Final report for GNE16-130
Project Information
Cold hardiness studies are of great importance for the Northeastern US, where relatively low temperatures can lead to damage to perennial crops. Grapevines can suffer from cold injury in several ways, of which one of the most common is the injury to fruiting buds. The purpose of the current study was to identify genes responsible for aspects of cold hardiness, and to adapt models for predicting cold hardiness in grapevine buds during the winter. To evaluate this, buds were collected from vineyards in the Finger Lakes Wine Region in NY for three studies. A field collection was conducted to evaluate changes in bud cold hardiness due to field air temperatures and to adapt predictive model for the cold hardiness phenotype. Additionally, an assessment of acclimation to cold was performed in controlled environments during the fall. Cold hardiness was measured using differential thermal analysis (DTA). Buds were also collected for RNA extraction in the acclimation study. Field data was used to assess the fitness of models created based on previous year’s data. The proposed model appears to have a better fit to data from this region compared to the currently available model, especially in the acclimation phase. Controlled acclimation did not respond to constant temperatures, but required temperature fluctuations. These findings are being used to further improve predictive models for field data. RNA Seq data have not been analyzed to date.
Introduction:
Viticulture in the Northeast growing regions of the United States, particularly in the Finger Lakes region in the state of New York, often faces risks related to cold weather. The occurrence of extreme cold weather during the winter, or frost and freeze events during early fall and late spring are common challenges for grapevine growers. Although overall winter temperatures are expected to rise due to climate change, a more erratic climate is expected. The polar vortexes of 2013 and 2014 are some examples, as well as the severe cold after a period of warm weather in the spring of 2016. While all these events resulted in damage to the vines, the mechanisms that resulted in such damage were different in nature. The long-term prediction of these events is difficult, and management practices to avoid damage from cold weather events are not yet available or are impractical on a vineyard scale, but efforts are underway to find methods for reducing weather related damage in the field. However, predicting damage from a cold event is required to guide growers in the correct employment of damage mitigation efforts that may be developed in order to avoid wasting resources. Cold hardiness varies throughout the dormant season based primarily on temperature. A model has been established for use in the Western growing region of the US, however this model does not appear to be predictive in the Northeast and must be adapted to different climates. Additionally, the identification of genes responsible for cold resistance is also important for the development of markers for temperature tolerance, which can be used in breeding efforts for cultivars adapted to a more variable climate. The purpose of this project is to adjust the prediction model for the effect of temperature shifts on the changes of cold hardiness of grapevine buds in the Northeast region and to identify genes responsible for cold resistance. The adjusted model will generate a tool which can be used by growers facing potential damage from cold events in the decision making process for the use of mitigation efforts. The new model could also aid in the choice of vineyard management protocols in terms of pruning and training methods employed. This prediction tool could also result in improved yields by reducing the risks imposed by the environment, while also improving the growers’ quality of life. The identification of genes used as markers for cold resistance will improve the understanding of how plants are able to cope with cold temperatures and can be used by breeders in the development of cultivars that are better adapted to the Northeast region, further improving productivity in the vineyards. Additionally, identification of specific genetic pathways can provide information for the development of new products that can enhance cold resistance in grapevines.
Two main objectives were proposed:
- Generate or adapt a prediction model for cold hardiness for the Northeastern United States, using DTA data from (a) the Finger Lakes wine region in New York and (b) experiments conducted in controlled environment.
- Evaluate gene expression changes in controlled acclimation as a response to temperature shifts to identify specific genes that may serve as markers for cold hardiness aspects and enriched pathways related to cold hardiness.
Cooperators
- (Researcher)
Research
Objective 1a – Field (De-)Acclimation. Buds of V. vinifera ‘Cabernet Franc’, ‘Cabernet Sauvignon’, and ‘Riesling’, and V. interspecific hybrids ‘Cayuga White’ and ‘Marquette’ were collected for 29 weeks (4 Oct 2016 – 18 Apr 2017) from local vineyards. Eight buds were collected for differential thermal analysis (DTA) at each time point. Briefly, the buds are placed on thermoeletric modules in plates. These plates are placed in a programmable freezer that drops the temperature to -40 °C. The freezing of water produces two peaks due to the release of heat: a high temperature exotherm, when extracellular water freezes, which is not lethal; and the low temperature exotherm (LTE), when the intracellular water freezes, and correlates with bud death (Mills et al., 2006). A model created using previous year’s (2015-2016) ‘Cab. Sauvignon’, ‘Riesling’, and ‘Marquette’ data that uses a temperature parameter (σT, as described in Londo and Kovaleski, 2017), and chill and heat accumulation was evaluated using the 2016-2017 data set and compared to the current prediction model for cold hardiness (Ferguson et al., 2014). The current model uses linear rates of acclimation and deacclimation, and different temperature limits for acclimation and deacclimation, all varying with dormancy state. Additionally, temperature boundaries that are cultivar dependent (maximum and minimum hardiness) and an associated logistic term which provides asymptotic bounds to the predictions. The proposed model used σT, chill accumulation, and growing degree hours as predictors of cold hardiness.
Objective 1b – Controlled Acclimation. To evaluate acclimation in a controlled environment, buds were collected on 26 Oct. 2016, cut into single or two node cuttings, and placed in growth chambers at constant 2, 7, or 11 °C on 27 Oct. 2016. On 22 Nov. 2016, subsets from the 3 temperatures were moved into a chamber with a temperature cycle ranging from 1.5 °C –12.5 °C, resulting in an average temperature of 7 °C (7 °C Cycle). The genotypes used were: V. vinifera ‘Cabernet Franc’, ‘Cabernet Sauvignon’, ‘Riesling’, and ‘Sauvignon blanc’; V. riparia PI588711 and PI588562; and V. aestivalis PI483143. Sets of 5 buds were collected for DTA on 1 Nov., 3 Nov., 9 Nov., 16 Nov., 22 Nov., 30 Nov., 7 Dec., 21 Dec., 4 Jan., and 20 Jan. ANOVA was used to compare the effects of temperature and cycling on acclimation as measured by primary bud LTEs.
Objective 2 - RNA-Seq of Controlled Acclimation. From the experiment described in obj. 1b, 8 buds were collected at the same time points from ‘Cabernet Sauvignon’ for RNA extraction. RNA extraction was performed in three replicates of 2 buds using Sigma Spectrum kits (Sigma-Aldridge, St. Louis, MO) for each temperature regime (cycle vs. constant). Briefly, buds are ground under liquid nitrogen, and a lysate is added. Samples are run through a binding column, washed twice with a DNAse digestion step in between, and then eluted. RNA samples were submitted for 3’RNAseq library preparation and sequencing at the BRC Genomics Facility (Cornell University). Reads had the 12 basepair barcodes trimmed off, and terminal low quality bases were also trimmed to a final read length of 50 bps using Trimmomatic. Reads were aligned to the genome and trasncriptome using the program STAR. Read counts per gene were computed using HTSeq Count. Output files were combined in a matrix for differential expression analysis using DESeq2. LTE values from obj. 1b were used as the explaining variable for differential gene expression. Gene pathway analysis was conducted using the VitisPathways online platform (Osier, 2016), using genes with false discovery rate <0.10. The pathways were considered to be enriched when their permuted P-value was <0.10.
Field (De-)Acclimation (obj. 1a)– ‘Marquette’ acclimated faster than the other genotypes, and reached a greater maximum cold hardiness (Fig. 1). All the vinifera genotypes appeared to have the same trends in acclimation and deacclimation, with their cold hardiness being very similar with the exception of a few data points in midwinter for ‘Cab. Sauvignon’. Both hybrids started losing hardiness before the vinifera genotypes in early March. ‘Marquette’ appeared to deacclimate faster than all other genotypes. Northern distributed wild species have a tendency to acclimate and deacclimate faster. This is due to their necessity to rapidly respond to environmental cues: at the onset of winter, rapid decrease in temperatures; in the spring, rapid deacclimation will result in a longer growth season. ‘Marquette’, being a hybrid, presents both of these characteristics, whereas ‘Cayuga White’ only seemed to lose more hardiness in midwinter.
The new proposed model used chill accumulation, σT, and growing degree-days (GDD) for prediction of the LTE. For this model, GDDs were accumulated after 1000 chill units (NC model). The model had the best fits for ‘Riesling’ and ‘Marquette’, while for ‘Cab. Sauvignon’ it appeared to fit very poorly (Table 1). Prediction based on the current (Ferguson et al., 2014) model for ‘Riesling’ had a similar R2, although both root mean square error (RMSE) and bias were greater in magnitude. For ‘Cabernet Sauvignon’, the current model had a greater R2 and lower RMSE, but the bias was greater in magnitude. The current model has no available parameters for ‘Marquette’.
The current model appears to overestimate cold hardiness in both acclimation and deacclimation phases for both genotypes (Fig. 2). The proposed model likely has poor fit for ‘Cabernet Sauvignon’ due to the gross overestimation of deacclimation, resulting in a predicted cold hardiness that is much lower than the observed (Fig. 2A). For ‘Riesling’, however, the proposed model underestimated deacclimation (Fig. 2B). In both cases, predicted acclimation appears to have a good fit.
Although the fit of the model was good, both cases can be problematic. In the case of ‘Riesling’, overestimating the cold hardiness can result in no mitigation practices in the vineyards (e.g., lower pruning intensity), leading to lower crops. In the case of ‘Cabernet Sauvignon’, leaving more buds during pruning may result in excessive growth and overbearing. Therefore, better understanding of the response to temperatures is necessary, especially in the deacclimation phase.
Table 1. Fitness evaluation for current and proposed models of cold hardiness prediction for grapes using R2, root mean square error (RMSE), and bias.
|
Cultivar |
Modela |
R2 |
RMSE |
Biasb |
|
‘Marquette’ |
Proposed |
0.813 |
2.73 |
-1.74 |
|
‘Riesling’ |
Current |
0.786 |
2.41 |
-1.96 |
|
Proposed |
0.830 |
2.15 |
-0.97 |
|
|
‘Cabernet Sauvignon’ |
Current |
0.658 |
2.73 |
-2.32 |
|
Proposed |
0.325 |
3.84 |
2.03 |
|
|
a Current model – Ferguson et al. (2014) |
||||
|
b bias = average(error) |
||||
Figure 1. Cold hardiness of 3 V. vinifera (full symbols) and 2 Vitis hybrids (open symbols) in Geneva, NY, through the 2016-2017 winter.
Figure 2. Evaluation of current and proposed models for cold hardiness prediction of (A) 'Cabernet Sauvignon' and (B) 'Riesling'.
Controlled Acclimation (obj. 1b)– Contrary to expectations, results demonstrated that constant temperature treatments did not lead to further acclimation, thus field levels of cold hardiness were not enhanced. However, buds that were exposed to the 7 °C Cycle increased in cold hardiness significantly, regardless of genotype and temperature from which they were moved (Fig. 3). This is perhaps most interesting when looking at the response of the material that was moved from 2 °C into the 7 °C Cycle, as it was experiencing temperatures higher than before for most of the day, but this is still signaling towards acclimation to cold of the material. Further experiments with this temperature response are required to identify what are the effects of the average temperature and the amplitude on acclimation of the tissues.
Figure 3. Acclimation of 'Cabernet Sauvignon' buds in controlled environment at 3 constant temperatures further subjected to cycling temperatures.
RNA-Seq of Controlled Acclimation (obj. 2) - RNA-Seq analysis was performed with buds of 'Cabernet Sauvignon', which showed the most dramatic differences between cycling and constant temperatures in terms of LTE (obj. 1b) . 3879 genes were differentially expressed in cycling temperatures compared to constant, of which 1178 were downregulated and 2701 were upregulated. Pathways of interest that were enriched were ABA signaling, auxin biosynthesis and transport, gibberellin signaling, regulation of actin cytoskeleton, NAC and WRKY transcription factors.
Contrary to expected, in more acclimated buds, the ABA signaling pathway was mostly downregulated, while some genes that are inhibitors of ABA responses (KEG and AIP2) were upregulated. In the auxin biosynthesis, both the synthesis related genes (e.g. amine oxidases) and those responsible for reactivation of auxin pools bound to aminoacids (IAA-aminoacid hydrolases) were downregulated. All the genes that were differentially expressed in the auxin transport pathway were downregulated in the cycle compared to constant temperatures. In gibberellin signaling, LTCOR11 (a low temperature correlated protein) was highly upregulated in the cycle vs. constant. This is similar to what was reported by Schrader et al. (2004) in dormant vs. active cambium.
In the regulation of actin cytoskeleton, several genes responsible for myosin synthesis were downregulated in cycle vs. constant. Additionally, gelsolin and katanin genes, responsible for disassembly of actin filaments were downregulated. However, cofilin and SP1L2, responsible for organization of actin, were upregulated, as well as CHUP1 - responsible for anchoring of chloroplasts to actin filaments. This suggests that there is a reorganization of the cytoskeleton as the buds acquire cold hardiness. Reorganization of actin filaments was also reported by Plohovska et al. (2018) as a response to low temperatures in Arabidopsis roots.
Both NAC and WRKY transcription factors were generally downregulated in the cycle vs. constant. This is a similar response to that seen in Londo et al. (2018) in acclimation of leaves. Both NAC and WRKY gene families are associated with regulation of the transcriptome in response to biotic and abiotic stresses, suggesting that acclimation may not be a result of stress by temperature.
We expect that this project will yield an improved model for the prediction of cold hardiness based on weather data, which will result in more reliable predictions for the NE than that of the currently available model. Although more work is required, the proposed model appears to be a good predictor of bud cold hardiness for the Geneva region. With the model developed, farmers can be made aware of damage predictions, and can then plan for pruning accordingly (e.g. leaving more buds per vine based on bud damage predictions) in order to reduce the impacts of the cold damage on yields. If methods are developed in the future for increasing cold hardiness through cultural practices, this model will help in the decision-making. Additionally, further analysis of the RNA-Seq data may lead to genes used as markers in the future to develop new cultivars for regions where there are difficulties associated with low temperatures. Although major impacts have not yet occurred, we have found that temperature oscillations might be playing an important role in the development of the hardiness phenotype observed, which will be important in the planning of future controlled acclimation experiments, as well as in further development of the prediction model.
Ferguson, J.C., M.M. Moyer, L.J. Mills, G. Hoogenboom and M. Keller. 2014. Modeling Dormant Bud Cold Hardiness and Budbreak in 23 Vitis Genotypes Reveals Variation by Region of Origin. Am. J. Enol. Vitic. Doi:10.5344/ajev.2013.13098
Londo, J.P. and A.P. Kovaleski. 2017. Characterization of Wild North American Grapevine Cold Hardiness Using Differential Thermal Analysis. Am. J. Enol. Vitic. Doi:10.5344/ajev.2016.16090
Londo, J.P., A.P. Kovaleski and J.A. Lillis. 2018. Divergence in the transcriptional landscape between low temperature and freeze shock in cultivated grapevine (Vitis vinifera). Hort. Res. (in press).
Mills, L.J., J.C. Ferguson and M. Keller. 2006. Cold-hardiness evaluation of grapevine buds and cane tissues. Am. J. Enol. Vitic. 57:194-200.
Osier, M.V. 2016. VitisPathways: gene pathway analysis for V. vinifera. Vitis 55:129-133. Doi:10.5073/vitis.2016.55.129-133.
Plohovska, S.H., Y.A. Kasylenko and A.I. Yemets. 2018. Nitric oxide modulates actin filament organization in Arabidopsis thaliana primary root cells at low temperatures. Cell Bio. Int. doi: 10.1002/cbin.10931
Schrader, J., J. Moyle, R. Bhalerao, M. Hertzberg, J. Lundeberg, P. Nilsson and R.P. Bhalerao. 2004. Cambial meristem dormancy in trees involves extensive remodelling of the transcriptome. Plant J. doi: 10.1111/j.1365-313X.2004.02199.x
Education & Outreach Activities and Participation Summary
Participation Summary:
Two presentations were made with data collected with funds provided by NE SARE:
- 2017 New York State Agricultural Experiment Station Student Symposium
- 2017 American Society for Horticultural Sciences Annual Conference
At these meetings, the presentations were in regard to how grapevine buds sense and react to temperatures experienced in the acclimation and deacclimation phase during the winter. At these meetings, educators and service providers were present. At the NYSAES Student Symposium, many extension agents from the Finger Lakes Grape Program were present.
We will be participating in the B.E.V. (Business, Enology and Viticulture - http://www.bevny.org/program2/#viticulture ) NY conference between Feb 28 and March 2, 2018, which is an annual conference for grape and wine industry. This conference is heavily attended by farmers, educators and service providers. There we will present the advances in our temperature relations in grapevines, as well as an early version of the modeling for prediction of cold hardiness during the winter.
Data analysis is still ongoing, but we expect to produce at least 3 journal articles from the data collected with partial or full funding from NE SARE.
Project Outcomes
The final prediction model that will be generated from the data collected during this project will be helpful in assessing damage from extreme cold events in the winter, and cold weather following warm periods in the spring. If mitigation methods are found, this tool will be very important in providing information for the timing of application of these methods. This may result in reduction of cold temperature damage, producing economic benefits for growers. As a social benefit, we also expect an increase in quality of life, with a reduction of concerns over temperature damage.
From the sequencing data, we expect to learn more about the mechanisms underlying cold hardiness. We have learned that abscisic acid plays a major role in the deacclimation process, and genes related to it are likely markers for the increased cold hardiness phenotype. Advances in this area can, in the future, be used for breeding new cultivars that are resistant to cold without the need for multiple years of field testing.
As we attempted to study acclimation, we hypothesized that constant temperatures would promote acclimation. We based this on the predictions of the currently available model, which predicts acclimation in temperatures below ~15 °C. Once we began the project and noticed that there was no acclimation happening, we looked at what were the field temperatures and noticed that they were generally above that of the treatment with lowest temperature. Learning that temperature oscillations are important for acclimation is a big step in planning future experiments. Although we may be able to predict field data in our region without acknowledging this, understanding the mechanism underlying acclimation can lead to a more widespread applicability of the model generated. The new grant just received is to further study the acclimation process, and to improve our understanding of how average temperatures, as well as the amplitude of temperature fluctuations, result in increased cold hardiness.