GWAS revealed novel loci associated with resistance to a short duration and a long duration drought stress
The GWAS analysis identified a total of 17 and 22 significant marker-trait associations for four traits (CW, LSEN, AGRmax-Yr, CH-Yr) in the short-duration and long-duration drought experiments respectively. None of these SNPs was common to the two experiments. This was not completely unexpected because the overall phenotypic response of the EUCLEG soybean collection in two experiments was different [23]. A total of 12 and 16 of the genes that are in LD with these SNPs (for SDS and LDS experiments respectively) are of particular relevance for drought stress responses including stomatal movement, root formation, photosynthesis, ABA signaling, cellular protection and cellular repair mechanisms. Furthermore multiple previously reported QTLs related to drought resistance traits such as WUE, chlorophyll content and photosynthesis, and QTLs for traits including plant height and other yield-related traits, disease resistance, nutrient content, seed composition also co-localize with the significantly associated SNPs.
Short duration and long duration drought treatment induce differential mechanisms of canopy wilting in soybean
Slow or delayed wilting is known to be a complex trait, that can be the result of several underlying mechanisms in soybean: (i) good water resource exploration by a large root system [61]; (ii) lower stomatal conductance, reduced transpiration rate and high water use efficiency [62]; (iii) maintenance of constant transpiration rate under vapor pressure deficit conditions above 2.0 kPa [63]; and (iv) lower radiation use efficiency [64]. Several of the candidate genes identified in the SDS experiment are putatively involved in lateral root development, and some are related to photosynthesis. Root system modification is an important response that shows a large phenotypic plasticity in soybean under drought stress [15]. This response is a drought avoidance mechanism enabling plants to sustain high plant water status or cellular hydration under drought [65]. Drought stress inhibits root growth in general but susceptible genotypes show more prominent effects [66,67] due to an overall decrease in newly synthesized cell wall polysaccharides such as pectin, hemicellulose, and cellulose [68]. [67] discuss how the taproot length and tertiary root length influence the root surface area, what in turn influences the plant nutrient and water absorption capacity. Similarly, soybean genotypes with high root length, surface area, diameter and volume achieved high net photosynthesis, attained higher plant height and biomass, and tended to perform better under water deficit conditions [66]. The well-described slow wilting soybean accession PI 796397 has a dense root system with a high number of root tips [69]. We have shown a large level of variation for drought index for canopy height (CH-Yr) and number of pods per main stem (PPS-Yr) in the EUCLEG soybean collection [23]. Some accessions were able to maintain high CH and displayed no reduction in the number of pods per plant (NPP), but the relationship between these traits and CW was not clear. This might be due to compensation during the period of stress recovery as discussed in [23]. Anyhow, the results obtained in this work regarding marker-trait associations support the hypothesis that the mechanism of slow wilting in the SDS experiment might be associated with the characteristics of the root system.
In the LDS experiment, the associated SNPs were in LD with genes related to stress-induced abscisic acid (ABA) accumulation, stability of the photosynthetic apparatus, oxidative stress response and ROS scavenging, and trans-membrane water and solute transport, along with one gene related to root development. Some candidate regions in LDS also overlapped with previously reported QTLs for WUE [70]. In [23] we have reported a drastic reduction of the growth rate, plant height and yield in the LDS treatment. These plant responses match with the dehydration tolerance mechanism that is activated under extreme drought stress conditions [71]. Dehydration tolerant plants tend to maintain metabolic activities at low tissue water potential through osmotic adjustment, antioxidant activities, and altered growth regulators [72]. Soybean plants subjected to drought for 20 days at V5 stage in [73] showed a higher accumulation of ABA, lower stomatal conductance and decreased rate of net photosynthesis, but resistant genotypes performed relatively better than the susceptible ones. In [74], drought stress disturbed the balance between ROS and antioxidant enzymes, and a severe stress led to overproduction of ROS causing cellular damage, low stomatal conductance and a decrease in photosynthesis parameters in soybean. Drought tolerant soybean genotypes in [22] developed an anti-oxidative defense mechanism for ROS scavenging by increasing the activity of antioxidant enzymes including the peroxidase superoxide dismutase, ascorbate peroxidase, glutathione reductase and catalase, which helped them to maintain a high photosynthetic efficiency and a high RuBisCo activity under heavy stress conditions. Our results suggest that under severe stress, canopy wilting response in soybean is regulated by a different mechanism than under short duration stress. Based on the marker-traits associations identified in the LDS experiment, slow wilting genotypes can achieve a high WUE and thus a high level of resistance to long-term drought stress through a reduction of stomatal conductance and transpiration rates.
Drought-induced leaf senescence is related to ABA and ROS responses in soybean
GWAS identified 3 and 13 significant SNPs for LSEN in the SDS and LDS experiments respectively. Premature leaf senesce is thought to be an indicator of stress vulnerability in plants [75]. [76] described a negative correlation between premature leaf senescence and plant survival in a perennial temperate grass. According to the mechanism of premature leaf senescence, accumulation of ABA under stress conditions promotes overproduction of ROS which may react with proteins, lipids and deoxyribonucleic acid, leading to oxidative damage and premature leaf senescence [77,78]. Several transcription factors that play a role in the control of age-induced leaf senescence, play also a role in plant stress tolerance, and chloroplast breakdown is a common feature in both cases [75]. Here we also identified multiple candidate genes for LSEN. Three candidate genes known to be involved in ABA-mediated stomatal closure, root cap development and ROS scavenging were identified in the SDS experiment. In the LDS experiment, six candidate genes had stress related functions such as ABA-mediated seedling survival, maintenance of seed shape and seed size at high ABA levels, drought and salt tolerance through proline accumulation and ROS scavenging, and stability of Photosystem I. In the work of [79] a higher WUE and an increased tolerance obtained through the activation of a cell wall invertase in tomato was associated with a low stomatal conductance, delayed senescence, increased source activity and a better control of ROS production under drought stress. In [80] minimizing the stress-mediated senesce by the over expression of a senescence associated gene enabled the plants to attain higher shoot and root biomass and to recover better after a period of drought stress. Our results also reveal that drought-induced premature senescence in soybean can be avoided by targeting the genes involved in ROS scavenging and stomatal conductance.
Furthermore, our results reveal that two candidate regions (LSEN_LDS_3 and LSEN_LDS_4a on Gm16) can be of particular significance as they co-localize with QTLs for net photosynthesis rate (qP16; [59]) and for total chlorophyll content (eChl-T; [58]). Moreover, the candidate gene Glyma.16G145800, having a putative role in chlorophyll synthesis [81], is also present in the candidate region LSEN_LDS_4a. It has been discussed that the drought-induced accelerated senescence affects source-sink relationship in wheat [82] and causes a significant yield reduction in soybean [83]. The candidate region LSEN_LDS_6 on Gm19 contains a gene encoding a heat shock protein which confers stress tolerance [84,85]. The same region also contains many previously reported QTLs for important agronomic traits including leaf growth, plan height, internode length, number of pods and seed amino acid content, indicating the relevance of this candidate region to improve drought resistance and agronomic performance in soybean.
QTLs related to WUE play a role in the maintenance of growth rate and canopy height under drought stress
Our GWAS analysis identified also candidate genes for drought response of maximum absolute growth rate (AGRmax-Yr) and drought response of maximum canopy height (CH-Yr). These traits were determined using UAV-based methods [23], and quantify the growth of the soybean accessions in the drought treatment relative to the control treatment. AGRmax-Yr and CH-Yr are therefore comprehensive indicators of drought resistance, and are derived by the actions of multiple physiological processes, including the rate of carbon assimilation, photosynthesis, respiration, source and sink relations, nutrient balances, cell differentiation and elongation, as well as belowground processes [43,86,87]. A total of four significant associations were identified by GWAS for AGRmax-Yr in the SDS experiment, and another four significant associations for CH-Yr in the LDS experiment. Allelic differences of the significant SNPs explained a low to high drought-induced reduction for maximum absolute growth rate and for maximum canopy height in the EUCLEG accessions. The comprehensive nature of these growth related traits was reflected in the fact that some of the candidate regions for AGRmax-Yr and CH-Yr are located closely to the other candidate regions for CW and LSEN (on Gm05, Gm06 and Gm16; Table 2).
Table 2 SNPs significantly associated with the four traits, genes of interest and previously reported QTLs in proximity of the significant SNPs. ‘Locus ID’ consists of the trait abbreviation followed by the name of experiment and an ordinal number per trait and experiment. ‘Candidate region’ consists of the chromosome ID followed by the start position and end position of the candidate region (defined by the LD decay distance around the significant SNP). ‘SNP ID’ consists of the ID of the significant SNP followed by the major allele and the minor allele of the significant SNP. ‘P’ is -log10 of the P-value of the significant SNP. ‘R2’ is Phenotypic variance explained by the significant SNP. ‘Allele effect’ is the difference between median trait value of genotypes carrying the major allele and minor allele. ‘Gene of interest’ refers to the gene of interest as in the reference genome Wm82.a2.v1. 'Arabidopsis' is the homolog gene ID and gene name (in parenthesis) from Arabidopsis thaliana. ‘Annotations’ is the gene transcript of the Arabidopsis homolog or its encoding protein retrieved from [101], and the biological function (in parenthesis). ‘Reference’ contains the literature references to the biological functions stated in 'Annotation'. ‘Known QTL’ contains QTL information retrieved from [102] [59] and [58]. The QTLs names are adapted. The original names of QTLs can be found in Additional file 4.
Locus ID
|
Candidate region
|
SNP ID
|
P
|
R2
|
Allele effect
|
Gene of interest
|
Arabidopsis
|
Annotation
|
Reference
|
Known QTL
|
CW_SDS_1
|
Gm03_7027565_7317565
|
AX-93988673_G_A
|
4.3
|
0.05
|
-0.3
|
N/A
|
N/A
|
N/A
|
N/A
|
N/A
|
CW_SDS_2
|
Gm03_16874434_17164434
|
AX-93991327_G_C
|
4.1
|
0.02
|
0.23
|
N/A
|
N/A
|
N/A
|
N/A
|
N/A
|
CW_SDS_3
|
Gm03_45259915_45549915
|
AX-93918502_C_A
|
4.1
|
0.07
|
-0.4
|
Glyma.03G261500
|
AT2G46820 (PSI-P)
|
Photosystem I P subunit (photosynthesis)
|
[103,104]
|
PubC 2, PubC 3
|
|
|
|
|
|
|
Glyma.03G261400
|
AT4G01410 (N/A)
|
LEA hydroxyproline-rich glycoprotein (lateral root formation)
|
N/A
|
N/A
|
|
|
|
|
|
|
Glyma.03G258300
|
AT3G61830 (ARF18)
|
Auxin response factor 18 (regulates cell growth and seed weight)
|
[105]
|
N/A
|
CW_SDS_4
|
Gm05_28547428_28867428
|
AX-93921335_C_T
|
5.2
|
0.06
|
0.27
|
N/A
|
N/A
|
N/A
|
N/A
|
SCN 1
|
CW_SDS_5
|
Gm06_14605039_14955039
|
AX-93923547_T_C
|
5.8
|
0.04
|
-0.18
|
N/A
|
N/A
|
N/A
|
N/A
|
PH 1, PH 2
|
CW_SDS_6
|
Gm10_7752764_8022764
|
AX-94072904_A_G
|
4.2
|
0.05
|
-0.3
|
N/A
|
N/A
|
N/A
|
N/A
|
PH 3, PH 4
|
CW_SDS_7
|
Gm12_268639_738639
|
AX-94092624_A_G
|
9.9
|
0.08
|
-0.29
|
Glyma.12G009100
|
AT5G09680 (RLF)
|
Reduced lateral root formation (lateral root formation)
|
[106]
|
|
CW_SDS_8
|
Gm13_30347735_30657735
|
AX-93643614_G_A
|
4
|
0.04
|
0.29
|
N/A
|
N/A
|
N/A
|
N/A
|
WUE 1
|
CW_SDS_9a
|
Gm15_50341499_50871499
|
AX-94142375_T_C
|
4
|
0.06
|
-0.38
|
Glyma.15G269200
|
AT2G32840
(N/A)
|
Proline-rich family protein (cell wall integrity and root elongation under drought)
|
[107]
|
Cu 1, Cu 2
|
CW_SDS_9b
|
Gm15_50369101_50899101
|
AX-93945641_A_G
|
4
|
0.06
|
-0.38
|
CW_LDS_1
|
Gm02_8486923_8816923
|
AX-93976757_G_A
|
4.7
|
0.09
|
0.62
|
Glyma.02G094700
|
AT4G01470 (TIP1-3)
|
Tonoplast intrinsic protein 1;3 (transmembrane channels for water and small uncharged solutes)
|
[108]
|
N/A
|
|
|
|
|
|
|
Glyma.02G094900
|
AT3G55330 (PPL1)
|
PsbP-like protein 1 (reapir of PSII damage)
|
[109]
|
N/A
|
|
|
|
|
|
|
Glyma.02G097700
|
AT1G25480 (N/A)
|
Aluminium activated malate transporter family protein (stomatal opening)
|
[110]
|
N/A
|
CW_LDS_2
|
Gm06_50717337_51067337
|
AX-93628808_T_C
|
4.7
|
0.14
|
0.73
|
Glyma.06G321900
|
AT1G55670 (PSAG)
|
Photosystem I subunit G (satbility of photosystem I complex)
|
[111,112]
|
Seedset 1, Seeds
|
CW_LDS_3
|
Gm08_47511280_47831280
|
AX-93761082_G_T
|
6.1
|
0.14
|
1.04
|
Glyma.08G365700
|
AT1G67080 (ABA4)
|
Abscisic acid (aba)-deficient 4 (stress-induced ABA accumulation)
|
[113]
|
WUE 2, WUE 3
|
CW_LDS_4
|
Gm20_5309031_5709031
|
AX-93900908_A_C
|
4.6
|
0.14
|
0.76
|
Glyma.20G037100
|
AT5G12330 (LRP1)
|
Lateral root primordium (LRP) protein-related (root development)
|
[114]
|
N/A
|
CW_LDS_5
|
Gm20_43251405_43651405
|
AX-93910009_G_T
|
7.3
|
0.04
|
0.6
|
Glyma.20G197600
|
AT2G29460 (GSTU4)
|
Glutathione S-transferase tau 4 (oxidative stress response)
|
[115,116]
|
N/A
|
LSEN_SDS_1
|
Gm03_35362990_35652990
|
AX-93917720_G_A
|
5.1
|
0.15
|
-0.53
|
Glyma.03G138000
|
AT5G19450 (CPK8)
|
Calcium-dependent protein kinase 19 (ABA-mediated stomatal closure and ROS reduction)
|
[117]
|
P, Mg, Zn
|
LSEN_SDS_2
|
Gm15_49950789_50480789
|
AX-94142205_C_T
|
4.9
|
0.14
|
-0.43
|
Glyma.15G266500
|
AT1G79580 (SMB)
|
NAC (No Apical Meristem) domain transcriptional regulator superfamily protein (root cap development)
|
[118,119]
|
Cu 1, Cu 2
|
LSEN_SDS_3
|
Gm19_2913123_3603123
|
AX-93886487_C_T
|
6.8
|
<0.01
|
-0.01
|
Glyma.19G027700
|
AT5G39360 (EDL2)
|
EID1-like 2 (ABA signalling and stress respnse)
|
[77]
|
SIFC 1
|
LSEN_LDS_1
|
Gm02_36400063_36730063
|
AX-93981961_T_A
|
6.6
|
0.14
|
0.54
|
Glyma.02G192700
|
AT5G04870 (CPK1)
|
Calcium dependent protein kinase 1 (ROS reduction and proline accumulation)
|
[78]
|
N/A
|
|
|
|
|
|
|
Glyma.02G193000
|
AT5G06410
|
DNAJ heat shock N-terminal domain-containing protein (maintenance of seed shape and size at high ABA; AtDjA3)
|
[120]Salas-Muñoz et al., 2016
|
N/A
|
LSEN_LDS_2
|
Gm16_25479906_26001906
|
AX-93852357_C_A
|
5.6
|
0.11
|
-0.6
|
N/A
|
N/A
|
N/A
|
N/A
|
N/A
|
LSEN_LDS_3
|
Gm16_29164663_29686663
|
AX-93853705_G_C
|
4
|
<0.01
|
-0.02
|
N/A
|
N/A
|
N/A
|
N/A
|
PodShat 1, PodShat 2, qP16*
|
LSEN_LDS_4a
|
Gm16_30272259_30794259
|
AX-93854127_A_T
|
4
|
<0.01
|
-0.14
|
Glyma.16G145800
|
AT3G54890 (LHCA1)
|
Photosystem I light harvesting complex 1 chlorophyll a/b binding protein (electron transport in photosynthesis)
|
[81]
|
eChl_T**
|
LSEN_LDS_4b
|
Gm16_30395381_30917381
|
AX-93650691_A_C
|
4
|
0.01
|
0.08
|
LSEN_LDS_5a
|
Gm18_48141658_48501658
|
AX-94178903_C_G
|
4
|
0.02
|
-0.24
|
Glyma.18G202900, Glyma.18G203500
|
AT1G09950 (RAS1), AT3G51810 (EM1)
|
RESPONSE TO ABA AND SALT 1 (Salt and ABA sensitivity), Em-like protein GEA1 (stress induced ABA response through ABI5, ROS reduction)
|
[121–123]
|
SCN 2, Oil
|
LSEN_LDS_5b
|
Gm18_48144745_48504745
|
AX-94178905_G_A
|
4
|
0.02
|
-0.24
|
LSEN_LDS_5c
|
Gm18_48157192_48517192
|
AX-93655968_A_C
|
4
|
0.02
|
-0.24
|
LSEN_LDS_5d
|
Gm18_48205607_48565607
|
AX-93655972_C_T
|
4.1
|
0.02
|
-0.21
|
LSEN_LDS_5e
|
Gm18_48257716_48617716
|
AX-93881368_A_C
|
4.1
|
0.02
|
-0.21
|
LSEN_LDS_5f
|
Gm18_48277688_48637688
|
AX-93952219_T_C
|
4.7
|
0.02
|
-0.24
|
LSEN_LDS_5g
|
Gm18_48322720_48682720
|
AX-93881396_G_T
|
4.1
|
0.02
|
-0.21
|
LSEN_LDS_6
|
Gm19_44805252_45495252
|
AX-94195039_A_G
|
7.2
|
0.12
|
-0.49
|
Glyma.19G191700
|
AT5G03720 (HSFA3)
|
Heat shock transcription factor A3 (thermotolerance and drought tolerance in combination with DREB2A)
|
[84,85]
|
AMIN, Seedset 2, PH 5, Pods, LeafArea, LeafWidth, StemEnd, Internod, interbrach 1, DTM, DFTM, Lod, Nod, TotalSeed
|
AGRmax-Yr_SDS_1
|
Gm05_31841551_32161551
|
AX-93720346_T_C
|
7.7
|
0.1
|
0.06
|
Glyma.05G128000
|
AT1G29930 (LHCB1.3)
|
Light harvesting chlorophyll a/b binding protein 1 (stomatal movement, drought response)
|
[124]
|
N/A
|
AGRmax-Yr_SDS_2
|
Gm06_19076651_19426651
|
AX-93730252_G_T
|
5.4
|
0.09
|
0.05
|
Glyma.06G204100
|
AT3G29575 (AFP3)
|
ABI5 binding protein 3 (negative regulator of ABA signalling)
|
[90]
|
WUE 4, WUE 5, interbrach 2, DTF
|
|
|
|
|
|
|
Glyma.06G204400
|
AT1G74160
|
LONGIFOLIA protein (cellulose deposition)
|
[88,89]
|
N/A
|
AGRmax-Yr_SDS_3
|
Gm17_39250897_39530897
|
AX-93866910_A_C
|
7.1
|
<0.01
|
0.01
|
|
|
|
|
|
AGRmax-Yr_SDS_4
|
Gm18_54299774_54659774
|
AX-93883955_T_A
|
4.6
|
0.09
|
-0.04
|
Glyma.18G259700
|
AT5G15250 (FTSH6)
|
FtsH protease 6 (thylakoid membrane biogenesis, PSII repair mechanism)
|
[125]
|
SCN 3, Fe
|
CH-Yr_LDS_1
|
Gm06_18615861_18965861
|
AX-93730063_A_G
|
6.7
|
0.06
|
-0.03
|
Glyma.06G202200
|
AT1G74310 (HSP101)
|
Heat shock protein 101 (tolerance to heat sterss)
|
[92]
|
PubC 1, PubC 4
|
CH-Yr_LDS_2
|
Gm08_15065114_15385114
|
AX-93754190_C_T
|
5.1
|
0.16
|
0.06
|
Glyma.08G191300
|
AT4G33050 (EDA39)
|
Calmodulin-binding family protein (stomatal opening)
|
[126]
|
PUE, SIFC 2
|
CH-Yr_LDS_3
|
Gm09_7236144_7576144
|
AX-94060112_T_C
|
4.2
|
0.16
|
0.05
|
Glyma.09G071400
|
AT1G19150 (LHCA6)
|
Photosystem I light harvesting chlorophyll a/b binding protein 6 (photosynthesis)
|
[95]
|
WUE 6
|
CH-Yr_LDS_4
|
Gm16_32550790_33072790
|
AX-93855040_G_A
|
5.7
|
0.16
|
0.06
|
N/A
|
N/A
|
N/A
|
N/A
|
BSR
|
* QTL information from [59]
** QTL information from [58]
Furthermore, two previously reported QTLs for WUE [70] coincided with the candidate region AGRmax-Yr_SDS_2, on Gm06. This region contains Glyma.06G204400, whose Arabidopsis orthologue TRM4 or LONGIFOLIA is considered essential for cellulose deposition [88], whereas cellulose is essential for stem growth [89]. The same region contains Glyma.06G204100, annotated as Arabidopsis AFP3, which is a negative regulator of ABA response as it causes proteomic degradation of ABI5 [90], a core element in ABA response [91]. Furthermore, another candidate region CH-Yr_LDS_1, on Gm06 contains Glyma.06G202200, annotated as Heat-shock Protein 101 in Arabidopsis (HSP101). Constitutive expression of HSP101 provided heat tolerance and a high survival rate without any detrimental effect on growth in Arabidopsis plants [92]. These two regions on Gm06 can be of particular interest for further research and to modulate drought resistance in soybean.
SNPs in candidate regions CH-Yr-LDS_2, _3 and _4 explained a relatively large proportion of the phenotypic variation for CH-Yr (R2 = 0.16; Table 2). The candidate gene Glyma.08G191300 (Arabidopsis EDA39) on Gm08 belongs to the calmodulin-binding family protein. EDA39, in coordination with AtWRKY21, promotes stomatal opening by down-regulating the ABA response [93]. Another interesting candidate gene is Glyma.09G071400 (Arabidopsis LHCA6), present on Gm09. LHCA6 was shown to play a role in the normal functioning of the chloroplast [94], and its reduced expression caused a lower photosynthesis in Arabidopsis plants [95].
Considering the high level of stress imposed in the LDS treatment, and the results presented above for canopy wilting in the LDS experiment, it is obvious that within the EUCLEG collection some accessions have mechanisms to minimize transpiration through stomatal regulation, ensuring survival under long-term stress treatment. Therefore, our results suggest that a high water use efficiency could be an effective means to improve drought resistance of this European soybean collection, in particular for environments in which the chance of long periods of drought is high. Further, our results indicate that for an effective drought resistance under long-term drought stress, modulating the genes related to photosynthesis could be promising in soybean.