The tammar wallaby holds a unique place in the natural history of Australia, for it was the first Australian marsupial discovered, and the first in which its special mode of reproduction was noted: 'their manner of procreation is exceeding strange and highly worth observing; below the belly the female carries a pouch into which you may put your hand; inside the pouch are her nipples, and we have found that the young ones grow up in this pouch with the nipples in their mouths. We have seen some young ones lying there, which were only the size of a bean, though at the same time perfectly proportioned so that it seems certain that they grow there out of the nipples of the mammae from which they draw their food, until they are grown up' 1 . These observations were made by Francisco Pelseart, Captain of the ill-fated and mutinous Dutch East Indies ship Batavia in 1629, whilst shipwrecked on the Abrolhos Islands off the coast of Geraldton in Western Australia. It is therefore appropriate that the tammar should be the first Australian marsupial subject to an in-depth genome analysis.

Marsupials are distantly related to eutherian mammals, having shared a common ancestor between 130 and 148 million years ago 2 3 4 . The tammar wallaby Macropus eugenii is a small member of the kangaroo family, the Macropodidae, within the genus Macropus, which comprises 14 species 5 (Figure 1). The macropodids are the most specialized of all marsupials. Mature females weigh about 5 to 6 kg, and males up to 9 kg. The tammar is highly abundant in its habitat on Kangaroo Island in South Australia, and is also found on the Abrolhos Islands, Garden Island and the Recherche Archipelago, all in Western Australia, as well as a few small areas in the south-west corner of the continental mainland. These populations have been separated for at least 40,000 years. Its size, availability and ease of handling have made it the most intensively studied model marsupial for a wide variety of genetic, developmental, reproductive, physiological, biochemical, neurobiological and ecological studies 6 7 8 9 10 11 12 13 .

In the wild, female Kangaroo Island tammars have a highly synchronized breeding cycle and deliver a single young on or about 22 January (one gestation period after the longest day in the Southern hemisphere, 21 to 22 December) that remains in the pouch for 9 to 10 months. The mother mates within a few hours after birth but development of the resulting embryo is delayed during an 11 month period of suspended animation (embryonic diapause). Initially diapause is maintained by a lactation-mediated inhibition, and in the second half of the year by photoperiod-mediated inhibition that is removed as day length decreases 14 . The anatomy, physiology, embryology, endocrinology and genetics of the tammar have been described in detail throughout development 6 11 12 13 15 .

The marsupial mode of reproduction exemplified by the tammar with a short gestation and a long lactation does not imply inferiority, nor does it represent a transitory evolutionary stage, as was originally thought. It is a successful and adaptable lifestyle. The maternal investment is minimal during the relatively brief pregnancy and in early lactation, allowing the mother to respond to altered environmental conditions 11 12 15 . The tammar, like all marsupials, has a fully functional placenta that makes hormones to modulate pregnancy and parturition, control the growth of the young, and provide signals for the maternal recognition of pregnancy 14 16 17 18 . The tammar embryo develops for only 26 days after diapause, and is born when only 16 to 17 mm long and weighing about 440 mg at a developmental stage roughly equivalent to a 40-day human or 15-day mouse embryo. The kidney bean-sized newborn has well-developed forelimbs that allow it to climb up to the mother's pouch, where it attaches to one of four available teats. It has functional, though not fully developed, olfactory, respiratory, circulatory and digestive systems, but it is born with an embryonic kidney and undifferentiated immune, thermoregulatory and reproductive systems, all of which become functionally differentiated during the lengthy pouch life. Most major structures and organs, including the hindlimbs, eyes, gonads and a significant portion of the brain, differentiate while the young is in the pouch and are therefore readily available for study 11 12 19 20 21 22 23 24 . They also have a sophisticated lactational physiology with a milk composition that changes throughout pouch life, ensuring that nutrient supply is perfectly matched for each stage of development 25 . Adjacent teats in a pouch can deliver milk of differing composition appropriate for a pouch young and a young-at-foot 26 .

Kangaroo chromosomes excited some of the earliest comparative cytological studies of mammals. Like other kangaroos, the tammar has a low diploid number (2n = 16) and very large chromosomes that are easily distinguished by size and morphology. The low diploid number of marsupials makes it easy to study mitosis, cell cycles 27 , DNA replication 28 , radiation sensitivity 29 , genome stability 30 , chromosome elimination 31 32 and chromosome evolution 33 34 . Marsupial sex chromosomes are particularly informative. The X and Y chromosomes are small; the basic X chromosome constitutes only 3% of the haploid genome (compared with 5% in eutherians) and the Y is tiny. Comparative studies show that the marsupial X and Y are representative of the ancestral mammalian X and Y chromosomes 35 . However, in the kangaroos, a large heterochromatic nucleolus organizer region became fused to the X and Y. Chromosome painting confirms the extreme conservation of kangaroo chromosomes 36 and their close relationship with karyotypes of more distantly related marsupials 37 38 39 40 so that genome studies are likely to be highly transferable across marsupial species.

The tammar is a member of the Australian marsupial clade and, as a macropodid marsupial, is maximally divergent from the only other sequenced model marsupial, the didelphid Brazilian grey short-tailed opossum, Monodelphis domestica 41 . The South American and Australasian marsupials followed independent evolutionary pathways after the separation of Gondwana into the new continents of South America and Australia about 80 million years ago and after the divergence of tammar and opossum (Figure 1) 2 4 . The Australasian marsupials have many unique specializations. Detailed knowledge of the biology of the tammar has informed our interpretation of its genome and highlighted many novel aspects of marsupial evolution.

The genome of a female tammar of Kangaroo Island, South Australia origin was sequenced using the whole-genome shotgun (WGS) approach and Sanger sequencing. DNA isolated from the lung tissue of a single tammar was used to generate WGS libraries with inserts of 2 to 6 kb (Tables S1 and S2 in Additional file 1). Sanger DNA sequencing was performed at the Baylor College of Medicine Human Genome Sequencing Center (BCM-HGSC), and the Australian Genome Research Facility using ABI3730xl sequencers (Applied BioSystems, Foster City, CA, USA). Approximately 10 million Sanger WGS reads, representing about 2 × sequence coverage, were submitted to the NCBI trace archives (NCBI BioProject PRJNA12586; NCBI Taxonomy ID 9315). An additional 5.9 × sequence coverage was generated on an ABI SOLiD sequencer at BCM-HGSC. These 25-bp paired-end data with average mate-pair distance of 1.4 kb (Table S3 in Additional file 1) [SRA:SRX011374] were used to correct contigs and perform super-scaffolding. The initial tammar genome assembly (Meug_1.0) was constructed using only the low coverage Sanger sequences. This was then improved with additional scaffolding using sequences generated with the ABI SOLiD (Meug_1.1; Table 1; Tables S4 to S7 in Additional file 1). The Meug_1.1 assembly had a contig N50 of 2.6 kb and a scaffold N50 of 41.8 kb [GenBank:GL044074-GL172636].

The completeness of the assembly was assessed by comparison to the available cDNA data. Using 758,062 454 FLX cDNA sequences [SRA:SRX019249, SRA:SRX019250], 76% are found to some extent in the assembly and 30% are found with more than 80% of their length represented (Table S6 in Additional file 1). Compared to 14,878 Sanger-sequenced ESTs [GenBank:EX195538-EX203564, GenBank:EX203644-EX210452], more than 85% are found in the assembly with at least one half their length aligned (Table S7 in Additional file 1).

The tammar genome size was estimated using three independent methods: direct assessment by quantitative PCR 46 ; bivariate flow karyotyping and standard flow cytometry; and genome analyses based in the Sanger WGS reads, using the Atlas-Genometer 47 . These three approaches produced quite different genome size estimates (Tables S11 to S13 in Additional file 1) so the average size estimate, 2.9 Gb, was used for the purposes of constructing the Meug_2.0 integrated genome assembly. The smaller genome size of tammar compared to human is unlikely to be due to fewer genes or changes in gene size (Figure S1 in Additional file 2), but may be accounted for by the greatly reduced centromere size of 450 kb/chromosome and number (n = 8) 48 compared to the human centromere size of 4 to 10 Mb/chromosome (n = 23).

Novel strategies were developed for the construction of physical and linkage maps covering the entire genome. The physical map consists of 520 loci mapped by fluorescence in situ hybridization (FISH) and was constructed by mapping the ends of gene blocks conserved between human and opossum, thereby allowing the location of genes within these conserved blocks to be extrapolated from the opossum genome onto tammar chromosomes 37 (JE Deakin, ML Delbridge, E Koina, N Harley, DA McMillan, AE Alsop, C Wang, VS Patel, and JAM Graves, unpublished results). Three different approaches were used to generate a linkage map consisting of 148 loci spanning 1,402.4 cM or 82.6% of the genome 49 . These approaches made the most of the available tammar sequence (genome, BACs or BAC ends) to identify markers to increase coverage in specific regions of the genome. Many of these markers were also physically mapped, providing anchors for the creation of an integrated map comprising all 553 distinct loci included in the physical and/or linkage maps. Interpolation of segments of conserved synteny (mainly from the opossum assembly) into the integrated map then made it possible to predict the genomic content and organization of the tammar genome through the construction of a virtual genome map comprising 14,336 markers 44 .

Mapping data were used to construct tammar-human (Figure 2) and tammar-opossum comparative maps in order to study genome evolution. Regions of the genome were identified that have undergone extensive rearrangement when comparisons between tammar and opossum are made. These are in addition to previously known rearrangements based on chromosome-specific paints 50 . For example, tammar chromosome 3, consisting of genes that are on nine human chromosomes (3, 5, 7, 9, 10, 12, 16, 17, 22; Figure 2) and the X have an extensive reshuffling of the gene order. Rearrangements on the remaining chromosomes are mostly the result of large-scale inversions. This enabled us to predict the ancestral marsupial karyotype, revealing that inversions and micro-inversions have played a major role in shaping the genomes of marsupials (JE Deakin, ML Delbridge, E Koina, N Harley, DA McMillan, AE Alsop, C Wang, VS Patel, and JAM Graves, unpublished results).

The Ensembl genebuild (release 63) for the Meug_1.0 assembly identified 18,258 genes by projection from high quality reference genomes. Of these, 15,290 are protein coding, 1,496 are predicted pseudo-genes, 525 are microRNA (miRNA) genes, and 42 are long non-coding RNA genes, though these are composed of just 7 different families: 7SK, human accelerated region 1F, CPEB3 ribozyme, ncRNA repressor of NFAT, nuclear RNase P, RNase MRP and Y RNA.

Since the coverage is low, many genes may be fragmented in the assembly or even unsequenced. The Ensembl genebuild pipeline scaffolds fragmented genes using comparative data and constructs 'GeneScaffolds'. There are 10,257 GeneScaffolds containing 13,037 genes. The annotation also contains 9,454 genes interrupted by Ns. To partially ameliorate the problems of missing genes, a number of BACs from targeted locations have been sequenced and annotated, including the HOX gene clusters (H Yu, Z-P Feng, RJ O'Neill, Y Hu, AJ Pask, D Carone, J Lindsay, G Shaw, AT Papenfuss, and MB Renfree, unpublished results), major histocompatibility complex (MHC) 51 , X chromosome (ML Delbridge, B Landsdell, MT Ross, TP Speed, AT Papenfuss, JAM Graves, unpublished results), pluripotency genes, germ cell genes, spermatogenesis genes 52 53 and X chromosome genes. Findings from these are summarized in later sections of this paper.

Sequencing of the tammar genome has been augmented by extensive transcriptomic sequencing from multiple tissues using both Sanger sequencing and the Roche 454 platform by a number of different groups. Transcriptome datasets collected are summarized in Table S17 in Additional file 1 and are described in more detail in several companion papers. Sequences from the multiple tissues have been combined to assess the assembly and annotation, and to provide a resource that supplements the low coverage tammar genome by identifying and adding unsequenced and unannotated genes.

Transcriptomes of the testis [DDBJ:FY644883-FY736474], ovary [DDBJ:FY602565-FY644882], mammary gland [GenBank:EX195538-EX203564, GenBank:EX203644-EX210452], gravid uterus [DDBJ:FY469875-FY560833], hypothalamus [DDBJ:FY560834-FY602565) and cervical and thoracic thymus [SRA:SRX019249, SRA:SRX019250] were sequenced. Each dataset was aligned to the assembly (Meug_1.0) using BLASTN. The proportion of reads that mapped varied between approximately 50% and 90% depending on the tissues of origin (Figure S2a Additional file 3). Of the successfully mapped reads, the proportion aligning to annotated genes (Ensembl annotation or 2 kb up- or downstream) were more similar between libraries (Figure S2b in Additional file 3). However, the lowest rates at which reads mapped to annotated genes in the genome were observed in transcripts from the two thymuses and the mammary gland. The former is unsurprising as a large number of immune genes are expressed in the thymus and are likely to be more difficult to annotate by projection due to their rapid evolution. The lower rate at which these ESTs aligned to annotated genes in mammary gland may reflect the highly sophisticated and complex lactation of marsupials (reviewed in 12 ), a conclusion supported by the large number of unique genes identified with whey acidic protein and lipid domains (Figure 3). The mammary transcriptome may also contain a large number of immune transcripts. Together, these findings suggest a high degree of innovation in immune and lactation genes in the tammar. Previous analyses revealed that about 10% of transcripts in the mammary transcriptome were marsupial-specific and up to 15% are therian-specific 63 . Conversely, the high proportion of reads mapping to annotated genes in the testis and ovary (> 80%) suggest that there is significant conservation of active genes involved in reproduction between mammalian species (see section on 'Reproductive genes'

The testis, ovary, hypothalamus and gravid uterus full-length cDNA libraries were end-sequenced at RIKEN to evaluate composition and complexity of each transcriptome. We produced 360,350 Sanger reads in total (Table S18a in Additional file 1). Reads were clustered and the ratio of the clusters to reads was used as an estimate of the tissue's transcriptomic complexity. The hypothalamus showed the highest complexity (44.3%), whereas ovary showed the lowest (18.8%). We then looked for representative genes in each library by aligning reads to the Refseq database using BLASTN. For example, homologues of KLH10 and ODF1/2, both of which function in spermatogenesis and male fertility, were found to be highly represented in the testis library (4.3% and 3.5% respectively). The hypothalamus library was rich in tubulin family genes (7.9% of reads), and hormone-related genes such as SST (somatostatin; 1.8% of reads) (see Table S18b in Additional file 1 for details).

The organization of the tammar MHC is vastly different from that of other mammals 81 82 . Rather than forming a single cluster, MHC genes are found on every chromosome, except the sex chromosomes (Figure 2). The MHC itself is found on chromosome 2q and contains 132 genes spanning 4 Mb 51 . This region was sequenced using a BAC-based Sanger sequencing strategy as it did not assemble well from the low-coverage sequencing. An expansion of MHC class II genes is accompanied by duplication of antigen-processing genes. The seven classical MHC class I genes are all found outside the core MHC region. KERVs may have contributed to this re-organization (Figure 2).

The tammar wallaby has two thymuses: a thoracic thymus (typically found in all mammals) and a dominant cervical thymus. Based on digital gene expression profiles both thymuses appear functionally equivalent and drive T-cell development 83 . Transcriptomic sequencing also shows that both thymuses express genes that mediate distinct phases of T-cell differentiation, including the initial commitment of blood stem cells to the T lineage (for example, IL-7R, NOTCH1, GATA3, SPI1, IKZF1), the generation of T-cell receptor diversity and development of the thymic environment (for example, TRAF6, TP63 and LTBR). In the thymus transcriptomes, we identified and annotated 34 cytokines and their receptors (10 chemokines, 22 interleukins and 2 interferons), 22 natural killer cell receptors (20 leukocyte receptor complex (LRC) genes and 2 natural killer complex (NKC) genes), 3 antimicrobial peptides (2 beta-defensins and 1 cathelicidin), post-switch immunoglobulin isotypes IgA and IgG and CD4 and CD8 T-cell markers.

At birth, the altricial pouch young is exposed to a variety of different bacterial species in the pouch. These include Acinetobacter spp., Escherichia coli and Corynebacteria spp. 84 . These bacteria remain in the pouch despite the female tammar extensively cleaning the pouch by licking before birth. To survive in this pathogen-laden environment, the immunologically naive neonate is reliant on immune factors, which are transmitted from the mother through the milk. The sequencing of the genome uncovered a family of cathelicidin genes, which are expressed in the mammary gland during lactation and encode powerful antimicrobial peptides. These peptides may provide unique opportunities to develop novel therapeutics against emerging multidrug-resistant superbugs.

Due to the rapid evolution of immune genes, a high proportion of tammar immune genes were not annotated using automated annotation pipelines. For this reason an Immunome Database for Marsupials and Monotremes has been established 85 . This database contains over 5,000 marsupial and monotreme immune sequences from a variety of EST projects, as well as expert-curated gene predictions. Marsupial chemokine, interleukin, natural killer cell receptor, surface receptor and antimicrobial peptide gene sequences are also available. Genomic evidence confirms that the marsupial immune system is on par with the eutherian immune system in terms of complexity.

Marsupial sex chromosomes have been shown to represent the ancestral sex chromosomes, to which an autosomal region was fused early in the eutherian radiation. Thus, the basic marsupial X shares homology with the long arm and pericentric region of the human X 35 36 . The tammar Y shares only five genes with the degraded eutherian Y 86 (Figure 5).

Marsupial sex chromosomes lack the autosomal addition and so are expected to be smaller than those of eutherian mammals. The opossum X is about 97 Mb (Table S12 in Additional file 1). The larger size of the tammar X (150 Mb) reflects the addition of a heterochromatic arm containing satellite repeats and the nucleolus-organizing region 59 . Of the 451 protein coding genes on the opossum X chromosome, 302 have orthologues in the tammar Ensembl gene build. Gene mapping indicates that the gene order within the tammar X is scrambled with respect to both the opossum and human X chromosomes 37 . This scrambling of the marsupial X contrasts to the eutherian X chromosome, which is almost identical in gene content and order between even the most distantly related taxa 87 88 . The rigid conservation of the eutherian X was hypothesized to be the result of strong purifying selection against rearrangements that might interrupt a chromosome-wide mechanism to effect X-chromosome inactivation. Consistent with this hypothesis, inactivation on the scrambled marsupial X is incomplete, locus-specific, and does not appear to be controlled by an inactivation center 54 56 .

In many marsupial species the Y chromosome is a minute element of about 12 Mb. The tammar Y is larger, as the result of the addition to the X and Y in the early macropodid radiation of a heterochromatic long arm that contained the nucleolar organizing region (NOR) and NOR-associated repeats 59 . Degradation of the Y removed active rDNA genes but left repetitive sequences with homology to the NOR-bearing short arm of the X 89 90 . The tammar Y chromosome bears at least ten genes, which are all located on the tiny short arm of the Y (reviewed in 91 ) (V Murtagh, N Sankovic, ML Delbridge, Y Kuroki, JJ Boore, A Toyoda, KS Jordan, AJ Pask, MB Renfree, A Fujiyama, JAM Graves and PD Waters, unpublished results). All ten have orthologues on the Y of a distantly related Australian dasyurid marsupial, the Tasmanian devil, implying that the marsupial Y chromosome is conserved (Figure 5). It has degraded more slowly than the eutherian Y, which retains only four (human) or five (other mammals) genes from the ancient XY pair 91 92 .

Like most genes on the human Y, all of these tammar Y genes have an X partner, from which they clearly diverged. Some tammar Y genes are expressed exclusively in the testis (for example, the marsupial-specific ATRY 93 ), but most have widespread expression. Phylogenetic analysis of the X and Y copies of these ten tammar XY genes indicate that marsupial Y genes have a complex evolutionary history.

Epigenetic silencing of one X chromosome occurs in female mammals as a means of dosage compensation between XX females and XY males. Classic work on kangaroos established that X inactivation occurs in marsupials, but is paternal, incomplete and tissue-specific 94 and apparently occurs in the absence of the XIST controlling element 95 96 . Using tammar sequence to isolate X-borne genes and study their expression at the level of individual nuclei using RNA in situ hybridization, it has been found that different genes have a characteristic frequency of expression from one or both loci, suggesting that it is the probability of expression rather than the rate of transcription that is controlled 54 . The absence of clustering of high- or low-expressing genes has not so far provided evidence for an inactivation center. It appears that X inactivation in marsupials, like eutherians, uses a repressive histone-mediated gene silencing, and although inactive marks are not identical 55 56 , they do have H3K27 trimethylation and targeting to the perinucleolar compartment 97 .

Marsupials differ from eutherian mammals primarily in their unique mode of reproduction. In contrast to mice and humans, in which sexual differentiation occurs in utero, the altricial 440 mg tammar neonate has indifferent gonads on the day of birth and does not undergo gonadal sex determination until approximately 2 days later (testis) and 8 days later (ovary) 22 . This postnatal differentiation of the gonads therefore provides an unparalleled model for studying sex determination and sexual differentiation and enables experimental manipulation not possible in eutherian species. We have shown that almost all genes critical for testis and ovarian development are highly conserved between the tammar, mouse and human at the molecular level 98 99 , but their precise role in gonadogenesis may differ between the mammalian groups.

The segregation of the first cell lineages and specification of embryonic and extra-embryonic cell lineages have been studied extensively in the mouse. However, the mouse has a highly specialized embryogenesis, quite different from that of other mammals. Unlike a typical eutherian blastocyst with its inner cell mass, the tammar conceptus forms a unilaminar blastocyst of approximately 100 cells that lacks a readily defined pluriblast in the form of an inner cell mass. It can undergo a prolonged period of diapause. Thus, these differences highlight the developmental plasticity of mammalian embryos and genome analysis may provide comparative data that clarify the underlying control mechanisms of early mammalian development.

Genomic imprinting is a widespread epigenetic phenomenon characterized by differential expression of alleles, depending on their parent of origin. Imprinted genes in eutherian mammals regulate many aspects of early growth and development, especially those occurring in the placenta. Most, but not all, genes that are imprinted in mouse and human have orthologues in the tammar genome; an exception is the Prader-Willi-Angelman syndrome region containing SNRPN and UBE3A, which does not exist in tammar, nor in monotremes, so was evidently recently constructed in eutherians by fusion and retrotransposition 129 . Some tammar orthologues of genes that are imprinted in eutherians are not imprinted 130 131 . So far the orthologues of 13 eutherian imprinted genes examined have a conserved expression in the marsupial placenta, but only 6 of these are imprinted in marsupials 132 133 .

Marsupial orthologues of the classically imprinted IGF-2 receptor (IGF2R), insulin (INS) or paternally expressed gene 1/mesoderm specific transcript (PEG1/MEST) also show parent-of-origin expression in marsupials. However, some genes that are imprinted in eutherians, such as Phlda2 in the KCNQ1 domain, a negative regulator of placental growth, are not imprinted in the tammar 134 . This demonstrates that acquisition of genomic imprinting in the KCNQ1 domain occurred specifically in the eutherian lineage after the divergence of marsupials, even though imprinting of the adjacent H19-IGF2 domain 135 arose before the marsupial-eutherian split. A similar scenario applies to DLK1, DIO3 and RTL1 (PEG11), which are not imprinted in marsupials 130 136 .

Differentially methylated regions (DMRs) are the most common signals controlling genomic imprinting in eutherian mammals. However, no DMRs were found near the tammar orthologues of the classically imprinted genes IGF2R, INS or PEG1/MEST, although these genes still showed parent of origin specific expression differences. Other marsupial imprinted genes (H19, IGF2 and PEG10) do have DMRs, indicating that this mechanism of gene control evolved in the common therian ancestor at least 140 million years ago 133 . Using comparisons with the tammar genome, we have been able to reconstruct the emergence of an imprinted gene - PEG10 137 . PEG10 is derived from a retrotransposon of the suchi-ichi family and was inserted after the prototherian-therian mammal divergence. This demonstrates that retrotransposition can drive the evolution of an imprinted region with a DMR 137 . In contrast, another retrotransposed gene also of the suchi-ichi family, SIRH12, has been identified specifically in the tammar genome but is not seen in eutherians. It appears to be tammar-specific since it is absent from the opossum genome. Its imprint status has yet to be ascertained 138 .

The insulator genes CTCF (CCCTC-binding factor) and its paralogue BORIS (brother of regulator of imprinted sites) have orthologues in the tammar genome, and as in mouse, CTCF is expressed ubiquitously and BORIS is expressed in gonads. The existence of both genes in the monotreme and reptile genomes but the ubiquitous expression of BORIS in these species suggests that this gene became gonad-specific in therian mammals, coincident with the evolution of imprinting 139 .

Although all imprinted genes so far identified in the mouse are expressed in the placenta, the few mouse genes that have been knocked out (for example, Grb10, Peg3 ) that are also imprinted in the fetal brain have marked behavioral effects 140 . We now know that there are additional autosomal genes in the cortex and hypothalamus with sex-specific imprinting 141 142 , so we can expect an increase in the identification of imprinted brain genes that influence behavior. Since a large proportion of known imprinted genes also have a role in postnatal growth and nutrient supply, and marsupials depend much more on lactation than most other mammals (see below), it is possible that genomic imprinting might function in the marsupial mammary gland as it does in the placenta. Transcription analysis has confirmed that two genes critical for the onset of lactation in the tammar, IGF2 and INS, are imprinted in the tammar mammary gland throughout the long period of lactation (JM Stringer, S Suzuki, G Shaw, AJ Pask, and MB Renfree, unpublished observations).

Lactation is a defining character of mammals 148 149 . This is especially true of marsupials that give birth to highly altricial young that depend upon milk for growth and development during a relatively long lactation period. The marsupial mother dramatically alters milk production and composition throughout lactation, specifically for each stage of development of the pouch young 26 150 151 . They are even able to produce milk of differing compositions from adjacent mammary glands, a phenomenon known as concurrent asynchronous lactation (reviewed in 152 ).

Lactation in the tammar extends for approximately 300 days and is divided into 3 phases based upon the sucking pattern of the young (phase 1 (late pregnancy-birth), lactogenesis; phase 2A (day 0 to 100), permanently attached to the teat; phase 2B (day 100 to 200), intermittently sucking and confined to the pouch; phase 3 (day 200 to 300), in and out of the pouch), accompanied by changes in milk composition and mammary gland gene expression 26 . The tammar mammary gland transcriptome consists of two groups of genes 63 . One group is induced at parturition and expressed throughout lactation, as in eutherians. These genes include the milk protein genes encoding α-, β-, and κ-casein (CSN1, CSN2 and CSN3) and the α-lactalbumin (LALBA) and β-lactoglobulin (LGB) whey protein genes. However, the tammar genome lacks additional copies of α- or β-like caseins that are present in monotremes and eutherians (Figure S3 in Additional file 4).

The second group of mammary genes is expressed only during specific phases of lactation. This group includes marsupial-specific milk protein genes such as the late lactation proteins (LLPA and LLPB) as well as others such as whey acidic protein (WAP) 153 that are also found in milk of many eutherians 154 but lacking in humans, goat and ewe 155 . Evidence is now emerging that changes in composition of the major milk proteins and many bioactives 156 157 contribute to a more central role of milk in regulating development and function of the mammary gland 158 to provide protection from bacterial infection in the gut of the young and the mammary gland 159 (A Watt and KR Nicholas, unpublished results) and to deliver specific signals to the young that regulate growth and development of specific tissues such as the gut 160 . There is also a novel putative non-coding RNA (PTNC-1) expressed in the mammary gland throughout lactation. PTNC-1 is derived from a region of the genome that is highly conserved in mammals, suggesting it may have an important functional role 63 . Tammar ELP (early lactation protein), originally thought to be marsupial-specific (phase 2A) 63 , has a eutherian orthologue, colostrum trypsin inhibitor (CTI), which is present in some eutherians but is reduced to a pseudogene in others (EA Pharo, AA De Leo, MB Renfree, and KR Nicholas, unpublished results). The ELP/CTI gene is flanked by single-copy genes that map to orthologous regions of the genome - strong evidence that ELP/CTI evolved from the same ancestral gene. ELP/CTI has not yet been detected in monotremes. Other marsupial-specific milk protein genes identified include trichosurin and the putative tammar milk proteins PTMP-1 and PTMP-2 63 . Remarkably, the tammar PTMP-1 gene has been identified in the tammar genome sequence, but does not seem to occur in the genome sequence of the short-tail grey opossum. Thus, PTMP-1 may be macropodid-specific.

The tammar, a small kangaroo species, is the model Australian marsupial that has played a particularly important role in the study of reproduction, development, immunity and the evolution of the mammalian sex chromosomes. Here, we have presented its genome sequence and associated resources, including transcriptome sequence data from a range of tissues. Together these data have provided new insights into a host of important gene families. We identified novel tammar-specific, as well as conserved but previously undiscovered, miRNAs that regulate the HOX genes, a novel SINE class that is rRNA-derived and a novel class of small RNAs. We show that there has been expansion of several gene families, especially of the MHC and OR genes, that there are features that are of specific importance to marsupials, such as the innovation of genes in lactation and the presence of genomic imprinting in the mammary gland. However, there is high conservation in testicular and ovarian genes, one of which, DHH, is only the second mammal-specific gonadal development gene so far identified. The Y chromosome is minute but relatively gene rich and conserved in marsupials. The X chromosome reflects the ancestral mammalian X and perhaps an ancestral stochastic dosage compensation that operates without an X chromosome inactivation center. These initial tammar genome analyses have already provided many unique insights into the evolution of the mammalian genome and highlight the importance of this emerging model system for understanding mammalian biology.

Materials and methods are briefly described in the body of the paper and extensively in the supplementary methods (Additional file 1).

Public database accessions are provided for all raw datasets where they are first mentioned in the text. The latest version of the genome assembly is available in NCBI under the GenBank accession ABQO000000000; Meug_1.1 has accession ABQO010000000; Meug_2.0 has accession ABQO020000000. All versions of the genome assembly are also accessible via the web 161 .

BAC: bacterial artificial chromosome; BCM-HGSC: Baylor College of Medicine Human Genome Sequencing Center; bp: base pair; crasiRNA: centromere repeat-associated short interacting RNA; DHH: Desert hedgehog; DMR: differentially methylated region; EST: expressed sequence tag; GH: growth hormone; IFITM: interferon inducible transmembrane protein; IGF: insulin-like growth factor; KERV: kangaroo endogenous retrovirus; LINE: long interspersed nuclear element; LTR: long terminal repeat; MHC: major histocompatibility complex; miRNA: microRNA; NOR: nucleolar organizing region; OR: olfactory receptor; PGC: primordial germ cell; piRNA: Piwi-interacting RNA; SINE: short interspersed nuclear element; VNO: vomeronasal organ; VNR: vomeronasal receptor; WGS: whole-genome shotgun.

All authors were members of the tammar wallaby genome sequencing consortium. Authors contributed to sequencing, assembly, analysis, experiments or writing as follows (team leaders are in bold). Joint lead authors: MBR, ATP. Principal investigators: JAMG, SMF, RAG, DWC, MBR, TPS, AF. Tammar genome size: AJP, WR, RJO'N, KCW, MFS. Physical and linkage mapping: JED, CW, FWN, KRZ, JAMG. Highly divergent or tammar-specific transcripts: AJP, AF, TH, YK, HY, MBR. Transcriptome: ATP, AF, AJP, MBR, YK, Z-PF, YSuz, SS, AT, YSak, SK, YN, ST. Genebuild: SMJS, SF. Sequence conservation and gene family expansion: PF, ESWW, KBl, JH. GC content: WR, RJO'N, JL. Analysis of Repeats and small RNAs: RJO'N, JL, DMC. Immunity: KB, ESWW, HVS, JED, MBR, KRS, CW, JW, BGC, ATP. Sex chromosomes: PDW, JAMG, JED. X chromosome inactivation: JAMG, PDW, JED, SAN. Reproductive genes: AJP, MBR, GS, DH, WO'H, YH. Developmental genes: MBR, AJP, GS, SRF, HY, K-YC, BRM, RJO'N, JL. Genomic imprinting: MBR, AJP, JAMG, GS, JS, SS, TAH. Olfaction: MBR, JAMG, SRF, AM, MLD, NYS, GS. Lactation: CML, KRN, EAP. Additional bioinformatics contributions: AH, BL, KAM, MJW. Australian Genome Research Facility Sanger sequencing: SMF, EK, AMG, PW, AM, JE, CT, DT, AS, LY, TL, MH-R, AH, JD, DW, SW, YSun. HGSC leadership: KCW, DMM, RAG. HGSC Sanger production: SNJ, LRL, MBM, GOO, SJR, JS, LN, AC, GF, CLK, HHD, VJ. HGSC SOLiD production team: HHD, CJ. HGSC 454 production: CLK, FL, RT. HGSC genome assembly and analysis: LC, JD, YL, JYS, X-ZS, GW, KCW. UConn sequencing and assembly improvement (Meug_2): RJO'N, AJP, JL, TH, IM. Senior authors: SMF, JAMG, RJO'N, AJP, KCW. Senior authors contributed equally and principal investigators contributed equally.