Mitotic-Chromosome-Based Physical Mapping of the Culex quinquefasciatus Genome

The genome assembly of southern house mosquito Cx. quinquefasciatus is represented by a high number of supercontigs with no order or orientation on the chromosomes. Although cytogenetic maps for the polytene chromosomes of this mosquito have been developed, their utilization for the genome mapping remains difficult because of the low number of high-quality spreads in chromosome preparations. Therefore, a simple and robust mitotic-chromosome-based approach for the genome mapping of Cx. quinquefasciatus still needs to be developed. In this study, we performed physical mapping of 37 genomic supercontigs using fluorescent in situ hybridization on mitotic chromosomes from imaginal discs of 4th instar larvae. The genetic linkage map nomenclature was adopted for the chromosome numbering based on the direct positioning of 58 markers that were previously genetically mapped. The smallest, largest, and intermediate chromosomes were numbered as 1, 2, and 3, respectively. For idiogram development, we analyzed and described in detail the morphology and proportions of the mitotic chromosomes. Chromosomes were subdivided into 19 divisions and 72 bands of four different intensities. These idiograms were used for mapping the genomic supercontigs/genetic markers. We also determined the presence of length polymorphism in the q arm of sex-determining chromosome 1 in Cx. quinquefasciatus related to the size of ribosomal locus. Our physical mapping and previous genetic linkage mapping resulted in the chromosomal assignment of 13% of the total genome assembly to the chromosome bands. We provided the first detailed description, nomenclature, and idiograms for the mitotic chromosomes of Cx. quinquefasciatus. Further application of the approach developed in this study will help to improve the quality of the southern house mosquito genome.


Introduction
Mosquito-borne infectious diseases pose unacceptable risks to public health and welfare [1]. Among mosquitoes, species of the genus Culex are the most taxonomically diverse and geographically widespread [2,3]. Mosquitoes within the Cx. pipiens complex are major vectors for lymphatic filariasis caused by nematode Wuchereria bancrofti in tropical and subtropical regions of Asia, Africa, Central and South America, and Pacific Islands. Cx. quinquefasciatus is also a primary vector for arboviral infections, such as West Nile virus, St. Louis encephalitis, Sindbis, and Rift Valley fever viruses. Members of the Cx. pipiens complex have great variation in their host range, feeding behavior, and female diapause. Sequencing of the genomes for three major mosquito taxa, Anopheles gambiae [4], Aedes aegypti [5], and Cx. quinquefasciatus [6], provides important insights into genetic diversity of mosquitoes and evolution of the mosquito-pathogen interactions [7]. However, compared to other mosquitoes, Cx. quinquefasciatus has the most fragmented genome. A total of 579 Mb (mega base pairs) is currently assembled into 3,171 supercontigs with the N50 size being *476 Kb (kilo base pairs). The N50 supercontig sizes are 12.3 Mb in the An. gambiae (PEST) genome and 1.5 Mb in the Ae. aegypti genome. A lack of a high-quality chromosome-based genome assembly for Cx. quinquefasciatus remains a significant impediment to further progress in Cx. quinquefasciatus biology and comparative genomics of mosquitoes. Fragmented unmapped genome assemblies create substantial problems for genome analysis. For example, unidentified gaps cause incorrect or incomplete annotation of genomic sequences; unmapped sequences lead to confusion between paralogous genes and genes from different haplotypes, and the lack of chromosome assignment and orientation of the sequencing contigs does not allow for studying chromosome organization and evolution [8]. Therefore, utility of the genome assembly for investigations on basic biology requires anchoring of the genomic supercontigs onto Cx. quinquefasciatus chromosomes.
Among other approaches, genetic linkage mapping has been so far the most effective method for the genome mapping of Cx. quinquefasciatus. Among closely related Cx. pipiens species, several morphological mutants have been described by Leonore Dennhofer [9]. Crosses involving different mutants permitted assignment of genes related to these mutations to three linkage groups. Use of deoxyribonucleic acid (DNA) markers as a new approach allowed the construction of a genetic map which originally consisted of 21 complementary DNA (cDNA) markers and covered 7.1, 80.4, and 78.3 cM on chromosomes 1, 2, and 3, respectively [10]. The sex determination locus was genetically mapped to the smallest linkage group 1 in Cx. pipiens. In addition, multiple quantitative trait loci (QTL), related to differences in reproductive diapause between species in the Cx. pipiens complex, were also genetically mapped [11]. The most recent genetic linkage map developed for Cx. quinquefasciatus includes 63 genetic loci [12]. This map covered 29.5, 88.8, and 65.6 cM on linkage groups 1, 2, and 3 and allowed integration of 10.4% of the genome with the genetic linkage map. Currently, this is the most representative map of the Cx. quinquefasciatus genome. However, this map has never been integrated with cytogenetic maps developed for this mosquito.
Physical mapping in Cx. quinquefasciatus is challenging because of the poor quality of the polytene chromosomes. Several attempts to create a cytogenetic photomap using Cx. quinquefasciatus polytene chromosomes have been made. The Malpighian tubule chromosome map for Cx. pipiens [13] and Cx. quinquefasciatus [14] and, more recently, the salivary gland chromosome map for Cx. quinquefasciatus [15], were developed. However, correspondence of arms and regions among these maps and the original drawn map published by L. Dennhofer [16] is uncertain. Almost no similarities between landmarks of different chromosome maps were found [15]. These problems occurred because of low levels of polyteny, high frequency of ectopic contacts or associations of nonhomologous chromosome regions, and poor spreading of Cx. quinquefasciatus polytene chromosomes in preparation. As a result, only two genes for esterase-and odorant-binding proteins were mapped to the polytene chromosomes of Cx. quinquefasciatus [15,17].
In contrast to polytene chromosomes, mitotic chromosomes do not form ectopic contacts and can be easily utilized for mapping DNA probes to the chromosome bands. A simple and robust technique for obtaining high-quality mitotic chromosomes from imaginal discs of 4 th instar larvae was recently developed for the yellow fever mosquito Ae. aegypti [18]. This work resulted in 13%, and more recently, in 45% of the genome placement to the chromosomes for this mosquito [19,20]. Mitotic chromosomes of Cx. pipiens, the closest relative of Cx. quinquefasciatus, have been briefly described in 1963 as three pairs of metacentric chromosomes and numbered in order of increasing size as chromosomes I, II, and III [21]. In some cases, chromosome I was identified as a submetacentric chromosome, meaning that the relative length of the shorter arm p was less than 35% of the total chromosome length. It was also determined that Cx. pipiens chromosomes are smaller than those in Ae. aegypti. Chromosome measurements also demonstrated that compared with Ae. aegypti chromosomes Cx. pipiens chromosome I was disproportionally smaller than chromosomes II and III. Unlike in anophelines that have heteromorphic X and Y sex chromosomes [22], sex-determining chromosomes in Cx. quinquefasciatus were considered as homomorphic. Only two genes, 18S and 28S ribosomal DNA (rDNA), have been physically mapped to the smallest mitotic chromosome of Cx. pipiens [23]. Chromosome maps suitable for the physical mapping have not been developed for the mitotic chromosomes of Cx. quinquefasciatus.
In this study, mitotic chromosomes of Cx. quinquefasciatus were described in details and directly linked to the previously established genetic linkage groups by hybridization of 26 Bacterial Artificial Chromosome (BAC) probes associated with 58 genetic markers [24] to the chromosomes. As a result, chromosomes were renumbered according to the existing genetic linkage groups [10,12]. We also developed idiograms or schematic representation of chromosome banding patterns for Cx. quinquefasciatus. In addition, we mapped an 18S rDNA probe and 9 large genomic supercontigs to the chromosomes. Thus, our study has demonstrated that a mitotic chromosome band-based technique can be utilized for further development a highresolution physical genome map for the Cx. quinquefasciatus.

Mosquito strain and slide preparation
The laboratory strain Johannesburg (JHB), used in this study was obtained from BEI Resources [25]. This colony was originated from the field population of Cx. quinquefasciatus near Johannesburg, South Africa [15]. The same strain was previously used for the genome sequencing project [6]. Adult mosquitoes were kept at 26°C and fed on artificial membrane blood feeders 4-5 days after emerging. Approximately 4 days after feeding, the eggs were collected and hatched at 26°C. After 4 days, 2 nd instar larvae were transferred to 16°C to obtain a high number of mitotic divisions in imaginal discs [18]. At 7-8 days, 4 th instar larvae were used for slide preparation. Our study utilized mitotic chromosomes from imaginal discs of 4 th instar larvae which develop into legs and wings at the adult stage. These imaginal discs are located immediately under the cuticle and can be easily dissected from the larvae. The morphology of the imaginal discs and details of their dissection from the larvae were previously described for three species of mosquitoes including Cx. quinquefasciatus [26]. Chromosome preparations were made using a routine technique based on hypotonic treatment and subsequent application of Carnoy's solution (3 parts of ethanol: 1 part of acetic acid) and 50% propionic acid [26].
The percentage of chromosome preparations suitable for further analysis, which contained more than 50 chromosome spreads, was *85%.

DNA probe preparation and fluorescent in situ hybridization (FISH)
Notre Dame Johannesburg (NDJ) BAC library [24] was used as a probe DNA source for FISH. BAC clone DNA isolation and sequencing were performed at the Clemson University Genomics Institute. BAC clone correspondence to the certain genomic supercontigs or genetic markers was determined by BAC library screening [26] or by BAC-end sequence comparison using Basic Local Alignment Search Tool (BLAST) against the genome assembly of Cx. quinquefasciatus available at Vectorbase [27]. Three polymerase chain reaction (PCR) fragments with sizes *1Kb from genomic supercontig 3.32 were amplified using primers: AAAACCC ATCTCCCTCGTAG forward, GCTTCTCCAAAACCTTCCTC reverse; TCAAACGACCAC AACTTTGA forward, TGGCCTTGTTCTTCTTCTTG reverse; and ATGAAGTTACGGTC GTCAGC forward, AGTGCATGATGACTCCCATT reverse. Probes were labeled by nick translation with Cy3-or Cy5-deoxyuridine 5-triphosphate (dUTPs) (GE Healthcare UK Ltd., Buckinghamshire, UK) as described before [26]. An 18S rDNA probe was amplified using forward primer CCTATATGGTGGCGCTTGAT and reverse primer AACTAAGAACGGCC ATGCAC. It was labeled by Cy3-or Cy5-dUTPs in a PCR reaction using PCR IMMOMIX (Bioline USA, Taunton, MA) with standard parameters. Nonspecific hybridization of BAC DNA probes to the chromosomes was prevented by pre-hybridization of the probe with unlabeled repetitive DNA fractions of genomic DNA [26]. Genomic DNA was extracted from adult mosquitoes using Qiagen Blood & Cell Culture DNA Maxi Kit (Qiagen Science, Germantown, MD, USA). Approximately 500 mg of adult mosquitoes were taken for extraction. Final outcome of repetitive DNA fractions accounts for *20% of genomic DNA. Approximately 250-350 ng of DNA probe were pre-hybridized with 4 mg of repetitive DNA. FISH of DNA probes was performed using a standard protocol [26]. Slides were analyzed using Zeiss LSM 510 Laser Scanning Microscope (Carl Zeiss Microimaging, Inc., Thornwood, NY, USA) at 600X magnification. For each probe, from 5-10 chromosome spreads were tested.

Image processing and measurements
For idiogram development, the best images of the chromosomes from imaginal discs stained with Oxasole Yellow (YOYO-1) iodide (Invitrogen Corporation, Carlsbad, CA, USA) were selected. The original images were converted into gray-scale images and contrasted as described previously [28]. These chromosome images were straightened and aligned for comparison using ImageJ program [18,29]. In total, 150 chromosomes at early metaphase were analyzed, and 25-30 images of each chromosome with reproducible banding patterns were used for idiogram development. To calculate exact proportions of chromosomes, we utilized standard curve measurements in Zen2009LightEdition software [30]. We utilized early metaphase and midmetaphase chromosomes to precisely assign signals to the particular chromosome band.

Culex quinquefasciatus chromosome nomenclature
According to an original chromosome nomenclature, three pairs of metacentric chromosomes of Cx. pipiens, the closest relative of Cx. quinquefasciatus, were numbered as I, II, and III in order of increasing size (Rai, 1963). In this study, we established correspondence between mitotic chromosomes and genetic linkage groups by direct placement of 26 genomic supercontigs associated with 58 genetic markers to the chromosomes (Table 1). These markers were previously mapped to smallest, largest, and intermediate linkage groups 1, 2, and 3 of Cx. pipiens, respectively [10]. Seven BAC clones for this mapping were identified by screening the NDJ BAC library [24] using PCR-amplified genetic markers [31]. Another 19 BAC clones were identified as belonging to the genomic supercontigs with known genetic markers by BAC-end sequencing followed by BLAST-based alignment to the genomic sequences of Cx. quinquefasciatus [27]. BAC clones corresponding to linkage group 1 carrying markers CX60, LF284, and The order of most markers on chromosomes exactly followed their positions in the genetic linkage map. We found only two discrepancies in the order of markers CX44 and LF203 on chromosome 2 and also a BAC clone with genetic marker LF108 was mapped on a different chromosome. Thus, we propose renumbering the mitotic chromosomes for Cx. quinquefasciatus in correspondence to the genetic linkage groups as follows: 1-smallest, 2-largest, and 3-intermediate chromosomes.
We also determined the average chromosome lengths at mid-metaphase as 4.04 μm for chromosome 1, 6.37 μm for chromosome 2, and 5.59 μm for chromosome 3 ( Table 2). Relative lengths of chromosomes were 25.3%, 39.8%, and 34.9% for each chromosome, respectively. Centromeric indexes (the relative length of the p-arm) were 47.4% and 46.9% for chromosomes 2 and 3. Thus our data confirmed that these two chromosomes are metacentric [21]. Measurements for the centromere position in chromosome 1 varied depending on size of the ribosomal locus determined by FISH of 18S rDNA probe on the chromosome between 43.1% or 48.1%, respectively (Fig. 1). Thus, both variants of chromosome 1 must be also considered as metacentric according to the modern chromosome nomenclature [32].

Idiograms of mitotic chromosomes for Culex quinquefasciatus
In addition to chromosome nomenclature, our study developed idiograms or drawn schematic representations of the banding pattern for mitotic chromosomes of Cx. quinquefasciatus at early-metaphase. From a whole range of the different stages of mitosis (prophase, prometaphase, metaphase, and anaphase), metaphase chromosomes have the most clear and reproducible banding patterns (Fig. 2). Similarly to Ae. aegypti, in Cx. quinquefasciatus homologous chromosomes are paired at prophase and prometaphase ( Fig. 2A, B). At these two stages, the visible chromosome number equals three. Chromosomes start segregating from each other at prometaphase and become completely segregated at metaphase (Fig. 2C). Visible chromosome number at metaphase equals six. For the idiogram development, we used images of the chromosomes at early metaphase stained with YOYO-1 iodide (Fig. 2D). This dye strongly stains euchromatin and therefore provides more detailed banding patterns than DAPI (4',6-diamidino-2-phenylindole), which preferentially stains AT-rich heterochromatic regions [18,26]. The original chromosome pictures were converted into gray-scale images. Chromosome images were then straightened and aligned for comparison. After that, unique and reproducible patterns for each chromosome were identified. Following human chromosome nomenclature [33], we determined four color intensities of the chromosome bands: intense (black), medium intensity (dark gray), low intensity (light gray), and negative (white). Chromosomes were subdivided into 20, 28, and 24 bands for chromosomes 1, 2, and 3, respectively (Fig. 2E). The total number of bands for all chromosomes of Cx.quinquefasciatus equals 72. Each chromosome of Cx. quinquefasciatus has unique features or landmarks for the arm recognition: large negative band containing ribosomal locus in the q arm region of chromosome 1, negative band separating intense and mediumintense bands in the p arm of chromosome 2, and large negative band in the middle of arm q on chromosome 3 (Fig. 2).

Physical mapping on mitotic chromosomes of Culex quinquefasciatus
In addition to BAC clones associated with genetic markers, 9 BAC clones from the largest genomic supercontigs and 18S ribosomal DNA (Table 2) were also mapped to the bands on idiograms by FISH (Fig. 3). An 18S ribosomal DNA probe hybridized above the 2 dark bands on the q arm of chromosome 1 in region 1q13 on the idiogram. In total, a majority of the DNA probes (17) hybridized to the largest chromosome 2, 9 BAC clones were found in intermediatesized chromosome 3, and 11 DNA probes hybridized to the smallest sex-determining chromosome 1. To simplify physical mapping, we optimized a landmark-guided approach developed for Ae. aegypti [20,34] for Cx. quinquefasciatus chromosomes (Fig. 4). We hybridized two BAC clones of interest in the presence of 3 landmark probes: 18S rDNA for 1q arm, telomere BAC clone with genetic marker LF334 on 2q arm, and a BAC clone with genetic marker CX112 close to telomere for 3q arm. Two BAC clones on 3q arm carrying genetic markers CX17 and CX112 were ordered within the band on 3q arm using a two-step mapping approach [19]. In addition to FISH on metaphase mitotic chromosomes (Fig. 4D), the FISH results on prophase and polytene chromosomes were also analyzed. This additional step permitted the ordering of these genetic markers within chromosome band (Fig. 4E, F).

Discussion
Knowledge about cytogentics of mosquitoes is important to better understand their genome organization and function. However, mitotic chromosomes for the major vector of lymphatic filariasis Cx. quinquefasciatus were only briefly mentioned, as three pairs of metacentric chromosomes, more than 50 years ago in a review on mosquito mitotic chromosomes [21]. For other mosquitoes such as An. gambiae and Ae. aegypti, chromosome work was performed back in the 1970s and 1980s, but it was not done for the Cx. quinquefasciatus. Here, for the first time, we described the details of morphology, length, and proportions for the mitotic chromosomes of Cx. quinquefasciatus. Our measurements demonstrated that a total chromosome length at mid-metaphase in Cx. quinquefasciatus is 1.5 times longer than in An. gambiae and 1.5 times shorter than in Ae. aegypti (Table 2) [26]. It reflects the difference in genome sizes of 264 Mb, 579 Mb, and 1376 Mb in An. gambiae, Cx. quinquefasciatus, and Ae. aegypti, respectively. Our measurements also indicated the presence of two variants of sex-determining chromosome 1 in Cx. quinquefasciatus that differ from each other by the size of ribosomal locus and the centromere position on this chromosome. This polymorphism can be potentially related to the sex determination in this mosquito.
Our study also, for the first time, integrated the cytogenetic map with the genetic linkage map [10,12] and created a new chromosome nomenclature for Cx. quinquefasciatus. Originally numbered as I, II, and III in order of increasing size [21], chromosomes were renumbered as 1-smallest, 2-largest, and 3-intermediate chromosomes in correspondence to the genetic linkage map of Cx. quinquefasciatus. The correspondence between genetic linkage groups and chromosomes was determined by direct positioning of 58 genetic markers on the chromosomes. The order of the markers on the chromosomes were basically the same as on the previously published genetic linkage map [12] with only two exceptions in order of the markers CX44 and LF203 on chromosome 2 and position of the genetic marker LF108 on a different chromosome. For Ae. aegypti, linkage groups were associated with chromosomes based on the analysis of X-ray-generated chromosome translocations in 1970 [35]. The smallest, largest, and intermediate chromosomes were also renumbered as 1, 2, and 3, according to the genetic linkage groups. Recent physical mapping of 100 genetic markers of Ae. aegypti helped to clarify the chromosome positions of 12 QTL related to pathogen transmission: the filarioid nematode [36], the avian malaria parasite [37,38], and dengue virus [39,40]. These markers were combined into five major clusters of QTL on the chromosome map suggesting that transmission of various pathogens can be controlled by the same genomic loci [19]. However, QTL related to the pathogen transmission for Cx. quinqufasciatus have not been identified yet. Only multiple QTL to diapause were described [11]. Thus, further development of integrated genetic linkage and chromosome map requires both physical mapping and additional QTL identification.
In addition to chromosome nomenclature, our study developed detailed idiograms for the Cx. quinquefasciatus chromosomes. The whole chromosome complement was subdivided into 19 regions and 72 bands with four different intensities. We demonstrated the utility of our idiograms for the physical genome mapping by placement of 37 genomic supercontigs to the chromosome locations. Chromosome idiograms for Cx. quinquefasciatus are comparable to that previously developed for the yellow fever mosquito Ae. aegypti [18]. The total number of bands (72) is slightly lower in Cx. quinquefasciatus than in Ae. aegypti (94). Physical mapping approach based on idiograms allowed assignment of 45% of the Ae. aegypti genome to chromosome bands [20]. Additional physical mapping, based on the idiograms developed by this study for Cx. quinquefasciatus, needs to be conducted to increase the genome assignment to the chromosome position.
Using mitotic chromosomes for physical genome mapping raises a concern about the low resolution of this mapping approach compared to traditionally used polytene chromosomes [41][42][43][44]. Unlike the subfamily Anophelinae, which have well-developed polytene chromosomes, mosquitoes from the Culicinae subfamily lack high-quality polytene chromosome spreads [14,45]. Nevertheless, our study determined that in addition to mitotic chromosomes low-polytenized chromosomes from salivary glands can be used for the ordering of closely located supercontigs of Cx. quinquefasciatus without assigning them to specific bands in polytene chromosomes (Fig. 2). This so called "two-step" mapping approach was successfully used for the ordering of 100 genomic supercontigs in Ae. aegypti [19]. This strategy significantly increased the final resolution of the physical map. The distance between two signals that can be distinguished from each other was estimated at 300 Kb for the polytene chromosomes of Ae. aegypti. The resolution of polytene chromosomes in Cx. quinquefasciatus is higher due to their better polytenization and comparable to the 100-Kb resolution of polytene chromosomes in An. gambiae [42].
Previous investigations of chromosome arm homology between Cx. quinquefasciatus, An. gambiae, and Ae. aegypti indicated whole-arm conservation between Cx. quinquefasciatus and An. gambiae, and a whole-arm translocation between chromosomes 2 and 3 of Cx. quinquefasciatus and Ae. aegypti [6]. This conclusion was based only on 9%, 31%, and 88% genome placement to the chromosomes for Cx. quinquefasciatus, Ae. aegypti, and An. gambiae, respectively [10,42,46]. The dramatic gene order reshuffling between homologous chromosomes of Ae. aegypti and An. gambiae was recently demonstrated based on 45% and 88% genome placement to the chromosomes for these two mosquitoes, respectively [20]. Additional physical mapping may provide some new insights into chromosome evolution in Cx. quinquefasciatus. For example, FISH results for 18S rDNA suggest an inverted position of the ribosomal locus in chromosome 1 of Cx. quinquefasciatus compared with Ae. aegypti [19]. This locus was mapped close to the centromere above the dark bands in Cx. quinquefasciatus (Fig. 1) but in the middle of the 1q arm below the dark band in Ae. aegypti. Our measurement data of Cx. quinquefasciatus chromosomes support the previous observation that the proportions between sex-determining chromosome 1 and autosomes 2 and 3 differ in Cx. pipiens and Ae. aegypti (Table 2). It is clear that the relative length of chromosome 1 is shorter in Cx. quinquefasciatus than in Ae. aegypti. These results support the idea of partial degradation of the sex-determining chromosome 1 in Cx. quinquefasciatus compared to chromosome 1 in Ae. aegypti. Degradation of sex chromosomes was described in different lineages of Drosophila [47]. However, a more advanced chromosome-based genome map for Cx. quinquefasciatus is required for clarifying the intimate details of chromosome evolution in mosquitoes.

Conclusion
Our study developed a mitotic-chromosome-based approach for physical mapping of the Cx. quinquefasciatus genome. We provided the first detailed description and offered a new nomenclature for mitotic chromosomes of Cx. quinquefasciatus. Based on the genetic linkage map, the smallest, largest, and intermediate chromosomes were numbered as 1, 2, and 3, respectively. We demonstrated the efficiency of our physical mapping approach by placing 37 genomic supercontigs and 58 genetic markers onto chromosome idiograms. This effort, together with previously conducted linkage mapping [12], resulted in the chromosome assignment of 13% of the total genome assembly. Further application of the approach described here will improve the current highly fragmented genome assembly of Cx. quinquefasciatus and will also stimulate research in vector biology and comparative genomics in mosquitoes.