REPORT on the First International Workshop on Chromosome 9 held at Girton College Cambridge, UK, 22–24 March, 1992

The meeting was attended by 70 participants whose names and addresses are included in an appendix to this report. Fifty-four abstracts were received and the data presented on posters. The main role of the meeting was seen as the construction of a consensus map of chromosome 9, together with a sharing of knowledge about resources such as hybrids, libraries and new polymorphisms. The group divided into separate working parties to consider 9p, 9cen-9q32, 9q33-9qter, global problems and comparative mapping. The findings of each of these groups are included in this report. There was a separate discussion of resources which is also summarized. Two databases were discussed and demonstrated (GDB by Bonnie Maidak and ldb by Andy Collins). The chromosome 9 workshop was preceded on the afternoon of Sunday, 22 March by a workshop on Tuberous Sclerosis. Because of the relevance of this to chromosome 9q34 a report on this workshop is included in that section. A further subgroup met on 24 March to discuss the CEPH chromosome 9 consortium map and a brief report on their progress was produced. In accordance with the recommendation of the Global group an attempt has been made to reference information in the maps but it is recognized that this is incomplete. Within the text of this report a name without a date refers to an abstract at this meeting (see end of report). A name with a date refers to a publication listed in the references and these are generally confined to very recent or in press references. A verbal communication at the meeting is identified as a personal communication. For authoritative referencing of published information the reader should consult GDB. It is also planned to enter information into ldb so that in future it may be possible to compare a map produced by artificial intelligence with that produced by a human consensus view ! It was decided to hold a further workshop on chromosome 9 in April

The meeting was attended by 70 participants whose names and addresses are included in an appendix to this report. Fifty-four abstracts were received and the data presented on posters. The main role of the meeting was seen as the construction of a consensus map of chromosome 9, together with a sharing of knowledge about resources such as hybrids, libraries and new polymorphisms. The group divided into separate working parties to consider 9p, 9cen-9q32, 9q33-9qter, global problems and comparative mapping. The findings of each of these groups are included in this report. There was a separate discussion of resources which is also summarized. Two databases were discussed and demonstrated (GDB by Bonnie Maidak and ldb by Andy Collins). The chromosome 9 workshop was preceded on the afternoon of Sunday, 22 March by a workshop on Tuberous Sclerosis. Because of the relevance of this to chromosome 9q34 a report on this workshop is included in that section. A further subgroup met on 24 March to discuss the CEPH chromosome 9 consortium map and a brief report on their progress was produced.
In accordance with the recommendation of the Global group an attempt has been made to reference information in the maps but it is recognized that this is incomplete. Within the text of this report a name without a date refers to an abstract a t this meeting (see end of report).
A name with a date refers to a publication listed in the references and these are generally confined to very recent or in press references. A verbal communication at the meeting is identified as a personal communication. For authoritative referencing of published information the reader should consult GDB. It is also planned to enter information into ldb so that in future it may be possible to compare a map produced by artificial intelligence with that produced by a human consensus view ! It was decided to hold a further workshop on chromosome 9 in April or May 1993 and David Kwiatkowski kindly agreed to organize this in Boston.

CHROMOSOME 9p
A number of groups co-ordinated by Jane Fountain combined to produce a n ordered map of 9p (Figs 1, 2 ) . This incorporates a partial map of 9p supported by multipoint linkage analysis, pulsed field gels and FISH (Fountain et al. and Fountain et al. 1992). Most of the detailed information on 9p comes from intensive effort by several workers to investigate the position of a potential tumour suppressor gene or genes in gliomas, leukaemias, non-small cell lung carcinomas (NSCLCs) and melanomas. Deletion mapping combined with isolation of the region on a 900 kb YAC contig has led to the determination of the physical position of    Existing YAC clones include a YAC extending 220 kb centromeric to IFNA (Nobori et ul.) which must lie close to the tumour suppressor region. The tumour suppressor for melanoma has been mapped to the same general area of p21 (Fountain et ul. ALAD, ORM, D9S 16 D9S112 S116 S117 s 1 2 1 S1 15 S1 18 s120 system to introduce copies of chromosome 9 originally derived either from normal tissues or tumours with deletions of 9 (other than those deleted for MTAP) into tumorigenic cell lines to investigate the effect on tumorigenicity. The results supported the existence of a tumour suppressor gene in the region. There is still a shortage of markers on 9p distal to D9S54. Two new highly polymorphic CA repeats at loci D9S129 and D9S132 thought to lie in this region (Kwiatkowski, personal communication) were described (Hudson, personal communication to GDB ; primers listed in GDB). Two other useful CA repeats mapped to 9p by translocation breakpoints were described by Yuille et aZ. and the primers are in Table 1. New regional assignments include 1GHEP2 to 9p (Yuille et d.). A YAC recognized by D9S50 was found to hybridize to 9p21-pl3 even though the distance between D9S50 and D9S15 is about 9 cM and D9S15 has previously been shown to lie in 9q13-21.1 (Shaw et d.). Useful resources on 9p include a well characterized radiation hybrid panel derived using the selectable marker histinol-resistanceintegrated into the region p13-pl2 (Jackson, C. et d.). Cosmids mapped against this panel of radiation hybrids include a number in p24-p23 (c 132, c 165, c 198, c 219, c 238, c 101, c 77), some in p23-p21 (c 79, c 129, c 23) and a large number in the more centromeric region (p21-qll) (c 3,c 76,c 69,c 80,c 65,c 43,c 49,c 87,c 29,c 30,c 31,c 109,c 100,c 148,c 103 (Graw et al.)). Many of these have been mapped more precisely by FISH, and the order in which they have been listed gives some rough indication of the most likely order. Three cosmids have already been found to have useful CA repeat polymorphisms (c 80, c 103 and c 123) and the primers are in Table 1. CHROMOSOME Seen to 9q32 The report of the working party on this region was prepared by Allen Bale.

Linkage mapping
Based on the published data of Kwiatkowski et al. (1992) and Wilkie, P. J. et al. (1992) with data presented a t this meeting from Goudie, D. et al., Shaw, J. et al., Furlong, R. A. and Slaugenhaupt, S., a consensus map was generated showing the locations of new polymorphic markers among established loci anchored by physical localization (Fig. 3). Where distances did not agree, an average was taken. There were no discrepancies with regard to order, although there were corrections since HGMl 1 based on new typing and physical mapping by PFGE and contigs.
D9S15 and D9S8 identify the same restriction fragments and are likely t o be identical loci (Shaw, personal communication).

PFGE and Contig mapping
D9S15, S110, S111,S5 and probe GS1 are within 460 kb (Shaw, J., personal communication) by PFGE and YAC contigs. Primers D9S12 and D9S56 identify the same polymorphism (David Goudie, personal communication) confirmed by construction of cosmid contigs and by sequencing. D9S16, ORM and ALAD all lie within 1 Mb (Harris, 1991, and David Goudie, personal communication) by PFGE and cosmid contigs.

Disease loci on chromosome 9cen to 9932
The gene for Friedreich's Ataxia (FRDA) is located on proximal 9q13-21 . l . Conservative estimates of the map location would include the region between D9S48 and ALDHl but the ESSl (multiple self-healing squamous epithelioma) shows linkage disequilibrium with markers in the 9q31 region. The most likely location is between D9S29 and D9S1 based on The nevoid basal cell carcinoma syndrome gene (Gorlin Syndrome, gene symbol : NBCCS) has been mapped to 9q22.3-q31 by three groups (Farndon et al. 1992;Gailani et al. 1992;Reis et al. 1992). Six recombinants place the gene proximal to D9S53 and two recombinants place the gene distal to D9S12 (Reis et al., Farndon et al.). The maximum lod score with marker D9S53, is 19.4 a t 0 = 0.05. The majority of basal cell carcinomas, both sporadic and hereditary, show allelic loss at this location and little or no allelic loss elsewhere in the genome (Gailani et al. 1992 ;Wainwright et al.).
Two groups provided evidence for a tumour suppressor gene on 9q important in bladder carcinoma. Cairns, P. et al. studied 252 patients and found allelic loss in 5 9 % including both high and low grade tumours. The exact localization of the necessary region of allelic loss was not clear. Most tumours showed allelic loss for all chromosome 9 markers. The few tumours with loss of portions of 9 suggested that the critical region was proximal to 9q34. Jones, P. A. et al.
described similar findings in 33 tumours and have also been unable to map the putative tumour suppressor gene on chromosome 9q. CHROMOSOME 9q32-qter The report of the working party on chromosome 9q32-qter was co-ordinated by Jonathan Haines.
Extensive data on genetic order and distances in a reference set of Venezuelan families were contributed by   Loci whose position is less clearly defined are shown on the right-hand side of Fig. 4. Yuille et al. provided new regional localization for C8G, GRP78 and PAEP. Walter et al. confirmed the long-standing provisional assignment of FPGS to 9q by using this locus as a selection system in somatic cell hybrids.

Disease loci on 9q34
One form of Torsion dystonia has been previously mapped to the region between AK1 and ASS, with a strong allelic association between the Torsion dystonia gene and ASS/ABL (Ozelius, 1991 and1991b). Data presented a t this meeting from 25 small families with Report idiopathic Torsion dystonia (4 Jewish, 21 non-Jewish) suggested heterogeneity. When three clearly non-linked families were excluded, the remaining families give a lod score with ASS of 2.64 at 8 = 0.01 (Warner et d).

Tuberous sclerosis
Progress has been made in refining estimations of the map position of the Tuberous sclerosis gene on chromosome 9q34. Progress is due to extensive analysis of haplotypes for a series of highly informative markers in the 9q34 region in several large families considered to map to chromosome 9 on the basis of linkage analysis. Families were considered most likely to map to chromosome 9 if the two-point lod score (Zmax) between TSC and a chromosome 9q34 marker was above 2.0.
A t this meeting Haines presented results of re-analysis of the data from Boston TSC families, using an affecteds-only analysis. This revised analysis revealed a higher proportion of 9 linked TSC families (50%) than was obtained in data analysis which utilized information from affected and unaffecteds. This finding indicates that penetrance is an important factor to consider in Tuberous sclerosis. It is therefore important to concentrate primarily on recombination events in affected individuals in defining the position of the TSC gene.
Key recombinants were considered to be those in affected individuals in families in whom two-point lod scores between TSC and a chromosome 9q34 marker were above 2.0.
In the Hou4 family described by Northrup in which the two-point lod score between TSC and a chromosome 9q34 marker was greater than 3, a recombination event in an affected individual placed the TSC gene below GSN. I n the same family another recombination event also in an affected individual placed TSC proximal to D9S14. A recombination event which placed TSC below D9S62 was described in an affected individual (Haines and Kwiatkowski); the chromosome 9q34 TSC two-point lod score in this family was 2.1. A recombination event which placed TSC below D9S64 occurred in Duke family 507 (Kandt and Pericak-Vance). I n this family the chromosome 9q34 TSC two-point lod score was 2.0.
A recombination event which placed TSC below D9S125 was described in an affected individual in Irvine family TS 26 (Smith, Kwiatkowski, personal communication). The maximum lod score between TSC and chromosome 9q34 markers in this family was 2.4. A recombination event in an affected individual which placed TSC below D9S64 in a family with a TSC 9q34 maximum lod score of 1.2 was described by Pitiot (personal communication).
Recombination events in unaffected individuals (Nellist et a2.) were consistent with assignment of TSC to the region between GSN and DBH in a family with a 3.9 lodscore between TSC and chromosome 9 markers. A recombination event in a n unaffected individual in a family with a lodscore of 1.9 would place TSC above D9S66 (Haines et d.).  (Haines et ul.).
In summary, a conservative definition of the position of the TSCl gene would be distal to D9S64 and proximal to D9S66 (or proximal to D9S14 if considering affecteds only). There is some evidence which will need to be substantiated, that the TSCl gene may be between D9S125 and DBH, i.e. within a region that is less than 4 centimorgans and less than 2 Mb (see Fig. 4 for estimates of genetic and physical distances in this region). It will, however, be important to identify additional recombinants in affected individuals in families which are clearly chromosome 9 linked.

Other TSC loci
It is clear that genetic heterogeneity occurs in Tuberous sclerosis. Other TSC loci are not definitively assigned a t this time.
Analysis of a collaborative data-set (Janssen and the TS collaborative group, this meeting) provided no support for a locus on chromosome 11 and very little (non-significant) support for a locus on chromosome 12. Analysis of the same data by John Edwards, using a different approach, also found definite evidence only for a locus on chromosome 9 and for heterogeneity.
There was some support for a locus on chromosome 11 in three studies presented at this meeting. Studies by Haines et al. revealed a maximum lod score of 2.0 between TSC and tyrosinase in families where linkage to chromosome 9 was excluded. Data in two families analysed by Smith et al. provided evidence for linkage to markers on chromosome 11. Multipoint analysis of TSC using stromolysin, DllS84 and tyrosinase revealed a Z,,, of 3.51 with a peak a t tyrosinase.
Data presented by Benham et al. also provided some support for a locus on chromosome 11 in a small number of families. Their analysis would suggest location of a putative TSC gene below DRD2.
It was generally agreed that there must be another locus determining TSC which has not yet been found. Dr Pericak-Vance presented preliminary results on analysis of 95 additional markers scattered through the genome which have been tested on several large non-9-linked families contributed by a number of US groups. So far no clear positive results are reported and an agreement was made to extend the number of families to be tested and the coverage of the genome by sharing DNA between all willing participants and allocating each laboratory a piece of the genome not yet excluded.

NEW GENES ON CHROMOSOME 9
New assignments to distal 9q from the literature since HGM 11 included a collagen gene COL5A1 to 9q34.2-9q34.3 (Greenspan et al. 1992). A homologue of the Drosophila notch gene, thought to function in cell lineage determination during embryonic development, has been cloned from a series of leukaemia breakpoints in 9q34.3 and named TAN1 (Ellisen et al. 1991). A new homeobox gene PBX3 has been mapped t o 9q3Sq34 (Monica et al. 1991).
Two other loci assigned to chromosome I) but without regional assignment include a new form of aldehyde dehydrogenase A'LDH5 (Hsu & Chang, 1991) and platelet erythroleukemia cell PGG/H synthase (Funk et al. 1991). A gene previously mapped to chromosome 9, HSPBL2, has been excluded from 9 by more recent work (Yuille et al.). (1) Documentation on generation of maps. It would be helpful if the consensus physical and genetic maps produced as a result of specific chromosome meetings would document the generation of any specific changes and additions to current maps.
(2) Sex-specific genetic maps. Genetic maps, illustrating density of meiotic crossingover/recombination along the length of the chromosome arms are unlikely to be identical in the two sexes in the human. Sex specific mouse chiasma density and interference maps (ChiDI maps), presented a t this meeting (M . Lawrie et al., personal communication, Fig. 5 ) indicate differences in the location of hot-spots of meiotic recombination between sexes. It would be valuable if the family linkage data are analysed for each sex separately so that eventually human sex specific rather than sex averaged linkage based genetic maps may be produced. This would : (1) allow an adequate comparison with the human male chiasma map presented at this meeting (M. Hult6n and M. Lawrie, personal communication, Fig. 6) which in turn (2) may indicate more precisely the physical distance to any gene locus in the segment concerned and thus the amount of work required to identify the gene, and (3) be useful for genetic counselling in situations where disease status is still dependent on family linkage analysis.
(3) Genetic interference. It is evident that genetic interference may preclude the formation of double recombinants over quite large chromosome segments. The human male chromosome 9 chiasma density and interference map (Fig. 6) indicates that double recombinants should normally be rare within 9p, as well as over the segments 9cen-9q32 and 9q31-9qter. Within these segments male recombination fractions, obtained by family linkage, would be additive and numerically correspond to genetic distance. Any double recombinants retrieved by family linkage analysis within these segments should therefore be regarded with caution and checked for typing errors, for example. It was recognized that the information on genetic interference provided by the chiasma maps cannot be readily incorporated into currently available linkage programs. Computer programs using null interference would tend to inflate genetic distances. Therefore any program taking account of genetic interference would be an improvement over those based on null interference.
(4) Computer analysis. A discussion took place regarding the potential advantage of simplified linkage analysis, using personal computers. No consensus was reached whether or not efforts should now be made in this direction.

CEPH CHROMOSOME 9 CONSORTIUM MAP
Previous to this meeting some identification of possible errors in the original data-set had been carried out by Mark Lathrop and, after communication with various contributors and some corrections, a revised data-set had been distributed in February 1992. Preliminary maps were produced by J. Haines, J. Attwood, S. Slaugenhaupt, N. Morton, J. Yates and C. Falk (C. Falk not present). It was encouraging that there was substantial agreement between the maps. After agreement on a working framework order it was decided to re-examine those remaining unlikely double recombinants which could be tested by PCR before distribution of the final data-set within the next two months. This data-set will be used for the definitive CEPH Consortium map.

COMPARATIVE MAPPING BETWEEN CHROMOSOME 9 AND THE MOUSE
This report was prepared by Cathy Abbott and Ian Jackson. Genes mapping to human chromosome 9 have homologues mapping to three different mouse chromosomes; 2, 4 and 19. The homology on chromosome 4 falls into three segments. The simplest explanation by which the three segments are related is that there has been an insertion/deletion event and an inversion with respect to the ancestral chromosome. Figure 7 shows the conserved mouse linkages and syntenies aligned to the chromosome 9 map. The comparison allows a number of predictions to be made: (A) Predictions for chromosome 9, based on mouse data (i) TYRP (human designation) would be expected to be centromeric of IFN (mapping (ii) ACOl most likely maps to 9p, since it maps close to Galt in the mouse. (iii) SPTANl is probably telomeric of ABL, and centromeric of DBH (based on mapping (iv) SURF is probably telomeric to ABL (again based on good backcross data).
telomeric would require an additional rearrangement to be invoked).
data from three independent, large backcrosses, all of which put Spnu-2 proximal of Abl).
(B) Prediction for the mouse map, based on chromosome 9 data (i) If6 would be expected to be telomeric to Ifa in the mouse, based on physical mapping in (ii) The mouse homologue of MTAP is most likely to be on mouse chromosome 4, close t o Ifu. (iii) Rmrpr is known to be on mouse 4 and human 9p, so will map either close to Tyrpllfa or to GgtblGalt in mouse.
(iv) Ass will be probably telomeric of Abl in mouse, although they have not been genetically separated, because ASS is known to be centromeric of ABL in humans, based on interphase mapping and the use of the Philadelphia translocation.
(v) Ak-I in the mouse, which has previously only been localized using somatic cell hybrids, is probably between AsslAbl and GsnlHc.
(vi) Xpa in the mouse will probably be in Lv, Orm, Hxb region of chromosome 4. human.
(C) Mouse mutants whose human homologues would be predicted to lie on human chromosome 9 MMU2 : Sd Danforth's short tail (kidney/notochord development) stb Stubby (mild achondroplasia) lh lethargic (neurological/immunological) cold region in comparison with the near-terminal hotspots. The total chromosome 9 genetic length is estimated at 116 cM. The arrows below the diagram indicate chromosome segments where there is apparently full interference and no risk of double chiasmata/recombinants. The exclusively single chiasma in the p arm means that recombination fractions are additive and numerically equivalent to genetic distances. The 9q arm is subdivided into two major interference segments within each of which recombination fractions are again additive and numerically correspond to respective genetic distances.   al. 1990), for hybrid 8 also known as CJ9q (Jones & Kao, 1984). DNA from hybrid 64M3a12 and from GM10611 has been labelled with biotin and used to paint normal human metapham spreads. In both cams signal was seen only on chromosome 9 (West & Povey, personal communication).

Somatic cell hybrids and radiation reduced hybrids
Somatic cell hybrids containing portions of chromosome 9 are listed in Fig. 8. Radiation reduced hybrid panels were described at this meeting by hybrids by irradiating a hybrid which contained chromosome 9q. The selection applied to the post-irradiation hybrids resulted in selection for hybrids containing the human ASS gene. One hybrid, E6B, contained 5 Mb of DNA from the region between AK1 and D9S66. Jackson et al. described production of a panel of radiation reduced hybrids from the MG10611 cell line. This cell line contained a human chromosome 9 which is selectively retained by culture in histidinol, since the histidinol dehydrogenase gene (HIS) was integrated into the short arm of chromosome 9 using a retroviral vector. All hybrids in this panel therefore contain the portion of chromosome 9p where the HIS gene is located and varying size fragments extending toward both 9pter and 9qter but being biased towards 9p. Slaugenhaupt et al. also described production of a radiation reduced hybrid panel from the hybrid PK87-9 (GM 10611).

Cosmid libraries
Shaw et al. described a cosmid library produced from the hybrid PK87.9 (NG10611). In this cosmid library of 10 thousand colonies they identified approximately 200 human DNA containing colonies. Cosmid clones were subsequently localized to all regions of chromosome 9 but they were especially concentrated in human chromosome 9q. Henske et al. developed a cosmid library from a radiation hybrid which contained only the human 9q34 region. Approximately one in 1000 clones in this library were human in origin. Wolfe reported that his group had developed 160 cosmids from chromosome 9q. Armour reported the development of 80 human cosmid clones from chromosome 9q.

YACs containing regions of human chromosome 9
In the 9p region Olopade and Diaz reported isolation of a 900 kb YAC contig in the interferon gene region. Fountain described isolation of an 850 kb YAC contig in the D9S126 region. A YAC containing the D9S50 locus was isolated by Shaw and Chamberlain. A YAC contig containing the loci D9S5 and D9S15 was described previously (Chamberlain). Three YACs from the chromosome 9q34 region were described. Affara described a YAC containing the AK1 locus; Ozelius described a YAC which contained the ASS locus. Pitiot described a YAC containing the ABL locus. Details concerning a YAC HG1 which contains the 9q telomeric sequences can be found in Guerrini et al. (1990).

Cell lines
There was a general view that it would be useful to have a better panel of hybrids with well defined regions of chromosome 9. A large number of cell lines carrying translocations of chromosome 9 are available through the NIGMS mutant cell repository and are not listed here. Dr J. Yates has supplied a list of potentially useful lines available in Cambridge (some of these also from the EHCB at Porton) and this list is reproduced in Table 2  Karyotype 46XY t(5;9)(p14.2;p13.1) 46XY t(5;9)(q22;p22) 46XX t( 9 ; 13)(9pl3q ;9q 13p) 46XY t( 9 ; 19)(9pl9q ;9q 19p) 9 + der(9)t (9 ; 15)( ql1 ;ql1)