 15. Klickstein, L. B., Neve, R. L., Golemis, E. A., and Gyuris, J. (1995) in Current Protocols in Molecular Biology (Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. D., Seidman, J. G., Smith, J. A., and Struhl, K., eds) pp. 2-14, John Wiley & Sons.

16. Sambrook, J., Fritsch, E. F., and Maniais, T. (1989) Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Laboratory Press, Plainview, NY 17. Altschul, S. F., Madden, T. L., Schaffer, A. A., Zhang, J., Zhang, Z., Miller, W., and Lipman, D. J. (1997) Nucleic Acids Research 25,3389-3402 18. Salas, F., Haas, J., Brunk, B., Stoeckert, C. J. J., and Overton, G. C. (1998) Nucleic Acids Research 26, 288-289 19. Grunewald-Jahno, S., Keesey, J., Leous, M., van Miltenburg, R., and Schroeder, C. (1996) in Nonradioactive In Situ Hybridization Application Manual (Grunewald-Jahno, S., Keesey, J., Leous, M., van Miltenburg, R., and Schroeder, C., eds) pp. 44-45, Boehringer Mannheim GmbH, Biochemica, Mannheim

20. Dijkman, H. B. P. M., Mentzel, S., de Jong, A. S., and Assmann, K. J. M. (1996) in Nonradioactive In Situ Hybridization Application Manual (Grunewald-Jahno, S., Keesey, J., Leous, M., van Miltenburg, R., and Schroeder, C., eds) pp. 152-156, Boehringer Mannheim GmbH, Biochemica, Mannheim 21. Ye, T.-Z., Gordon, C. T., Lai, Y.-H., Fujiwara, Y., Peters, L. L., Perkins, A. C., and Chui, D. H. K.

(2000) Gene 242,337-345 22. Uochi, T. and Asashima, M. (1998) Gene 211,245-250