Chromosome 17 evolution was studied, initially, by two-color FISH of 12 single copy human BAC clones evenly distributed along the chromosome (Table 1). The probes were hybridized on metaphase chromosomal spreads of great ape species (chimpanzee (Pan troglodytes), gorilla (Gorilla gorilla) and orangutan (Pongo pygmaeus)), rhesus macaque (Macaca mulatta) as a representative Old World monkey, and three New World monkeys (marmoset (Callithrix jacchus), dusky titi (Callicebus moloch) and woolly monkey (Lagothrix lagothricha)). We performed a parallel analysis on Felix catus to serve as a mammalian outgroup. We designed 'overgo' probes corresponding to conserved sequence within each human BAC probe 10 and retrieved corresponding large insert genomic clones by hybridization against a cat genomic BAC library (RP-86). This approach facilitated comparative mapping by assembling a panel of cat probes orthologous to each of the human BAC loci (Additional data file 1). The cat clones were used in FISH experiments on metaphase spreads of F. catus and marker order was determined. In addition, we also compared the organization of human chromosome 17 and the finished sequence of the mouse (Mus musculus) orthologue (chromosome 11), the first finished mouse chromosome 13, in an effort to identify the likely mammalian ancestral state. If the centromere position is excluded, then orangutan, rhesus macaque and the New World monkeys share the same marker order. F. catus differs from this form for an inversion between the markers A and B. Zody et al. 13 have reported the same marker order arrangement on mouse chromosome 11.

The macaque organization, therefore, could be considered to represent the ancestral hominoid organization, with a paracentric inversion of the long arm responsible for the current human chromosome 17. During our FISH analysis, we narrowed the breakpoints of this rearrangement between FISH marker probes E and F and between marker probes J and K (Table 1 and Additional data file 2). Our analysis confirmed the location of the pericentric inversion in chimpanzee and chromosomal translocation in gorilla 1137. Moreover, we found evidence of two centromere repositioning events 3839 in F. catus and L. lagotricha. Figure 1 schematically summarizes our FISH results and the most parsimonious chromosomal changes necessary to reconstruct chromosome 17 evolution in primates.

The most basic approach to investigate the molecular mechanism underlying evolutionary chromosomal rearrangements is to characterize the breakpoint regions at the molecular level. We used two different approaches to define and analyze the breakpoint regions of the paracentric inversion in 17q. First, we further refined the proximal and distal breakpoints using additional human BAC probes, between marker probes E and F and between marker probes J and K. We tested 23 BACs against human and macaque in search of a clone that produced a single signal on human chromosome 17 and a double signal on both sides of the inverted segment in macaque (termed a split signal). As a result, we were able to localize the proximal breakpoint region (PBR) between the BACs RP11-115K3 (E, chr17:33,140,726-33,322,352) and RP11-923C2 (E1, chr17:33,713,298-33,818,972). Both the BACs gave a single signal in 17q12 on human metaphases. In contrast, in the macaque, E gave a single signal in a proximal not inverted position while E1 gave a signal in a distal position and it was thus included in the inverted segment (Figure 2a).

We refined the distal breakpoint region (DBR) between RP11-465I18 (J1, chr17:57,474,072-57,660,715) and RP11-50G1 (J2, chr17:57,765,687-57,954,835). Probe J1 gave a signal on 17q23 in human, but it was mapped proximally in macaque, while the BAC J2 produced a FISH signal in a distal location in both human and macaque, thus suggesting it was external to the inverted segment (Figure 2b). The most informative probes are reported in Additional data file 2 and the overall results are reported in Figure 1.

Complex duplication blocks mapping to the interval in E-E1 (390 kb chr17:33,322,352-33,713,298, named DUPA) and J1-J2 (100 kb chr17:57,660,715-57,765,687, named DUPB) complicated precise definition of the inversion breakpoints using this strategy. Four human BAC clones (Ea, Eb, Ec and Ed) spanning DUPA were tested by FISH on human and macaque. Each BAC gave duplicated signals on human chromosome 17 on both sides of the inverted segment. Notably, they also showed duplicated signals in M. mulatta in the proximal region except for the clone RP11-678G7 (Ed), which gave double signals in proximal and distal positions also on the macaque homologous chromosome (Figure 2a and Additional data file 2). Likewise, we tested three overlapping BACs (Ja, Jb, and Jc) spanning the distal DUPB breakpoint region by FISH. The BAC clone RP11-473H20 (Ja) showed duplicated signals in human 17q12 and 17q23, with duplicated signals present only in a proximal position in macaque. The remaining two clones gave double signals on both sides of the inverted segment in human and macaque (Figure 2b and Additional data file 2).

To further refine inversion breakpoints and duplication organization, we used a panel of 27 macaque-specific clones obtained from the macaque library by in silico 40 and overgo-probes/STS library screenings. Overlapping clones covering the PBR and DBR were assembled by end-sequence similarity searches against the human genome (Additional data file 3). All clones were tested by FISH.

One overgo probe, BP1/BP2 (chr17:33,612,761-33,612,796) was designed from human sequence for the PBR. We obtained only two positive clones mapping in the inverted segment (green in Additional data file 3). Gaps in the assembled human sequence prevent the further design of more useful overgo probes. BAC clone CH250-269k7 (AC140608) was mapped by BLAST sequence similarity searches, proximally (located closer to the centromere) respect to the overgo probe BP1/BP2. It produced the same hybridization pattern of the human E clone (not inverted clone) on human 17q12 and on the orthologous region in macaque. It is noteworthy that other macaque BAC clones covering the PBR gave duplicated signals both in proximal and distal regions on human chromosome 17 and proximally in macaque, thus preventing the further refinement of the PBR by FISH (Figure 2c and Additional data file 3).

We designed two STSs to the DBR: 20g10-51l5 (chr17:57,729,109-57,729,131) and SHGC-78807 (chr17:57,826,604-57,826,825). Three positive clones were obtained using 20g10-51l5 STS (blue in Additional data file 3) and further tested by FISH. They produced signals on human 17q23 and proximally on macaque, internal to the inversion. In addition, three BAC clones obtained by SHGC-78807 screening produced FISH signals distally both in human and macaque, thus mapping external to the inversion (Additional data file 3). BAC clone CH250-466D2 (only one end mapped by BLAST against human to chr17:57,734,101-57,734,926 distal to the 20g10-51l5 STS) produced signals distally in both human and macaque.

Combined, our approaches allowed us to further refine the PBR to an approximately 290 kb region between human BAC clone E, RP11-115K3 (33,322,352) and the overgo probes BP1/BP2 (33,612,761) and the DBR to an approximately 5 kb region between STS 20g10-51l5 (57,729,131) and the macaque BAC CH250-466D2 (57,734,101).

We performed a series of bioinformatics and comparative FISH analyses to provide further insight into the evolutionary history of the duplication blocks mapping to the PBR and DBR. Human duplication blocks are typically organized as mosaic structures composed of duplications of diverse evolutionary origin 41. We first considered the evolutionary architecture of each of the regions based on a recently developed algorithm designed to delineate the most likely ancestral duplication events (duplicons) within each block (Figure 3). In silico analysis was performed in the regions flanking the PBR and DBR and two more duplication clusters: DUPA' (chr17:42,332,487-42,612,217) and DUPB' (chr17:55,074,551-55,498,114) mapping between proximal and DBRs. Each duplicated block, DUPA, DUPA', DUPB' and DUPB showed a very complex organization, having unique and shared duplicon modules. The most common duplicon corresponded to the rapidly evolving TBC1D3 gene family, which mapped multiple times to DUPA, DUPB and DUPB'. We found four copies in DUPA, one in DUPB' and two entire copies of the gene in DUPB. Interestingly, by comparing the different copies of this gene both inside each block of duplication and between the two clusters, we found that the sequence similarity was higher inside each cluster (approximately 90%) than between DUPA and DUPB (87-88%). In addition, sequence analyses showed copies of the same gene, or part of it, also in 17p11/12 with a lower sequence similarity (85-86%). Similar analyses were performed on the Rhesus genome in the UCSC browser. We found five copies of the TBC1D3 gene mapping in the region orthologous to 17q12, where the PBR was previously defined, but no homologous region was detected in the orthologous region of 17q23.

To estimate the age of each duplicon within DUPA and DUPB, we analyzed the sequence identity between derived duplicons mapping within each duplication block and the presumptive ancestral loci. Within DUPA, we found that most of the duplicons showed 85-99% (mean 90.83%) sequence identity to their ancestral loci (Figure 4a), while in DUPB a shorter range of sequence identity was observed (range 87-99%; mean 95.0%; Figure 4b). Assuming a relative neutral molecular clock of evolution and the average degree of sequence identity between macaque and human (94.5%) 42, these data suggest that duplicative transposition events began to form these regions prior to the separation of human and macaque lineages in a common catarrhine ancestor. Our data further suggest that DUPA may be more ancient, composed of duplicons that show greater divergence with respect to ancestral loci when compared to DUPB. Surprisingly, an analysis of pairwise alignments between DUPA and DUPB show many segments with a high degree of sequence identity (97-99%), perhaps as a result of gene conversion or recurrent reciprocal duplication events (Figure 4c). We note, however, that these high-identity alignments between DUPA and DUPB are significantly shorter when compared to ancestral-derivative duplication events. We found no evidence of segmental duplication between DUPA' and DUPA or between DUPB' and DUPB.

In order to further define the organization and the evolutionary history of duplication blocks, we performed a series of comparative FISH experiments on great ape, Old World monkey and New World monkey metaphase and interphase chromosomes using human BAC clones corresponding to the duplication blocks (Additional data file 2). Complex hybridization patterns were observed in chimpanzee and gorilla, while the orangutan showed a pattern similar to macaque. The exact order and localization of signals were defined by FISH on interphase and stretched chromosomes. The overall results are reported in Additional data file 4. Comparative FISH analysis using the BAC clones covering DUPA' showed duplicated signals on human chromosome 17q, but not on macaque. BAC clone RP11-178C3, spanning DUPB', gave duplicated signals also on macaque in the region orthologous to 17q12 (Additional data files 2, 4 and 5).

In the New World monkey, only the BAC RP11-678G7 (Ed) mapping in DUPA gave double signals in the proximal and distal region (orange in Additional data file 4). In order to understand if the cross-hybridization signal was due to an existing duplication or a splitting signal, BAC clones flanking clones to Ed were tested. Proximal human BAC RP11-493E8 (Ec) produced signals corresponding to the orthologous 17q12 region external to the inversion. The distal BAC, RP11-923C2 (E1), gave a signal in an orthologous region to 17q23, mapping internal to the inversion. FISH experiments using BAC clones covering DUPA', DUPB' and DUPB did not detect any duplication in New World monkeys. Finally, overgo probes mapping in the duplicated regions were used to screen a cat genome BAC library for evidence of segmental duplications. No evidence of duplications was found. Zody et al. 13 have suggested a lower rate of duplication in mouse chromosome 11 orthologous to human chromosome 17, while a burst of segmental duplications in the primate lineage on chromosome 17 has been reported by She et al. 19 and Bailey et al. 1441.

Metaphase preparations were obtained from cell lines (lymphoblasts or fibroblasts) from the following species. Great apes: common chimpanzee (P. troglodytes), gorilla (G. gorilla), and Borneo orangutan (P. pygmaeus pygmaeus). Old World monkeys: rhesus macaque (M. mulatta, Cercopithecinae). New World monkeys: wooly monkey (L. lagothricha, Atelinae), common marmoset (C. jacchus, Callitrichinae), and dusky titi (C. moloch, Callicebinae); and cat (F. catus).

DNA extraction from BACs was previously reported 38. FISH experiments were performed essentially as described by Lichter et al. 47. Digital images were obtained using a Leica DMRXA2 epifluorescence microscope equipped with a cooled CCD camera (Princeton Instruments, Trenton, NJ, USA). Cy3-dCTP, FluorX-dCTP, DEAC, Cy5-dCTP and DAPI fluorescence signals, detected with specific filters, were recorded separately as gray scale images. Pseudocoloring and merging of images were performed using Adobe Photoshop™ software.

STSs (Additional data file 1) used for CH250 high density filter library-screening were chosen from the University of California Santa Cruz database (UCSC; May 2004 release) 48. Library screenings, using the human PCR products, were carried out according to a published protocol from Pieter De Jong 49. The first segment of the CH250 macaque genomic library is 6.0× redundant 50. The identification of additional BAC clones specific for rhesus macaque took advantage of specific genome browsers 4048.

Fifteen overgo probes of 36-40 bp each were designed based on conserved sequences between the human and mouse genomes according to the HomoloGene database 51, using a previously described protocol 52. The probes were hybridized to high-density filters of F. catus BAC libraries (RPCI-86; see Results) and the images were analyzed with ArrayVision Ver6.0 (Imaging Research Inc., Linton, UK). The sequence and location of overgo probes, along with clones they identified, are reported in Additional data file 2. Some overgo probes were also used to screen the rhesus macaque library (CHORI-250) to investigate the breakpoint regions (see Results).

Marker order reconstruction took advantage of the GRIMM software package, designed to outline the most parsimonious scenario of evolutionary marker order changes 5354.

In order to show the sequence homology between the putative breakpoints, we analyzed segmental duplication pairwise alignments (size ≥10 kb and sequence identity ≥95%; build35 UCSC human genome Browser) defined by the whole genome assembly comparison method 55. The ancestral origin of the duplications was determined as described 56. The duplicons were color-coded based on the cytogenetic band location of their ancestral loci. The age of the duplications at each breakpoint was also estimated by calculating sequence identity between predicted ancestral and derived duplications.