Turkish Journal of Biology
Turk J Biol (2014) 38: 880-897 © TÜBİTAK doi:10.3906/biy-1404-1
http://journals.tubitak.gov.tr/biology/
Research Article
Identification of new genes required for the maintenance of chromosome integrity in Drosophila melanogaster 1
1
2,
Francesca CEPRANI , Franco SPIRITO , Roberto PIERGENTILI * Department of Biology and Biotechnologies, Sapienza University of Rome, Rome, Italy 2 Institute of Biology, Molecular Medicine, and Nanobiotechnologies at the National Research Council (CNR), Sapienza University of Rome, Rome, Italy 1
Received: 01.04.2014
Accepted: 21.07.2014
Published Online: 24.11.2014
Printed: 22.12.2014
Abstract: Although genome-wide RNA interference (RNAi) screens for mitotic genes are not new in the literature, most of them lack the cytological characterization of the cell karyotype as for chromosome integrity. Here, the effects of RNAi on chromosome structure in S2 cultured cells of Drosophila melanogaster were analyzed. The cytological phenotype of 1132 genes selected by coexpression with known mitotic genes was scored. Cytological and statistical analysis of the treated cells allowed the identifying of 81 loci whose inactivation brings a level of chromosome breakage significantly higher than in the control. Many of the genes characterized in the present work had never been associated with a cellular function; in other cases, their putative role is apparently unrelated to the chromosome breakage phenotype. These results suggest novel biological roles for the proteins encoded by the identified genes and indicate that the number of loci required for chromosome integrity is much larger than expected. Moreover, these results strongly suggest that the compilation of a list of coexpressed genes for any given function would result in a largely incomplete set of data, and that quite surprisingly a more complete collection of loci may be obtained using screening criteria different from selection. Key words: Cancer, functional coexpression, genome instability, karyotype, S2 cultured cells, segmental aneuploidy, RNAi
1. Introduction RNA-interference (RNAi) is experimentally used for gene silencing via a double-stranded RNA (dsRNA) targeting a complementary messenger RNA (mRNA) and promoting its degradation by a nuclease-dependent cut. Although this acronym was first used by Fire et al. (1998) in studying the model organism Caenorhabditis elegans, this biological phenomenon had already been observed before, though not completely understood, in several organisms such as transgenic plants (Ecker and Davis, 1986; Napoli et al., 1990), Neurospora crassa (Romano and Macino, 1992), Caenorhabditis elegans (Guo and Kemphues, 1995), and Drosophila melanogaster (Pal-Bhadra et al., 1997), indicating that RNAi is widely present in most eukaryotes. Notably, some eukaryotes lack all or most of the RNAi machinery; among others, these include some protozoa (Robinson and Beverley, 2003; DaRocha et al., 2004), and several fungi including Saccharomyces cerevisiae (Aravind et al., 2000; Nakayashiki et al., 2006; Drinnenberg et al., 2009). Some researchers suggest that this mechanism – probably a defense against exogenous, potentially harmful dsRNA such as that of viruses or transposable elements * Correspondence:
[email protected]
880
– evolved very early and was already present in the first, primitive eukaryotes; as a result, organisms not showing RNAi probably lost it during evolution (Cerutti and CasasMollano, 2006). The use of RNAi has been largely and successfully used for the analysis of single gene silencing as well as for the screening of genome-wide collections of genes, showing that the resulting phenotype is usually highly specific and penetrant (Mohr et al., 2010). This approach allowed achieving extremely important results, especially in the investigation of basic cell life phenomena such as metabolism, mitosis and cytokinesis, chromosome structure and behavior, and mitotic spindle functions. Remarkably, in most screenings RNAi is not used to identify genes required to maintain mitotic chromosome integrity, principally because this type of analysis cannot be automated (Conrad and Gerlich, 2010). Thus, genomewide screenings for this phenotype are largely missing from the scientific literature. The DNA of living cells is subject to many types of molecular lesions, including base modifications, singleand double-strand breaks, and intra- and interstrand crosslinks between bases. Double-stranded DNA breaks (DSBs)
CEPRANI et al. / Turk J Biol
are probably the most deleterious lesions, as they can result in chromosome aberrations (CAs), cell death, and neoplastic transformation (Khanna and Jackson, 2001; van Gent et al., 2001; Mills et al., 2003). To counteract the effects of DSBs, living organisms have evolved 2 main mechanisms for repairing their DNA: homologous recombination (HR) (San Filippo et al., 2008) and nonhomologous end joining (NHEJ) (Lieber, 2010). In the HR pathway the broken ends undergo a recombinational process with the undamaged sequence of either the sister chromatid or the homologous chromosome, which is used as a template for accurate DSB repairs. In the NHEJ pathway, after a limited degradation (Huertas, 2010), broken ends are ligated irrespective of homology, thus resulting in a small sequence deletion at the joining site. The latter mechanism is intrinsically errorprone, and in addition to the aforesaid deletion it may also lead to various types of chromosomal rearrangements, such as transpositions and reciprocal translocations (Lieber et al., 2006; Weinstock et al., 2006). Experimental analyses using ionizing radiation and restriction enzymes have shown that DSBs are the principal lesions leading to the formation of CAs (Natarajan and Obe, 1978; Obe et al., 1992; Vamvakas et al., 1997; Richardson and Jasin, 2000; Obe et al., 2002; Tsai and Lieber, 2010). Interestingly, after a first burst, a second round of radiation-derived CAs may appear several cell generations after the first genomic insult, indicating that its effects on genome stability might become evident even after a long time (Streffer, 2010). Because of the lack of homology, the erroneous ligation of 2 centromere-containing broken DNA ends by NHEJ leads to the formation of a dicentric chromosome able to start the break-fusion-bridge cycle (McClintock, 1951), which in turn makes CAs more complex. CAs were associated with neoplastic transformation in man a long time ago (Nowell and Hungerford, 1960; Levan, 1967; Rowley, 1973; Zech et al., 1976; Fukuhara et al., 1979; Hatano et al., 1981; Tsujimoto et al., 1984; Finger et al., 1986; Rabbitts et al., 1988; Le Beau et al., 1993), frequently because of up-, down- or misregulation of genes important for DNA replication, DNA repair, checkpoint control, and mature ribonucleoprotein (mRNP) biogenesis (Aguilera and Gomez-Gonzalez, 2008; Clémenson and Marsolier-Kergoat, 2009; Kerzendorfer and O’Driscoll, 2009; Mitelman et al., 2010). Acentric fragments, lacking a centromeric region, are not able to correctly segregate during the cell division and can either be inherited by any of the daughter cells – irrespective of their gene content – or be lost. In most cases, the final output is the formation of a cell having a quantity of DNA that is larger or smaller than the normal complement; this situation is usually called segmental aneuploidy because only a fraction of one or a few chromosome(s) is genetically unbalanced. Interestingly, cells cope better
with polyploidy than aneuploidy; although chromosome number abnormalities are frequently associated with the neoplastic transformation (Torres et al., 2008; Williams and Amon, 2009), aneuploidy is a potential tool to target cancer cells, since even transformed cells are sensitive to genomic unbalances (Bannon and Mc Gee, 2009; Williams and Amon, 2009). As a general mechanism, aneuploidy – including segmental aneuploidy – leads to alterations of the gene copy number, and these alterations may influence not only the gene itself (and/or the protein it encodes) but also its molecular or functional interactors (Torres et al., 2008; Veitia et al., 2008; Henrichsen et al., 2009), which leads to the potential deregulation of tens of genes. In this perspective, it becomes crucial to identify those genes that determine the genome stability by controlling the chromosome integrity, since the effects of even a single un- or misrepaired DNA break may have harmful consequences. Of great help is the fact that these genes are widely conserved in most eukaryotes, so it is conceivable that studying them in a model system, such as cultured Drosophila S2 cells interfered with by RNAi, might provide important suggestions about the role of their human orthologs, with the advantage that in Drosophila the interference is easily achieved and the karyotype is simpler than in humans due to a reduced chromosome number. In 2008 Somma et al. identified a number of genes having a role in mitosis. To achieve this result, they created a list of genes coexpressed with other, known mitotic genes. In order to evaluate the coexpression, they used the Pearson correlation coefficient and, since this variable goes between –1.00 and + 1.00 and its highest values express increasing levels of positive correlation, they focused on the range of [0.85, 1.00]. Specifically, Somma et al. evaluated the frequency of contemporary expression of a reference gene (first variable) against any other gene (second variable) of the D. melanogaster genome using microarray data from 89 different experiments. They repeated this analysis 6 times, using 6 reference genes connected to mitosis, and created a merged table of coexpression using the average value of each gene against the 6 reference genes. Finally, they analyzed the first approximately 1000 genes that had a Pearson correlation coefficient in the above mentioned range. The screening was performed by studying the cytological phenotype of cultured S2 cells after RNAi targeting of these genes. In the present work we applied RNAi against the same genes, but specifically focused on the presence of chromosome instability as a consequence of gene silencing. Our approach allowed the identification of a group of at least 81 genes whose silencing results in a chromosome breakage phenotype. Surprisingly, only less than one-third of them can be directly related to DNA metabolism, i.e. DNA replication and/or repair, nucleotide biosynthesis, and chromosome structure. This indicates
881
CEPRANI et al. / Turk J Biol
that the majority of the loci we identified could not be predicted to produce such a phenotype upon silencing. We were able to group some of them into discrete classes, according to their function, and to identify a new phenotype, which we called multifragmented cells, showing extensive chromosome breakage upon RNAi-mediated silencing, and that is particularly strong for genes that have DNA replication-related functions. Unexpectedly, another class of genes was also highlighted by our experiments; they are apparently involved in the mitosis/apoptosis cell fate, but also destabilize the chromosome structure after silencing. Taken together, our results suggest that the genes involved in genome integrity are much more numerous than expected. 2. Materials and methods 2.1. Double-stranded RNA synthesis and cell treatment Genes were amplified by polymerase chain reaction (PCR) either using a mixed embryonic cDNA library (Brown and Kafatos, 1988) or genomic DNA. Each gene-specific primer used contained the 35-nt sequence for the T7 RNA polymerase binding site (5’-TAATACGACTCACTATAGGGAGG-3’) at the 5’ end. dsRNA was synthesized with an average length of 750 bp (minimum length: 600 bp) and analyzed as previously described (Somma et al., 2002, 2008). S2 cells were cultured at 25 °C in Shields and Sang M3 medium (Sigma) supplemented with 10% heat-inactivated fetal bovine serum (FBS, Invitrogen). Cells reared at 25 °C in complete medium were centrifuged, then resuspended at a concentration of 106 cells/mL with serum-free medium and finally plated in a 6-well culture dish (Sarstedt). Each culture was subsequently inoculated with 15 µg of dsRNA. Control cultures were prepared in the same way, in parallel, but no dsRNA was added. After 1 h of incubation at 25 °C, 2 mL of medium supplemented with 15% FBS was added to each culture. Both dsRNA-treated and control cells were grown for 72 h at 25 °C. 2.2. Cytological preparations and image collection Cells from 3-mL cultures were interfered with for 3 days and then resuspended in the medium, and 1 mL of cell suspension was treated for 2 h with colchicine (colcemid at a final concentration of 10–5 M). Cells were then centrifuged at 1000 rpm for 5 min. Pelleted cells were washed in 10 mL of phosphate-buffered saline, spun down by centrifugation, and resuspended in 5 mL of hypotonic solution (0.5 M sodium-citrate) for 7 min. After further centrifugation at the same speed, pelleted cells were fixed in 5 mL of methanol and acetic acid (3:1), spun down again, and resuspended in the small volume of fixative left after the gentle removal of the supernatant. Ten microliters of this suspension was dropped onto a microscope slide and air-dried. All slides were mounted in Vectashield
882
with DAPI (Vector) to stain DNA. Images were captured using a CoolSnap CCD camera (Photometrics, Tucson, AZ, USA) connected to a Nikon Eclipse E600 fluorescence microscope equipped with an HBO 50-W mercury lamp. 2.3. Statistical methods For the proportion of cells with aberrations, the comparison between each line with interference and the control entails the analysis of a 2 × 2 contingency table. The chi-square value with Yates’ correction for continuity and its associated probability were calculated. When the expected number in 1 or more of the table cells was less than 10, Fisher’s exact test was carried out and the 2-sided probability values were calculated using the method of small P-values, as described (Agresti, 1992). The expected number of aberrations per cell in every pair of samples compared (the treated line considered and the control) was calculated assuming that the 2 samples were extracted from the same population with a Poisson distribution. The chi-square test (which is based on the difference between observed and expected values) was performed using Yates’ correction. When the expected number of aberrations was less than 10 in at least 1 of the cells of the table, the exact test based on the binomial distribution was performed. Additionally, in this case the 2-sided probability was calculated with the method of small P-values. For genes whose interference caused a significant effect, the possible nonrandom distribution inside the coexpression list was checked by the Mann– Whitney test, while the possible nonrandom location within the chromosome bands was established evaluating the significance of the deviations from a Poisson distribution (variance test). Finally, whenever multiple tests were involved, in order to achieve a protection against errors of the first type, Holm’s sequential Bonferroni procedure was applied to check the statistical significance at the level of a P ≤ 0.05, as described (Holm, 1979). 3. Results 3.1. Preliminary screening Our initial analysis of S2 cells revealed that there is a considerable number of spontaneous CAs, including both chromatid/chromosome breaks and exchanges. This phenotype is not influenced by the use of non-Drosophila random dsRNA. In order to evaluate the absolute number of CAs and their variation, we performed a series of 45 different control experiments, in parallel with RNAi treatments, using cells grown for the same time and with the same medium, but without adding dsRNA. The results of these experiments are listed in Table 1. As shown, controls averaged 21 aberrant cells (ACs) and 23.7 CAs per 100 scored metaphases – the latter number being higher since occasionally 2 CAs are present in the same
CEPRANI et al. / Turk J Biol Table 1. Control S2 cells without dsRNA treatment (total: 45 independent experiments). Each row summarizes the results of a single experiment, performed in parallel with the RNAi; the last row (bold font) shows the total values (columns 1, 2, and 4) and the average values (columns 3 and 5) of the experiments. ACs: Aberrant cells; CAs: chromosome aberrations. Number of metaphases scored per experiment 80 55 100 100 65 56 30 50 50 50 55 40 100 80 56 30 50 50 50 50 40 50 40 20 22 31 20 30 25 30 20 65 100 56 100 20 70 11 20 30 58 40 20 30 20 2165
Number of ACs 17 13 23 23 13 12 5 11 12 11 13 8 23 17 12 5 12 11 11 9 9 11 9 6 3 7 3 4 5 6 4 13 23 5 22 4 17 2 4 5 12 4 5 6 4 454
ACs per 100 scored metaphases 21.3 23.6 23.0 23.0 20.0 21.4 16.7 22.0 24.0 22.0 23.6 20.0 23.0 21.3 21.4 16.7 24.0 22.0 22.0 18.0 22.5 22.0 22.5 30.0 13.6 22.6 15.0 13.3 20.0 20.0 20.0 20.0 23.0 8.9 22.0 20.0 24.3 18.2 20.0 16.7 20.7 10.0 25.0 20.0 20.0 21.0
Total CAs 22 14 26 29 15 14 6 12 12 12 14 8 23 22 14 6 12 12 12 9 10 13 10 6 4 7 3 6 6 7 4 15 23 6 26 5 20 2 4 6 14 6 5 7 4 513
CAs per 100 scored metaphases 27.5 25.5 26.0 29.0 23.1 25.0 20.0 24.0 24.0 24.0 25.5 20.0 23.0 27.5 25.0 20.0 24.0 24.0 24.0 18.0 25.0 26.0 25.0 30.0 18.2 22.6 15.0 20.0 24.0 23.3 20.0 23.1 23.0 10.7 26.0 25.0 28.6 18.2 20.0 20.0 24.1 15.0 25.0 23.3 20.0 23.7
883
CEPRANI et al. / Turk J Biol
metaphase. Analysis of these data using the chi-square test demonstrated that these numbers are very similar among different experiments (P > 0.9999 for both ACs and CAs), so the variability of this parameter is very low. In conclusion, it can be assumed that the basal level of ACs and CAs is a characteristic feature of this cell lineage and may be considered as a constant when evaluating the possible clastogenic effects of RNAi on a particular gene. As a starting point for the search for genes that are important for chromosome integrity, the experiments described by Somma et al. (2008) were replicated to reproduce the same working conditions, i.e. the same interfering RNA obtained with the same primers was used,
and a corresponding amount of dsRNA was added to the cells, which were treated for an equivalent time. In total, the chromosome complement of 1132 interfered genes was cytologically analyzed. The quantitative data were then studied using the chi-square test with Yates’ correction. We found that none of the tested dsRNA was able to lower the number of either ACs or CAs compared to our controls. Instead, genes increasing the number of CAs and/or ACs upon interference, with a P-value of ≤0.05 in the single comparison with the control, are reported in Table 2. For each cell lineage, at least 50 metaphases in at least 2 different experiments were examined. The experiments characterized by less than 50 metaphases are those for
Table 2. List of the genes that impair chromosome integrity upon RNAi-mediated silencing, with 137 genes identified by RNAi and validated using the chi-square test with Yates’ correction. ACs: Aberrant cells; MCs: multifragmented cells; CAs: chromosome aberrations. Columns – 1: Rank of the gene according to the list by Somma et al. (2008); 2: locus identifier according to FlyBase; 3: other names of the gene/locus, according to the available literature; 4: human orthologous gene; 5: function of the gene as reported in FlyBase, Release FB2013_04 (see below for the meaning of each code); 6: total number of scored metaphases; 7: cells having at least one chromosome break/rearrangement; 8: number of cells with multiple, unscorable breaks and already counted in Column 7; 9: the asterisk marks loci positive for ACs after applying the Holm–Bonferroni correction; 10: total number of CAs, excluding those in MCs; 11: the asterisk marks loci positive for CAs after applying the Holm–Bonferroni correction. Rows in bold text highlight genes/loci that are positive only for ACs or for CAs according to chi-square test with Yates’ correction. Codes for the genetic functions in alphabetical order – A: General RNA metabolism; C: chromatin/chromosome structure; D: DNA damage repair, DNA recombination; E: import-export from nucleus; K: protein phosphorylation and dephosphorylation; M: mitotic spindle assembly; N: nucleotide/nucleoside metabolism; O: other functions; P: phagocytosis, engulfment; R: DNA replication; S: splicing, mRNA maturation/modification; T: transcription; U: ubiquitin and sumo metabolism, protein degradation; X: unknown function; Y: apoptosis; Z: translation. Each functional class was arbitrarily created when at least 3 genes fell inside it; in all other cases, the genes were assigned to the O (other functions) class.
Scored metaphases
ACs
MCs
Holm– Bonferroni correction for ACs
Total CAs
Holm– Bonferroni correction for CAs
2165
454
0
---
513
---
257
111
17
*
163
*
R
100
38
8
47
*
RRM2
NY
165
90
7
*
168
*
BEAF-32
---
CT
120
47
0
*
65
*
4978
mcm7
MCM7
R
174
121
24
*
164
*
20
9273
rpA2
RPA2
R
103
31
1
35
41
3178
rrp1
APEX1
D
142
47
2
62
46
8142
---
RFC4
O
100
23
0
35
84
7055
dalao
SMARCE1
CT
113
35
0
35
104
1966
acf1
BAZ1A
C
181
54
1
72
112
3642
clipper
CPSF4L
A
20
9
0
10
120
11979
rpb5
POLR2E
T
83
31
0
45
*
126
2925
noisette
SF3A3
EM
183
84
3
126
*
135
14999
rfC4
RFC4
CDR
142
51
0
79
*
142
9241
mcm10
MCM10
CR
104
35
0
151
12359
ulp1
SENP1
U
170
88
0
Rank
Locus identifier (CG)
Aliases
Human orthologous gene
Function
0
control
---
4
9193
mus209
PCNA
MR
5
15220
RPA3
---
11
8975
rnrS
12
10159
18
884
*
*
*
40 *
101
*
CEPRANI et al. / Turk J Biol Table 2. (Continued). 152
4206
mcm3
MCM3
R
71
24
2
30
159
5553
DNApolα60
---
R
110
36
0
161
4654
TFDP1
TY
322
110
2
164
8749
SNRNP70
MS
100
28
0
168
12050
dp snRNP-U170K ---
WDR75
X
213
116
2
183
12113
intS4
INTS4
A
36
15
0
202
5452
dnk
ENS400000260851 N
103
49
0
*
80
*
219
8068
su(var)2-10
PIAS1
C
250
84
2
*
143
*
231
13427
---
---
X
126
55
0
*
80
*
245
2848
trn-SR
TNPO3
ES
156
54
1
68
*
264
7769
piccolo
DDB1
D
201
72
5
99
*
271
11920
---
NDOR1
X
80
26
1
31
276
18528
---
GTPBP3
A
150
39
0
54
277
9633
rpA-70
RPA1
MR
70
63
47
*
294
10354
rat1
XRN1
O
274
99
3
*
133
*
300
31671
tho2
THOC2
AE
91
31
0
44
*
306
5198
holn1
CD2BP2
P
44
16
0
21
307
7833
orc5
ORC5L
CMR
150
45
0
70
*
312
6349
DNApolα180 POLA1
R
159
76
1
*
106
*
337
17938
midway
DGAT1
Y
125
54
0
*
58
*
353
14749
GLE1
GLE1
AE
151
64
1
*
78
*
386
17383
jigr1
---
T
237
80
0
*
101
*
387
6011
prp18
PRPF18
S
43
16
0
20
391
10667
orc1
ORC1
R
91
33
1
36
395
3736
okra
RAD54B
D
152
68
2
*
104
*
406
11906
---
---
X
151
63
0
*
76
*
407
7989
wicked
UTP18
A
191
81
0
*
94
*
414
6249
csl4
EXOSC1
A
100
29
0
419
2260
---
WDR46
X
130
52
1
420
6121
tip60
KAT5
CD
100
36
422
2199
---
---
X
200
426
33095
---
---
X
105
427
8274
megator
TPR
M
429
5965
woc
---
430
7993
---
445
8711
454
9348
456
47 *
128
*
35 *
141
*
20
*
54
*
38 *
63
*
0
45
*
54
1
79
*
28
0
36
86
43
0
*
79
*
COT
156
61
0
*
80
*
RPF2
X
100
31
0
55
*
cullin-4
CUL4A
EU
200
108
8
*
204
*
taf6
TAF6
T
198
79
0
*
106
*
8962
paf-AHα
PAFAH1B3
O
150
49
0
462
9677
int6
EIF3E
PZ
280
149
4
*
175
*
474
5370
dcp-1
CASP2
UY
117
51
1
*
67
*
490
8989
his3.3b
HIST1H3B
C
188
90
2
*
125
*
57
885
CEPRANI et al. / Turk J Biol Table 2. (Continued). 493
1078
fip1
FIP1L1
A
150
58
0
*
67
*
495
2161
regena
CNOT2
T
121
60
0
*
86
*
502
7014
rpS5b
RPS5
Z
80
28
1
39
*
516
8878
---
---
K
171
71
1
91
*
517
4281
---
---
X
188
63
0
79
*
519
4788
---
---
X
150
43
0
54
520
13329
cid
---
M
50
20
1
22
523
6340
---
RSRC2
X
85
26
0
32
529
9900
mit(1)15
ZW10
O
10
6
0
11
539
10754
---
3F3A2
M
102
45
1
66
*
555
6197
---
XAB2
PS
70
25
1
40
*
561
2213
msd5
---
M
139
40
0
48
562
8426
l(2)NC136
CNOT3
T
106
35
6
40
585
5649
kin17
KIN
S
150
39
0
56
595
17161
grapes
CHEK1
DKM
61
24
1
28
596
3181
ts
TYMS
N
135
62
1
605
31111
---
TBRG1
X
233
69
0
*
*
*
98
*
*
78
607
11266
caper
RBM39
S
251
72
0
611
5193
TfIIB
GTF2B
T
140
51
1
613
6693
---
DNAJC9
X
90
27
0
35
622
8950
---
GTF3C3
X
151
45
0
48
627
1017
Mfap1
MFAP1
OS
35
10
0
15
632
6453
---
PRKCSH
X
225
65
1
84
652
8980
niPp1
PPP1R8
K
50
21
0
27
*
656
10689
l(2)37Cb
DHX8
A
50
25
1
39
*
661
9473
MED6
MED6
T
75
24
0
672
8243
---
SMAP1
O
268
110
0
674
32708
---
ABT1
X
85
20
0
32
677
13849
nop56
NOP58
X
50
20
0
24
686
1939
dpck
DCAKD
O
160
67
0
*
85
*
689
18041
dgt1
KANSL2
X
50
24
0
*
38
*
693
17446
cfp1
CXXC1
C
99
39
0
*
50
*
708
5786
peter pan
PPAN
X
180
89
0
*
107
*
715
10542
bre1
RNF20
CP
100
31
0
40
725
10903
---
WBSCR22
X
118
37
0
51
729
5933
ime4
METTL3
A
210
81
0
734
5149
---
TLDC1
X
80
25
0
743
12135
c12.1
CWC15
M
146
80
1
758
9601
---
PNKP
X
200
62
0
760
6946
glorund
---
CZ
252
113
0
*
129
*
763
6480
frg1
FRG1
X
200
103
0
*
110
*
886
*
*
100
*
63
*
32 *
*
139
91
*
*
32 *
123
*
73
CEPRANI et al. / Turk J Biol Table 2. (Continued). 768
6937
---
NIFK
X
245
91
0
*
129
*
772
7757
prp3
PRPF3
S
180
80
0
*
133
*
788
10042
MBD-R2
---
T
201
95
6
*
205
*
807
3351
mRpL11
MRPL11
Z
138
39
0
808
2685
---
WBP11
S
150
59
0
810
15019
---
---
X
121
36
0
45
811
12249
miranda
---
O
93
32
1
36
823
4916
me31B
DDX6
A
175
60
0
824
1676
cactin
CACTIN
X
301
107
0
829
4326
mRpS17
---
Z
86
24
0
36
843
6686
---
SART1
X
90
28
0
28
850
5408
tribbles
TRIB1
K
189
59
0
68
853
6057
SMC1
SMC1A
C
180
63
1
860
13625
---
BUD13
S
82
29
862
2177
zip103B
SLC39A9
O
150
867
10139
---
---
X
878
7003
msh6
MSH6
882
32066
CG6487
883
5102
885
51 *
73
*
66 *
*
77
*
1
44
*
48
0
62
*
160
46
0
50
D
138
50
0
*
62
*
FAM49A
X
159
73
0
*
107
*
daughterless
---
DT
265
82
0
99
*
5923
DNApolα73
POLA2
R
164
50
0
56
893
4449
---
---
X
300
107
4
*
139
*
900
4785
intS14
VWA9
X
80
35
1
*
58
*
927
17528
---
DCX
K
110
41
0
50
*
930
3735
---
DIEXF
X
110
29
0
48
932
4173
septin-2
SEPT11
O
109
34
0
43
949
1710
hcf
HCFC1
CT
130
46
0
51
967
17068
---
---
X
126
41
0
59
*
978
6854
CTP synthase CTPS1
N
120
54
0
91
*
983
6977
cadherin 87a
---
O
140
40
0
53
986
7656
---
UBE2R2
U
76
20
2
28
994
4866
---
IMP3
Z
102
36
1
42
995
7081
pex2
PEX2
O
170
42
0
59
998
8169
pms2
PMS2
D
125
40
0
58
*
1002
18005
beag
IK
O
122
49
1
56
*
1021
6744
---
EXD2
D
157
42
0
1023
2050
modulo
---
X
234
86
0
1057
18190
---
MAPRE1
O
152
48
0
61
1061
5454
snRNP-U1-C SNRPC
MS
218
65
1
79
1073
9680
dbp73D
A
176
52
0
62
DDX51
*
121
*
*
52 *
110
*
887
CEPRANI et al. / Turk J Biol
which the interfered cells show a mitotic index much lower than that of the control; in this case, RNAi was performed at least 3 times (and up to 4 times) and, if metaphases were still not enough, data were also analyzed using Fisher’s exact test (for ACs) and binomial distribution (for CAs) to confirm the results of the chi-square test. This allowed the identification of 137 genes (i.e. more than 12% of the 1132 scored) in which the number of ACs, the number of CAs, or both are significantly higher than in the control (P ≤ 0.05 in the single test). It is noteworthy that more than 80% of these genes (112 / 137) have a human counterpart (Table 2). Of great interest is the discovery that 34 of them (~25%), to date, have a completely unknown function, and 66 (~48%) have a function not directly related to DNA metabolism, since their annotation in FlyBase, the fruit fly database, does not report (as to Release 2013_04) for them a role in any of the following groups: i) chromatin/ chromosome structure, ii) DNA damage repair, iii) DNA recombination, iv) DNA replication, v) nucleotide/ nucleoside metabolism (Table 2). Consequently, only 27% (37/137) of the genes in the list could have been predicted, on the basis of their established function or of their homology with other known genes, as capable of altering
the karyotype structure. In conclusion, the number of genes important to maintain the stability of the genome is much greater than expected. For the majority of the identified genes (117/137, 85.4%), both the number of ACs and the number of CAs per 100 metaphases is significantly increased, compared to control, using the chi-square test. However, for 20 genes (14.6%), this is not true (Table 2, see the genes highlighted in bold font). Of them, 16 (11.7%) showed a significant increase only in CA absolute number, and 4 (2.9%) only in the number of ACs. An analysis of the function of these 20 genes indicates that also in this case, many (8/20, 40%) are of unknown function, and only a few of them (2/20, 10%) are directly related to DNA metabolism. Thus, also in these cases it would have been impossible to guess their role in genome stability without some types of experimental evidence. Finally, we also identified a phenotype that is not present in control cells and that we called multifragmented cells (MCs), i.e. cells showing such a high number of CAs that it does not allow the discerning of single chromosomes. This phenotype is easily recognizable: an average karyotype of S2 cells contains 12 chromosomes (Figure 1a), while
Figure 1. Cytology of S2 cells. A) Untreated control cell; note the chromosome complement, composed of 12 elements. B) An aberrant cell (AC) showing 1 chromosome involved in 2 rearrangements (CAs) at the same time, an X-type symmetric exchange (arrow) and a U-type asymmetric exchange (arrowhead). C) A multifragmented cell (MC); each asterisk marks a chromosome aberration (CA), either a single or a double chromatid break. Metaphases in panels B and C come from mus209-interfered cells. D, E, F) MC cells from rpA70interfered cells; note the increasing level of chromosome fragmentation.
888
CEPRANI et al. / Turk J Biol
in these cells the number of chromosomal structures, including acentric fragments, is far higher that 20 (our arbitrary threshold to separate ACs and MCs). Notably, in control cells this phenotype was never scored in more than 2000 metaphases. Instead, during the screening it was possible to identify 49/137 genes (35.8%) positive for this phenotype, up to a MC/AC ratio value of ~75% (47/63) as in the case of the RpA-70 coding gene (Tables 2 and 3; Figure 1). Notably, in previous reports this phenotype was identified only in RNAi experiments targeting RpA-70.
3.2. Identification of new genes involved in chromosome integrity In the previous paragraph, the results of single comparisons between each interfered line and the control were shown. However, the analysis of hundreds of tests may cause the emergence of several false positives by chance. In order to obtain protection against an inflated error of the first type, and to maintain a P-value of ≤0.05 as our level of statistical significance, Holm’s sequential Bonferroni procedure for multiple testing was used. Genes significant at the 0.05
Table 3. Genes showing the MC phenotype after RNAi-mediated silencing. Genes are ranked according to their descending MC/AC ratio. Column 1: Gene locus; Column 2: gene aliases (FlyBase, Release 2013_04); Column 3: functions (they are the same as in Table 2); Column 4: MC/AC ratio; Column 5: Holm–Bonferroni correction positivity. Locus (CG)
Aliases
Function
MC/AC × 1000
Positive after Holm–Bonferroni correction
9633
rpA-70
MR
746
yes
15220
---
R
211
yes
4978
mcm7
R
198
yes
8426
l(2)NC136
T
171
no
9193
mus209
MR
153
yes
7656
---
U
100
no
4206
mcm3
R
83
no
8975
rnrS
NY
77
yes
8711
cullin-4
EU
74
yes
7769
piccolo
D
69
yes
10042
MBD-R2
T
63
yes
13329
cid
M
50
no
3178
rrp1
D
43
yes
17161
grapes
DKM
42
no
6197
---
PS
40
yes
10689
l(2)37Cb
A
40
yes
11920
---
X
38
no
4449
---
X
37
yes
2925
noisette
EM
36
yes
7014
rpS5b
Z
36
yes
13625
---
S
34
yes
9273
rpA2
R
32
no
12249
miranda
O
31
no
10354
rat1
O
30
yes
10667
orc1
R
30
no
3736
okra
D
29
yes
4785
---
X
29
yes
889
CEPRANI et al. / Turk J Biol Table 3. (Continued). 4866
---
Z
28
no
9677
int6
PZ
27
yes
8068
su(var)2-10
C
24
yes
8989
his3.3b
C
22
yes
10754
---
M
22
yes
5370
dcp-1
UY
20
yes
5193
TfIIB
T
20
yes
18005
beag
O
20
yes
1966
acf1
C
19
yes
2848
trn-SR
ES
19
yes
2260
---
X
19
yes
2199
---
X
19
yes
4654
dp
TY
18
yes
12050
---
X
17
yes
14749
GLE1
AE
16
yes
3181
Ts
N
16
yes
6057
SMC1
C
16
yes
6453
---
X
15
no
5454
snRNP-U1-C
MS
15
no
8878
---
K
14
yes
6349
DNApola180
R
13
yes
12135
c12.1
M
13
yes
level after this correction are indicated by an asterisk in Table 2. This allowed the identification of at least 81 genes with the same characteristics, i.e. either CAs only or both ACs and CAs are significantly higher than in the control after RNAi. Their relative proportions as for the functions resulted similarly to those previously described, and also in this shorter list more than 80% of identified genes has a human ortholog. Interestingly, and differently from what was found with the single comparisons described before, with this statistical method none of the identified loci show a significant increase of ACs only. Consequently, at the moment it is not possible to assess whether the increase of ACs only is a true phenotype for some silenced genes, or if it is just a byproduct of randomness. 3.3. Genes causing AC-CA are randomly distributed as for ranking and map position To find out whether there is any correlation among the 81 genes identified in the present screening, 3 different variables were analyzed separately using statistical approaches.
890
The first variable is their ranking position inside the original list created by Somma et al. (2008) (reported in Table 2, Column 1). Their original work was based on the coexpression of genes having a role in mitosis, with the rationale that genes involved in the same biological process tend to be transcriptionally coexpressed. The ranking position of any given gene (Somma et al., 2008) consequently reflects its average coexpression value with respect to 6 reference genes. To get a general picture of the present data, the 1132 selected genes were arbitrarily split into 11 blocks of 103 genes (with the exception of the last block, which contains 102 genes) and the number of genes positive for ACs and/or CAs after RNAi treatment was evaluated. As shown in Figure 2, the genes seem to be almost uniformly distributed and there is no evidence of a tendency to concentrate in any portion of the coexpression list. A comparison between the distribution of the 81 loci and that of the remaining 1051 genes using the Mann– Whitney test confirmed the above intuitive conclusion (P = 0.1168). Therefore, genes causing genome instability
CEPRANI et al. / Turk J Biol 16
14
Number of genes
14 12 10 8 6
8
8
6
8
7
9
9
7 5
4 2 0
0 001103
104206
207309
310412
413- 516- 619515 618 721 Rank position
722824
825927
9281030
10311132
Figure 2. Distribution of the genes causing chromosome breakage upon RNAi. The genes analyzed in the present screening were split into 11 classes of 103 genes each (with the exception of the last class, which has 102 genes) according to their rank position. For each class, the numbers of genes positive after applying the Holm–Bonferroni correction are reported. Note that the identified genes are almost uniformly distributed as for their rank position (see text for the statistical analysis).
upon silencing are randomly and uniformly distributed throughout the original list. The second variable considered is the mapping position of the identified genes inside the D. melanogaster genome, to investigate if they share their chromosomal location, creating syntenic groups (Figure 3). For this purpose the available data from FlyBase about the 102 chromosomal bands that characterize the polytene chromosomes of the fruit fly were collected. In particular, the focus was on the number of protein-coding genes, which are the target of RNAi experiments. In order to check whether the 81 significant genes tend to concentrate on particular bands, an analysis was performed on the possible presence of departures from a Poisson distribution. The variance test
did not show any significant deviation (the 1-sided p-value equals 0.4648). Similarly, the number of positive genes on any given chromosome arm was proportional to the overall number of protein-coding genes mapped on that arm (Table 4), the single gene on the fourth chromosome being excluded in computing the p-value; the chi-square test did not show any significant deviation from the expected numbers (P = 0.5487). In conclusion, the 81 genes causing genome instability upon silencing are not clustered inside the genome of D. melanogaster. 3.4. Identification of gene clusters inside the list according to their cellular function The third variable evaluated is the cellular function of the 81 genes identified in the present screening, to check
Figure 3. Genes required for the chromosome integrity are not clustered inside the D. melanogaster genome. Asterisks mark the approximate cytogenetic position of the 81 genes positive after applying the Holm–Bonferroni correction for multiple tests; note the almost uniform distribution of the genes along the polytene chromosomes of the fruit fly.
891
CEPRANI et al. / Turk J Biol Table 4. Map distribution – chromosome arms (euchromatin only). Data according to FlyBase, Release FB2010_06. Chromosome arm
Length (kb)
Number of genes (all)
Protein-coding genes
Genes positive for AC-CA
X
22775
2341
2188
12
2L
22366
2899
2626
14
2R
20973
3299
2758
19
3L
23550
2960
2719
11
3R
28842
3701
3395
24
4
1339
84
84
1
Total
119,845
15,284
13,770
81
whether it is somehow related to their rank position. For this aim, the function reported in FlyBase (Release FB2013_04) was used as a reference. Sixteen arbitrary functional classes were created, after setting the rule that a class must be represented by at least 3 genes/loci; in all other cases, the genes/loci were assigned to the class “other functions” (Table 2). These functional classes were then further grouped into 5 macroclasses, which can be related to more general metabolic and/or cellular mechanisms. The 5 macroclasses are: i) DNA metabolismrelated functions; ii) RNA metabolism-related functions; iii) subcellular functions; iv) cell cycle progression-related functions; v) unknown functions (Figure 4). Interestingly, the Mann–Whitney test revealed a significant deviation from randomness for the distribution of the positive genes belonging to 2 of the macroclasses: the DNA metabolismrelated genes (P = 0.0070) and the cell-cycle progressionrelated genes (P = 0.0088). This means that these genes tend to be clustered toward the higher portion of the list. Notably, data of these 2 classes of genes are significant at the 0.05 level also after the Holm–Bonferroni correction for multiple tests. 3.5. Genes causing high frequencies of MCs upon silencing mainly have functions related to DNA replication and tend to have high rankings As described before, genes causing extensive chromosome fragmentation (MCs) upon RNAi-mediated silencing are well represented among those causing CAs and ACs – indeed, almost half of the genes (37/81, 45.7%) show this peculiar phenotype. Genes positive for the MC phenotype are reported in Table 3, which includes also those genes that are not positive after applying the Holm–Bonferroni method (in the present section, we will refer to the entire content of Table 3 as an “extended list”). Notably, 4 of the 8 genes positive after applying this statistical correction and with a stronger phenotype (i.e. with a ratio of MC/
892
AC of ≥5%), namely rpA-70 (whose ratio MC/AC equals 75%) CG15220 (MC/AC = 21%), mcm7 (MC/AC = 20%), and mus209 (MC/AC = 15%), are involved in the DNA replication. Additionally, these 4 genes are also the top 4 as for the strength of their MC phenotype. Looking at the extended list, 3 more loci fall inside the MC/AC ≥5% group, i.e. l(2)NC136 (MC/AC = 17%), CG7656 (MC/AC = 10%), and mcm3 (MC/AC = 8%). Of these, mcm3 (position 7 in the extended list) is involved in DNA replication as well. Taken together, these data suggest that an extreme MC/ AC ratio phenotype could be typical of genes involved in DNA replication-related processes. Interestingly, among the genes inducing the appearance of MCs after silencing, the number of genes involved in the mitotic spindle assembly is also relatively high (5 loci positive with the Holm–Bonferroni method out of 8 total loci), although for them there is no particular accumulation in any part of the extended list. To characterize further this class of genes, their rank position inside the list reported in Table 2 was considered, in order to check if their distribution is random (Figure 5). An analysis with the Mann–Whitney test revealed that these genes are not randomly distributed (P = 0.0243). These data indicate that genes causing extensive chromosome damage tend to be coexpressed with the mitotic genes originally described by Somma et al. in 2008. 4. Discussion 4.1. Overview of the screening In the present work, we analyzed the karyotype of S2 cells treated with dsRNA against a chosen pool of genes. These genes had previously been selected based on their coexpression with other genes involved in various aspects of mitosis (Somma et al., 2008). The cited study and the data shown here demonstrate that this methodology allows the identification of genes that play a role in chromosome
CEPRANI et al. / Turk J Biol 6
6 CDNR
5
4
4
3
3
3
3
2
2
2 1
1 0
Number of genes
Number of genes
6
1
1
3 2 1 0
2
Number of genes
Number of genes
3
AST
4 3
2
2
1
1
3
0
2
2 1
1
1
0
0
001- 104- 207- 310- 413- 516- 619- 722- 825- 928- 1031103 206 309 412 515 618 721 824 927 1030 1132 Rank position
5
MY
4 3
2
2
2
2
2 1
1
1 0
0
001- 104- 207- 310- 413- 516- 619- 722- 825- 928- 1031103 206 309 412 515 618 721 824 927 1030 1132 Rank position
0
0
0
001- 104- 207- 310- 413- 516- 619- 722- 825- 928- 1031103 206 309 412 515 618 721 824 927 1030 1132 Rank position
6 Number of genes
3
2
1 0
EKOPUZ
6
6 4
3
3
0
001- 104- 207- 310- 413- 516- 619- 722- 825- 928- 1031103 206 309 412 515 618 721 824 927 1030 1132 Rank position
4
4
1 0
5
5
5
X
5 4
3
3
3
3
2
2
1
1
1
1
2
1 0
0
0
001- 104- 207- 310- 413- 516- 619- 722- 825- 928- 1031103 206 309 412 515 618 721 824 927 1030 1132 Rank position
Figure 4. Function distribution. The functions identified by the 81 genes selected with the Holm–Bonferroni method (Table 2) were grouped into 5 macroclasses. Genes/loci were split into the same 11 blocks reported in Figure 2. There is not a one-to-one relationship between genes and functions: a gene having 2–3 functions in the same macroclass was scored once, while a gene having 2–3 functions in different macroclasses was scored once for each macroclass; thus, the number of functions analyzed here (98) does not correspond to the number of genes selected with the Holm–Bonferroni method (81). CDNR: DNA metabolism-related functions (for the letter meaning, see Table 2); AST: RNA metabolism-related functions; EKOPUZ: other cellular functions; MY: cell cycle progression-related functions; X: unknown functions. Note that the CDNR and MY functions are not evenly distributed along the list of coexpression (see text for the statistical analysis). 8
7
Number of genes
7 6 5
5
5
5
4
4
4 3
3
2
2 0
1
1
1
0 001103
104206
207309
310412
413- 516- 619515 618 721 Rank position
722824
825927
9281030
10311132
Figure 5. Distribution of the genes causing extreme chromosome fragmentation (MCs) upon RNAi. Note that these loci tend to accumulate toward the top half of the list described by Somma et al. (2008) (see text for the statistical analysis).
893
CEPRANI et al. / Turk J Biol
integrity. S2 cells show a relatively high, yet constant, level of spontaneous CAs (Table 1); this fact, and the relatively simple chromosome complement of this cell lineage, makes S2 cultured cells a suitable model for the karyotype analysis and for the identification of genes whose function is related to the maintenance of chromosome integrity. In their original analysis, Somma et al. found a lower number of genes causing chromosomal damage upon RNAi treatment. This difference may be explained by the different statistical approach we used. In their report, positive genes were identified by applying a high threshold to the total AC/CA values, to compensate for lower cell scores. Our increased number of observations, obtained by repeating the interference up to 4 times per gene, allowed us to perform a double, sequential statistical analysis, which permitted the identification of at least 81 genes positive for this phenotype. We are aware that, apparently, there is a main limitation in this screening, due to the method used to shrink the number of examined genes. The reference list (Somma et al., 2008) relies on the average coexpression value (evaluated with Pearson’s correlation coefficient) of D. melanogaster genes with other 6 mitosis-related genes. The choice of 6 reference genes is arbitrary; any other list derived averaging 6 (or even more) different lists could have been used, and at the moment it is not known if this would have created a significantly different record of coexpressed genes. Moreover, our screening was performed for a phenotype (chromosome integrity) that was not straightly related to their gene selection method. For these reasons, it is not surprising that genes directly involved in the DNA damage recognition/repair, or in general DNA metabolism (nucleotide biosynthesis, DNA replication, chromatin/chromosome structure), are poorly represented among those identified – they are approximately 32% of the total after the Holm–Bonferroni correction. Instead, a lot of found genes either have unknown function (~21%) or are assigned to functions not apparently related to chromosome integrity (~47%). Given that segmental aneuploidy is potentially harmful for cell survival and is a recurrent step during tumorigenesis, all genes causing such a phenotype are potentially “indirect” oncogenes, since their clastogenic action is able to alter the genetic equilibrium of the cell. Consequently, the possibility to have a list of such genes is crucial for the comprehension of the neoplastic transformation and for the study of the possible role of these genes in this phenomenon. Moreover, these genes may be used for patient screening and as potential targets of antitumor therapy. In addition, the fact that 2 of the reference genes encode proteins having important roles also outside of mitosis (Cid/CENPA deposition occurs during the S phase and EB1 plays important roles during the interphase) further broadens the record of interesting genes found.
894
Since the coexpression-based selection worked fine for the isolation of new genes involved in mitosis (Somma et al., 2008), and since the cytogenetic methodology works fine as well for the study of genes involved in chromosome structure, merging these 2 approaches (i.e. creating coexpression lists using genes known to be related to chromosome integrity) should allow the identification of novel genes whose cellular role is still unknown. At this point, the goal is not trying to understand why a given gene causes a certain phenotype, but rather to identify how many and which genes are responsible, directly or indirectly, for this phenotype. For example, there are no clues about why 7/81 (8.6%) genes have a described function connected to mitotic spindle assembly. However, knowing that these 7 genes may also cause chromosome instability is certainly an important discovery and indicates that, potentially, other genes with the same function might induce a comparable phenotype on chromosomes; this in turn indicates that genes responsible for mitotic spindle assembly are potentially able to cause neoplastic transformation by induction of segmental aneuploidy. Similar conclusions may be drawn for the other classes of genes described. The reason why these genes cause the CA phenotype is, in many cases, unknown. Several scenarios may be depicted to explain these data, and probably more than one should be applied to the genes here described. The easiest explanation is that these genes encode proteins that play multiple roles inside the cell, one of which is maintaining the chromosome integrity – a function not described before and that partly addresses the so-called “g-value paradox” (Hahn and Wray, 2002). This is reasonable, since available analyses of mutant flies did not necessarily cover all aspects of cell biology. Another possibility is that the reported function is simply wrong. In some cases no mutants are available for a given locus, and the gene role is inferred only on the basis of sequence similarity with other genes of known function. Finally, it may be also assumed that the gene function is not “directly” related to DNA damage, and that the chromosome breakage occurs only after one or more intermediate metabolic steps that, for some reasons, are not visible in mutants (if available). Indeed, cell cultures are not complete organisms, and RNAi is a very powerful technique: it is possible that these phenotypes are just not visible in mutant flies, either because of the presence of redundant functions, or because cells undergo cell cycle arrest that does not allow the analysis of the karyotype, or even because some genes are not active in certain tissues (the fruit fly karyotype is usually analyzed in larval neuroblasts, and S2 is an undifferentiated, embryo-derived cell lineage). In any case, the conclusion is that the list of genes that potentially may alter the chromosome structure is surely much longer
CEPRANI et al. / Turk J Biol
than anticipated, and one way to compile such a list is, unexpectedly, to search for genes not related to genome integrity. 4.2. Identification of gene clusters A great effort of the present screening was the search for relationships among the identified genes, especially for their relative position inside the coexpression list, for their physical position inside the genome, and for their role inside the cell. The result is that the 81 loci found are almost evenly distributed for both their relative position in the coexpression list, and for their map position inside the D. melanogaster genome. This roughly uniform scattering suggests a functional and evolutionary meaning. As described in Section 1, the loss of a genomic portion is frequently associated with the neoplastic transformation. Consequently, an even localization of the genes causing chromosomal breaks upon malfunction determines the fact that any deletion of any genomic region will impair the entire genome stability. Thus, it is possible to postulate some kind of “positional information” related to these genes that allows the cell to indirectly control the overall stability of the genome. In other words, any random genomic damage – independently of its cause – will frequently induce the associated loss of at least one gene controlling the overall genomic stability; this would further destabilize the genome, inducing cell death in most cases. Further studies are needed to verify if this almost even distribution is typical of Drosophila or is evolutionary conserved; if it is conserved, this would partly explain, among other things, why some genomic rearrangements in cancer cells are more frequent than others. When the positive genes were grouped into functional macroclasses, other interesting results arose. At least for 2 macroclasses, the positive genes tend to be clustered toward the top of the list: those connected to DNA metabolism, and those connected to cell cycle progression. The discovery of the first class of genes was not surprising, but the second class of loci, comprising genes involved in apoptosis and mitotic spindle assembly, was unpredictable. We would expect that impairing mitotic spindle assembly would result in an activation of the spindle checkpoint able to stop the cell cycle; alternatively (if this checkpoint is the interfered function), we would expect the cell to go on dividing, but would also expect that a subsequent interphase checkpoint would induce its cycle arrest. Similar hypotheses could be made for the genes involved in apoptosis. Instead, RNAi on these genes produces a potent effect on the chromosome integrity, and the high level of CAs that we found suggests that their formation is probably irrespective of other cellular control mechanisms. These results may be explained in several ways, and we suggest here 3: i) the silencing of these genes causes an inactivation cascade targeting also other checkpoint
proteins, so the entire cellular control machinery is impaired; ii) these genes play a role in the activation of one or more still-working checkpoints that, if not activated at the proper time, are not able to function anymore; iii) the silencing of these genes destabilizes the chromosome structure immediately before the mitotic chromosome condensation, so checkpoints do not have enough time to be activated prior to the metaphase plate establishment. Further analysis of these genes is required to discriminate among these possibilities, or to postulate others. The present study also allowed the identification of a class of genes causing extensive chromosome breakage after RNAi and recognizable by the presence of MCs, a phenotype absent in controls. Analysis of these loci revealed that many of them, and particularly those showing the stronger phenotype (as measured by the MC/ AC ratio), belong to the DNA replication functional class. DNA replication, strictly speaking, is not a mitosis-related function, but probably these genes were selected because, as described before, 2 of the 6 reference genes also act during interphase. This fact allowed description of what the effect of RNAi is on this type of genes and the drawing of some conclusions. First, many genes related to DNA replication, after silencing, may show a similar phenotype. This may be easily verified by performing RNAi against genes belonging to this functional group. Second, a given gene showing, upon silencing, a high MC/AC ratio, will likely take part in DNA replication. This of course will be useful for genes either with an unrelated or without a known function, and in this case the verification should be based on the cytogenetic analysis of mutants (if available) and/or on biochemical evidences (for example, interaction with known DNA replication proteins or in vitro functional analysis). Ideally, this should provide a role to at least some genes with unknown function but similar phenotype. Third, according to the present data, these genes tend to be coexpressed with each other (accumulation in the top half of the list), so in principle the creation of appropriate coexpression lists using known genes of this functional class should give good results in the search of new DNA replication-related loci. 4.3. Final remarks and future perspectives The aim of the present screening was to find genes causing chromosome integrity failure upon silencing. The results were obtained using, as a selection criterion, coexpression with genes not related to this phenotype. Our results indicate that, at least in some cases, these genes encode proteins with multiple, different roles. Indeed, this is a relatively new and not completely surprising finding: searching for new gene functions implies that the “previous” function should not be considered during the screening. These data in part fill the so-called “g-value paradox” (Hahn and Wray, 2002), according to which
895
CEPRANI et al. / Turk J Biol
there are too many functions compared to the number of protein-coding genes. Our data show that the genes whose function is preventing chromosome breakage are far more numerous than expected, and potentially they might be involved in tumorigenesis through the induction of segmental aneuploidy. Since many genes are apparently unrelated with this phenotype, a screen of the entire D. melanogaster genome would be invaluable for the identification of such loci; in fact, our data strongly suggest that any selection criterion would result in an incomplete record of genes.
Acknowledgments This work was achieved thanks to the generosity and kindness of Emeritus Professor M Gatti (Sapienza University of Rome), who permitted us to use his laboratory facilities and reagents to perform the experiments. We are thus deeply grateful to him and his staff. We also gratefully acknowledge Dr Alessandro Porrello (Lineberger Comprehensive Cancer Center, University of North Carolina-Chapel Hill, Chapel Hill, NC, USA) for providing us with important comments and suggestions and for the careful review of this manuscript.
References Agresti A (1992). A survey of exact inference for contingency tables. Statist Sci 7: 131–153. Aguilera A, Gómez-González B (2008). Genome instability: a mechanistic view of its causes and consequences. Nat Rev Genet 9: 204–217. Aravind L, Watanabe H, Lipman DJ, Koonin EV (2000). Lineagespecific loss and divergence of functionally linked genes in eukaryotes. P Natl Acad Sci USA 97: 11319–11324. Bannon JH, McGee MM (2009). Understanding the role of aneuploidy in tumorigenesis. Biochem Soc Trans 37: 910–913. Brown NH, Kafatos FC (1988). Functional cDNA libraries from Drosophila embryos. J Mol Biol 203: 425–437. Cerutti H, Casas-Mollano J (2006). On the origin and functions of RNA-mediated silencing: from protists to man. Curr Genet 50: 81–99. Clémenson C, Marsolier-Kergoat MC (2009). DNA damage checkpoint inactivation: adaptation and recovery. DNA Repair (Amst) 8: 1101–1109. Conrad C, Gerlich DW (2010). Automated microscopy for highcontent RNAi screening. J Cell Biol 188: 453–461. DaRocha W, Otsu K, Teixeira S, Donelson J (2004). Tests of cytoplasmic RNA interference (RNAi) and construction of a tetracycline-inducible T7 promoter system in Trypanosoma cruzi. Mol Biochem Parasitol 133: 175–186. Drinnenberg IA, Weinberg DE, Xie KT, Nower JP, Wolfe KH, Fink GR, Bartel DP (2009). RNAi in budding yeast. Science 326: 544–550. Ecker JR, Davis RW (1986). Inhibition of gene expression in plant cells by expression of antisense RNA. P Natl Acad Sci USA 83: 5372–5376. Finger LR, Harvey RC, Moore RC, Showe LC, Croce CM (1986). A common mechanism of chromosomal translocation in T and B cell neoplasia. Science 234: 982–985. Fire A, Xu S, Montgomery M, Kostas S, Driver S, Mello CC (1998). Potent and specific genetic interference by double-stranded RNA in Caenorhabditis elegans. Nature 391: 806–811.
896
Fukuhara S, Rowley JD, Variakojis D, Golomb HM (1979). Chromosome abnormalities in poorly differentiated lymphocytic leukemia. Cancer Res 39: 3119–3128. Guo S, Kemphues K (1995). par-1, a gene required for establishing polarity in C. elegans embryos, encodes a putative Ser/Thr kinase that is asymmetrically distributed. Cell 81: 611–620. Hahn MW, Wray GA (2002). The g-value paradox. Evol Dev 4: 73–75. Hatano M, Roberts CW, Minden M, Crist WM, Korsmeyer SJ (1991). Deregulation of a homeobox gene, HOX11, by the t(10;14) in T cell leukemia. Science 253: 79–82. Henrichsen CN, Chaignat E, Reymond A (2009). Copy number variants, diseases and gene expression. Hum Mol Genet 18: R1–R8. Holm S (1979). A simple sequential rejective multiple test procedure. Scand J Statist 6: 65–70. Huertas P (2010). DNA resection in eukaryotes: deciding how to fix the break. Nat Struct Mol Biol 17: 11–16. Kerzendorfer C, O’Driscoll M (2009). Human DNA damage response and repair deficiency syndromes: linking genomic instability and cell cycle checkpoint proficiency. DNA Repair (Amst) 8: 1139–1152. Khanna KK, Jackson SP (2001). DNA double-strand breaks: signaling, repair and the cancer connection. Nat Genet 27: 247–254. Le Beau MM, Larson RA, Bitter MA, Vardiman JW, Golomb HM, Rowley JD (1983). Association of inv(16)(p13q22) with abnormal marrow eosinophils in acute myelomonocytic leukemia: a unique cytogenetic-clinicopathologic association. N Engl J Med 309: 630–636. Levan A (1967). Some current problems of cancer cytogenetics. Hereditas 57: 343–355. Lieber MR (2010). The mechanism of double-strand DNA break repair by the nonhomologous DNA end-joining pathway. Annu Rev Biochem 79: 181–211. Lieber MR, Yu K, Raghavan SC (2006). Roles of nonhomologous DNA end joining, V(D)J recombination, and class switch recombination in chromosomal translocations. DNA Repair (Amst) 5: 1234–1245.
CEPRANI et al. / Turk J Biol McClintock B (1951). Chromosome organization and genic expression. Cold Spring Harb Symp Quant Biol 16: 13–47. Mills KD, Ferguson DO, Alt FW (2003). The role of DNA breaks in genomic instability and tumorigenesis. Immunol Rev 194: 77–95. Mitelman F, Johansson B, Mertens F, editors (2010). Mitelman Database of Chromosome Aberrations and Gene Fusions in Cancer. Bethesda, MD, USA: National Cancer Institute. Available online at http://cgap.nci.nih.gov/Chromosomes/ Mitelman. Mohr S, Bakal C, Perrimon N (2010). Genomic screening with RNAi: results and challenges, Annu Rev Biochem 79: 37–64. Nakayashiki H, Kadotani N, Mayama S (2006). Evolution and diversification of RNA silencing proteins in fungi. J Mol Evol 63: 127–135. Napoli C, Lemieux C, Jorgensen R (1990). Introduction of a chimeric chalcone synthase gene into Petunia results in reversible cosuppression of homologous genes in trans. Plant Cell 2: 279– 289.
Romano N, Macino G (1992). Quelling: transient inactivation of gene expression in Neurospora crassa by transformation with homologous sequences. Mol Microbiol 6: 3343–3353. Rowley JD (1973). Identification of a translocation with quinacrine fluorescence in a patient with acute leukemia. Ann Genet 16: 109–112. San Filippo J, Sung P, Klein H (2008). Mechanism of eukaryotic homologous recombination. Annu Rev Biochem 77: 229–257. Somma MP, Ceprani F, Bucciarelli E, Naim V, De Arcangelis V, Piergentili R, Palena A, Ciapponi L, Giansanti MG, Pellacani C, Petrucci R, Cenci G, Vernì F, Fasulo B, Goldberg ML, Di Cunto F, Gatti M (2008). Identification of Drosophila mitotic genes by combining co-expression analysis and RNA interference. PLoS Genet 4: e1000126. Somma MP, Fasulo B, Cenci G, Cundari E, Gatti M (2002). Molecular dissection of cytokinesis by RNA interference in Drosophila cultured cells. Mol Biol Cell 13: 2448–2460. Streffer C (2010). Strong association between cancer and genomic instability. Radiat Environ Biophys 49: 125–131.
Natarajan AT, Obe G (1978). Molecular mechanisms involved in the production of chromosomal aberrations. I. Utilization of Neurospora endonuclease for the study of aberration production in G2 stage of the cell cycle. Mutat Res 52: 137–149.
Torres EM, Williams BR, Amon A (2008). Aneuploidy: cells losing their balance. Genetics 179: 737–746.
Nowell P, Hungerford D (1960). A minute chromosome in human chronic granulocytic leukemia. Science 132: 1497.
Tsujimoto Y, Finger LR, Yunis J, Nowell PC, Croce CM (1984). Cloning of the chromosome breakpoint of neoplastic B cells with the t(14;18) chromosome translocation. Science 266: 1097–1099.
Obe G, Johannes C, Schulte-Frohlinde D (1992). DNA doublestrand breaks induced by sparsely ionizing radiation and endonucleases as critical lesions for cell death, chromosomal aberrations, mutations and oncogenic transformation. Mutagenesis 7: 3–12. Obe G, Pfeiffer P, Savage JR, Johannes C, Goedecke W, Jeppesen P, Natarajan AT, Martínez-López W, Folle GA, Drets ME (2002). Chromosomal aberrations: formation, identification and distribution. Mutat Res 504: 17–36. Pal-Bhadra M, Bhadra U, Birchler J (1997). Cosuppression in Drosophila: gene silencing of Alcohol dehydrogenase by whiteAdh transgenes is Polycomb dependent. Cell 90: 479–490. Rabbitts TH, Boehm T, Mengle-Gaw L (1988). Chromosomal abnormalities in lymphoid tumors: mechanism and role in tumor pathogenesis. Trends Genet 4: 300–304. Richardson C, Jasin M (2000). Frequent chromosomal translocations induced by DNA double-strand breaks. Nature 405: 697–700. Robinson K, Beverley S (2003). Improvements in transfection efficiency and tests of RNA interference (RNAi) approaches in the protozoan parasite Leishmania. Mol Biochem Parasitol 128: 217–228.
Tsai AG, Lieber MR (2010). Mechanisms of chromosomal rearrangement in the human genome. BMC Genomics 11: S1.
Vamvakas S, Vock EH, Lutz WK (1997). On the role of DNA doublestrand breaks in toxicity and carcinogenesis. Crit Rev Toxicol 27: 155–174. van Gent DC, Hoeijmakers JH, Kanaar R (2001). Chromosomal stability and the DNA double-stranded break connection. Nat Rev Genet 2: 196–206. Veitia RA, Bottani S, Birchler JA (2008). Cellular reactions to gene dosage imbalance: genomic, transcriptomic and proteomic effects. Trends Genet 24: 390–397. Weinstock DM, Richardson CA, Elliott B, Jasin M (2006). Modeling oncogenic translocations: distinct roles for double-strand break repair pathways in translocation formation in mammalian cells. DNA Repair (Amst) 5: 1065–1074. Williams BR, Amon A (2009). Aneuploidy: cancer’s fatal flaw? Cancer Res 69: 5289–5291. Zech L, Haglund U, Nilsson K, Klein G (1976). Characteristic chromosomal abnormalities in biopsies and lymphoid-cell lines from patients with Burkitt and non-Burkitt lymphomas. Int J Cancer 17: 47–56.
897