﻿A critical review on cytogenetics of Cucurbitaceae with updates on Indian taxa

﻿Abstract The cytogenetic relationships in the species of Cucurbitaceae are becoming immensely important to answer questions pertaining to genome evolution. Here, a simplified and updated data resource on cytogenetics of Cucurbitaceae is presented on the basis of foundational parameters (basic, zygotic and gametic chromosome numbers, ploidy, genome size, karyotype) and molecular cytogenetics. We have revised and collated our own findings on seven agriculturally important Indian cucurbit species in a comparative account with the globally published reports. Chromosome count (of around 19% species) shows nearly three-fold differences while genome size (of nearly 5% species) shows 5.84-fold differences across the species. There is no significant correlation between chromosome numbers and nuclear genome sizes. The possible trend of evolution is discussed here based on molecular cytogenetics data, especially the types and distribution of nucleolus organizer regions (NORs). The review supersedes the scopes of general chromosome databases and invites scopes for continuous updates. The offline resource serves as an exclusive toolkit for research and breeding communities across the globe and also opens scope for future establishment of web-database on Cucurbitaceae cytogenetics.


Introduction
The family Cucurbitaceae contains an extensive range of diversity consisting of about 1000 species spread over 96 genera (Renner and Schaefer 2017). The diversity of plant families is associated with variation in genome sizes and chromosome numbers as a result of enormous adaptive radiation (Soltis et al. 2004;Lysák and Schubert 2013). The viewpoint of evolution has been changed with the understanding of whole genome duplication (WGD) (Soltis et al. 2014) followed by core-eudicot hexaploidy . A cytogenetic database is essential to gain insights into evolution by supplementing phylogeny trees with chromosome number information (Mota et al. 2016) to upgrade knowledge on plant systematics (Soltis et al. 2014;Viruel et al. 2021). Cucurbitaceae, being the fourth most important and one of the earliest consumed vegetables yielding family, has coped with extreme climates, extensive human intervention and a huge domestication syndrome (Chomicki et al. 2020). Considerable advances have been made in molecular phylogeny (Renner and Schaefer 2016;Bellot et al. 2020;Chomicki et al. 2020;Guo et al. 2020) and genomics (CuGenDB, http://cucurbitgenomics.org) (Zheng et al. 2019).
We had previously discussed about the gaps in cytogenetic studies (Bhowmick and Jha 2015b) which has been surmounted with the advent of molecular cytogenetics.
Currently, we have collated the cytogenetic reports of Cucurbitaceae globally and integrated our own findings for a collective interpretation. The review attempts to address i) the trend of chromosome evolution in specific tribes and species based on available information, ii) correlation between chromosome numbers and ploidy or genome size in the studied taxa and iii) the requirement of an exclusive cytogenetic catalogue for genome researchers, taxonomists and breeders working on Cucurbitaceae.

Chromosome analysis in the Cucurbit species ocurring in India
Presently an enzymatic maceration and air drying (EMA) method followed by flurochrome banding has been employed as per our previous protocols (Bhowmick et al. 2012(Bhowmick et al. , 2016Bhowmick and Jha 2015a to represent fresh karyotypes of seven agriculturally important cucurbit species (Table 1) belonging to Benincaseae and Sicyoeae. Fresh and healthy roots were used from different sources (like germinating seeds, seedlings and underground root stocks). Roots were pretreated with 0.002 M hydroxyquinoline and fixed in 1:3 aceto-methanol solution. The standardization of EMA-fluorescence banding was conducted for the different species. In brief, fixed roots were digested in enzyme mixture [1% Cellulase (Onozuka RS), 0.75% Macerozyme (R-10), 0.15% Pectolyase (Y-23), 1 mM EDTA] for 40-45 min at 37 °C, macerated on slides, air-dried, stained with 2% Giemsa solution (Merck, Germany) and plates selected for karyotyping. After de-staining, slides were kept in McIlvaine buffer, stained with 0.1 µg mL -1 DAPI for 15-20 min in darkness. For CMA staining, slides were incubated in 0.1 mg mL -1 CMA for 15-25 min in darkness. For meiotic chromosomes, fixed anthers were digested in enzyme mixture for 5-8 min, macerated on slides and DAPI staining protocol was followed with minor modifications. All slides were mounted in non-fluorescent glycerol and chromosome plates were observed under a Zeiss Axioscop 2 fluorescence microscope (using UV and BV filter cassettes for DAPI and CMA stains, respectively). Images were captured using the attached ProgRes MFscan Jenoptik D07739 camera and ProgRes CapturePro 2.8.8 software.

Statistical analyses
Statistical analysis involving foundational cytogenetic parameters have been demonstrated to imply significant knowledge on chromosomal evolution within a group (Winterfeld et al. 2020). Considering the lack of hypotheses, we have tested for correlation between the dependent variables (2C genome size, MCL and HCL) and predictor variables [chromosome number (2n) and ploidy level (pl)] and also calculated linear models for regression analysis using IBM SPSS (v23, free).

The modern cytogenetic catalogue of cucurbitaceae
Along with the global review, fresh EMA based somatic plates and idiograms (Figs 1-3) of Indian species are presented here. We retain the previous designation of 10 tribes as 'understudied' (Bhowmick and Jha 2015b), excluding Indofevilleeae, having no cytological reports.

Nuclear genome contents
Nuclear genome sizes are reported in 49 species (~5% of total species) belonging to 15 genera (~16% of total genera) of Cucurbitaceae. Among the understudied tribes, 2C genome content is known for one species each from Gomphogyneae and Coniandreae (Table 2). Within the Momordiceae species of India, significant interspecific genome size differences have been reported (Ghosh et al. 2021). The species differed 5.19-fold in their genome sizes (2C = 0.72-3.74 pg) (Table 3) (Ghosh et al. 2021). Interestingly, the species with lowest chromosome number (M. cymbalaria, 2n = 18) contained highest nuclear DNA content among the four Momordica species (Table 3). In Bryonieae, flow cytometric genome size of Bryonia shows a 2.2-fold increase than Ecballium (Table 4).
In case of Sicyoeae, flow cytometric 2C DNA content ranges from 1.49-2.32 pg/2C, indicating 1.55-fold differences in genome size. Echinocystis lobata Michaux, 1803, in spite of tetraploid condition, shows lowest genome size (Table 5). There is no significant difference in genome size between the genders of Trichosanthes dioica Roxburgh, 1832   (Table 6). The divergence in genome size between genders was found to be highest in dioecious C. grandis (Table 6), a sharp contrast to dioecious Trichosanthes dioica (Table 5). Benincaseae shows a 3.07-fold overall difference in genome size. Genome sizes are known in eight species of Cucurbita Jussieu, 1789. Flow cytometric genome size ranges from 0.686-0.933 pg/2C, indicating a 1.36-fold variation (Table 8). Despite polypoidy, the nuclear DNA content of Cucurbita species is comparable to many diploids.
Karyotypes and chromosome sizes are reported in ten species of Momordiceae (Table  3). Interspecific differences have been observed and found to correlate with phylogenetic relationship within Momordica (Ghosh et al. 2021). Infraspecific delimitation of Indian M. charantia varieties was based on fluorochrome banding pattern and genome size divergence (Table 3), corresponding to infraspecific distinction reported in the Japanese bitter gourd cultivars (Kido et al. 2016). FISH in three Momordica species revealed 45S and 5S rDNA sites to be localised on different chromosomes (Table 3). In context of the genome sequence of bitter gourds (Matsumura et al. 2020), further scopes for cytogenetic and genomic investigation remain open.
Benincaseae generally reveal two distal 45S rDNA loci of which at least one locus is either adjacent to 5S rDNA locus (Table 6) or co-localized in the same chromosome as in most of the Cucumis species (Table 7). Exceptionally, a wild species of Benincasa (B. fistulosa Stocks, 1851) has non-adjacent 45S and 5S signals . GC rich satellites were observed in the 12 th pair of chromosomes showing CMA + bands in cultivated Indian ashgourd (B. hispida) (Fig. 3 A-C, J, Tables 1, 6). Lagenaria siceraria Molina, 1782 and Cucumis melo Linnaeus, 1753 are the other two genera having similarity in rDNA hybridization profile, agreeing with phylogenetic affinity .
The genus Cucumis is the largest in Benincaseae with 65 species of which 39 have been studied (Table 7). Among the Cucumis species with x = 12, co-localization rDNA loci (45S and 5S rDNA) have been documented in 14 species, including C. melo (Table  7). However, the number of 45S sites is generally four, which may be six or eight in some cases (Table 7). rDNA hybridization data strongly corroborated with the 'fusion' theory for derivation of x = 7 (C. sativus) from x = 12 (C. melo) (Waminal and Kim 2012) which is substantiated by genomic studies (Li et al. 2011). There are ten pericentromeric/ centromeric 45S and two distal 5S rDNA sites in C. sativus while six 45S rDNA sites were reported in C. sativus var. hardwickii Royle, 1835 (Koo et al. 2005;. Comparative chromosome painting (Lou et al. 2014) and GISH  proved high colinearity between cucumber and melon. Based on chloroplast and nuclear DNA (ITS) phylogeny, C. melo (melon) has been found to be sister to a clade comprising C. sativus and related genera (Dicaelospermum Clarke, 1879 and Mukia Arnott, 1840) (Renner et al. 2007). rDNA site co-localization was found to coincide with geographical origin of 12 Cucumis species (Zhang et al. 2016). The chromosomal affinity between C. metuliferus Schrader, 1838, C. anguira Linnaeus, 1753, C. zeyheri Sonder, 1862, C. myriocarpus Naudin, 1859 and polyploid C. heptadactylis Naudin, 1859 (dioecious) (Yagi et al. 2015) can be substantiated by their phylogenetic proximity based on chloroplast and nuclear DNA (ITS) sequences (Renner et al. 2007). rDNA distribution of C. metuliferus was also the reason to consider proximity with Citrullus naudinianus Sonder, 1862, (previously Acanthosicyos naudinianus Sonder, 1862) (Reddy et al. 2013). Infraspecific differences were documented in Cucumis melo on the basis of 45S-5S rDNA signals (linked or separated) which also possessed unique centromeric satellites (Setiawan et al. 2018(Setiawan et al. , 2020. Moreover, chromosome painting method elucidated chromosomal rearrangement in some Cucumis species (Lou et al. 2014;Li et al. 2018).
The dramatic evolution of Y chromosome was validated in karyotypes ( Fig. 3 D-I, K-L) of Coccinia grandis ( Table 6). The 45S rDNA sites enabled confirmation of NORs in the 8 th and 12 th pair containing distal GC rich CMA + signals in C. grandis (Fig. 3 D-I, K-L, Tables 1, 6). 45S and 5S rDNA hybridization pattern was similar in three other Coccinia species and Diplocyclos palmatus Linnaeus, 1753 (Table 6). The three closely related dioecious species of Coccinia accumulated Y chromosome repeats and displayed sex chromosome turnover . Strong centromeric CMA bands (Fig. 3 D-I, K-L, Table 1) were observed in C. grandis except Y chromosome (Fig. 3 I, L), presenting a possibility that CgCent (CL1) is a feature of centromeres of dioecious Coccinia species ). In addition, non-nucleolar CMA + heterochromatin might be associated with sexual differentiation of autosomes in dioecious C. grandis (Fig. 3) which is also a marker in Trichosanthes dioica (Fig. 2, Table 1), opening good scope for further study.
Distinct 45S rDNA sites are higher in number than 5S rDNA sites in Cucurbitaceae (Fig. 4) (Waminal and Kim 2012). The distal 45S rDNA loci are conserved genomic landmarks (Fig. 4) while 5S rDNA loci are relatively diverse (Fig. 4). Based on the literature reports, some NORs (Type I) included chromosomes showing noncolocalized 45S and 5S rDNA sites in seven species of Benincaseae, one species each from Cucurbiteae and Momordiceae and two species of Sicyoeae. The rearrangement of 45S rDNA site in Cucumis sativus, probes for chromosome number reduction which may be a consequence of diploidization. The second type (Type II) shows colocalised 45S and 5S rDNA loci, either adjacent or distant, but always on the same chromosome and found in one species each of Benincaseae, Sicyoeae and Actinostemmateae. The third type (Type III) was characterized by chromosomes with non-colocalized and colocalised 45S and 5S rDNA loci, as in 14 species of Benincaseae and one species each of Sicyoeae and Thladiantheae. The rDNA sites of majority of Cucumis species were of non-adjacent type. Hence, type III NORs in majority of Benincaseae genera advocates conservation of the marker chromosomes having distal NOR (45S rDNA). Gynostemma pentaphyllum and some polyploid Cucumis reveal rDNA loci reduction after polyploidization (Zhang et al. 2016;Pellerin et al. 2018).

Correlation between parameters
Chromsome numbers in Cucurbitaceae range from x = 5 to x = 16. The most prevalent number x = 12 (Fig. 5) is considered ancestral (Xie et al. 2019b), followed by x = 11, 13, 14 and 10 (Fig. 5). The present regression analyses for 41 taxa (including 16 Indian taxa) (Table 9) revealed significant linear correlation between 2n and HCL, between ploidy and genome size and between ploidy and HCL (Fig. 6). Therefore, an increase in ploidy/ 2n number is linked with increase in HCL. There was no significant correlation between 2C genome size and chromosome numbers. Cytogenetic parameters may not reflect residual evidence of CCT in Cucurbitraceae at present, as reasoned by Alix et al. (2017). The numbers in brackets beside names of genera signify the number of species whose chromosome counts are reported. The % of genera and species with a particular chromosome number, is indicated at the end arrow (out of a total of 44 genera and 188 species with chromosome counts).

Future directions
Chromosome number and genome size information in the basal clades (understudied tribes) should be given attention to infer ancient base numbers. The parameters of fundamental and molecular cytogenetics are inevitable for genomic interpretation (Weiss-Schneeweiss and Schneeweiss 2013; Deakin et al. 2019) and hence relevant to spot genetic resources and relationships with wild relatives. The current review is not exhaustive but supersedes the scopes of general web resources and brings an offline resource exclusive for Cucurbitaceae.