1. Introduction

One of the distinctive features of blueberry relative to many other commercially-available fruits is that it accumulates a particularly broad range of compounds demonstrated to have health-protective properties, including fibers, folate, ascorbate, carotenoids, and a diverse and highly-concentrated array of flavonoids (flavanols, flavanones, flavonols, flavones, anthocyanins) and non-flavonoids (phenolic acids, stilbenes) ^[1,2,3,4]. This complex, concentrated and diversified bioactive phytochemical profile is responsible for blueberry’s selection as an intervention in clinical trials related to cognitive function, diabetes, cardiometabolic diseases, blood pressure and exercise performance ^[2,5,6,7].

Significant variations in levels of individual health-relevant flavonoid classes have been recorded for many popular highbush (V. corymbosum) and rabbiteye (V. virgatum) varieties ^[2,4], and recently the positive and negative influences of interspecific introgression on blueberry anthocyanin levels and acylation and glycosylation patterns were reported ^[8]. Chlorogenic acid (5-O-caffeoylquinic acid), one of the major phenolic acids in blueberry, has been linked to anti-obesity benefits, as well as a protective role against oxidative stress ^[9,10]. Elevated levels of chlorogenic acid have been associated with reduced incidence of Parkinson’s and Alzheimer’s diseases and diabetes ^[11,12]. Chlorogenic acid levels in blueberry are known to vary depending on genotype, but also stage of fruit maturity, year-to-year climatic differences, storage of fruit postharvest, and organic versus conventional cropping practices can all exert significant influence on levels of accumulation ^[2,13].

This study was designed to investigate the variations in chlorogenic acid (CA) content in a large number of blueberry commercial cultivars, breeding selections and breeding populations grown in North Carolina over two consecutive years. These blueberry genotypes are composed of different genetic backgrounds and ploidy levels (4×, 5×, 6×) and are currently part of the ongoing public NC State University blueberry breeding program.

---

2. Materials and methods

---

2.1. Plant material

A total of 16 commercial cultivars, 13 breeding selections (clones), and 23 breeding populations (F₁ cross progenies) of blueberry were evaluated for chlorogenic acid (CA) content in this study. The commercial cultivars included: Columbus, Ira, Montgomery, Onslow, Powderblue, Premier, Tifblue and Yadkin (rabbiteye blueberry, 6×), Arlen, Legacy, Lenoir, O'Neal, Ozarkblue, Pamlico, and Sampson (southern highbush blueberry (SHB), 4×), and Robeson (5×). The breeding selections were SHB, developed through NC State University’s breeding program and included: NC 3961, NC 4263, NC 4365, NC 4385, NC 4398, NC 4900, SA-10:135 (NC 4399), SA-13:75 (NC 4807), SA-4:2 (NC 4563), SHF2B1-20:21 (NC 5018), SHF2B1-21:3 (NC 5021), SHF2B1-25:25 (NC 5042), and SHF2B1-25:41 (NC 5043). The breeding populations, also developed through the same breeding program, consisted of varying numbers of progeny plants generated from the following crosses:“Arlen”דGeorgiagem”(74 plants), CHID2-14:73×open pollinated (OP) (28 plants), NC 1223דColumbus”(90 plants), NC 2873×OP (11 plants), NC 2898×G-615 (37 plants), NC 3147דLegacy”(six plants), NC 3147×NC 4562 (two plants), NC 3252×OP (six plants), NC 3958×OP (40 plants), NC 4165×OP (six plants), NC 4295דArlen”(34 plants), NC 4297דOzarkblue”(69 plants), NC 4299דOzarkblue”(164 plants), NC 4302דGeorgiagem”(two plants), NC 4302דSunshine Blue”(nine plants), NC 4562×NC 3476 (31 plants), NC 4562×NC 4179 (47 plants), NC 4562×NC 4361 (50 plants), NC 4812×OP (33 plants), NC 81-10-2דColumbus”(69 plants), NJ 89-158-24דColumbus”(37 plants), “Reveille”×NC 3476 (55 plants), and“Reveille”×NC 3920 (six plants). Detailed pedigree information (ploidy levels, origin, and % of parental species contribution) for the commercial cultivars, breeding selections and breeding populations were reported previously ^[8]. Blueberries had trickle irrigation, and no pesticides were applied because the genotypes were being screened to evaluate natural resistance to insects and fungi. Each genotype was replicated at minimum three times. Populations were composed of single plant progenies generated from crossing two heterozygous parents. Breeding selections and populations were created and selected for fruit firmness, yield, suitability for mechanical harvest and adaptation to the NC environment in the current blueberry breeding program at NC State University.

---

2.2. Fruit sample preparation

Blueberry field planting and establishment in the Piedmont Research Station, Salisbury, NC were detailed previously ^[8]. Fully ripe fruit were harvested at a uniform stage of maturity (when 75% of berries were fully ripe on any plant) and from comparable locations on each plant in the summer of 2010 and 2011. In both years, approximately 500 g of fruit were harvested all at once from each plant and packed on ice. Fruit were transported to the adjacent laboratory at the Plants for Human Health Institute (PHHI), NC Research Campus (NCRC), Kannapolis, NC. After transport, fruit were frozen in liquid nitrogen and stored at −80°C, then lyophilized using a freeze dryer (VirTis 24Dx48; SP Scientific, Stone Ridge, NY) with a temperature controlled chamber for samples. Freeze-dried fruit were then stored at −80°C until extraction. Weights of samples were taken before and after lyophilization to estimate dry matter content (DM percent) in the fruit.

---

2.3. Extraction and HPLC-DAD analysis

For each plant, lyophilized tissue (2.5-3.0 g/plant) was extracted with 30 mL of 0.3% acetic acid in MeOH:H₂O (70:30, v:v). Ground samples were transferred into 50-mL tubes, vortexed, and then centrifuged for 15 min at 3400 g_n. The supernatants were decanted into 100-mL volumetric flasks, extraction was repeated two more times and the final volume was brought to 100-mL. Extract was filtered into 2-mL amber vials using 0.2-µm polytetrafluoroethylene syringe filters (Fisher Scientific, Pittsburg, PA). A 10 µL aliquot was injected into a 1200 high-performance liquid chromatography (HPLC) system (Agilent Technologies Inc., Santa Clara, CA) equipped with an ultra violet-visible spectrophotometry (UV-VS) diode array detector (DAD), controlled-temperature autosampler (4°C), and column compartment (30°C). Chemstation software (Agilent Technologies Inc., Santa Clara, CA) was used as the system run controller and for data processing. Chlorogenic acid separation and quantification was performed using a reversed-phase column (250 mm×4.6 mm×5 μm (Supelcosil LC-18; Supelco, Bellefonte, PA)). The mobile phase consisted of 5% formic acid in H₂O (A) and 100% methanol (B). A step gradient of 10%, 15%, 20%, 25%, 30%, 60%, 60%, 10%, and 10% of solvent B at 0, 5, 15, 20, 25, 35, 36, 37 and 40 min, respectively, at a constant flow rate of 1 mL/min applied for samples and CA commercial standard (Sigma Aldrich, St. Louis, MO). CA in samples was quantified using a standard curve with peak areas recorded at 325 nm associated with concentrations of 1.0, 0.5, 0.25, 0.125, 0.063, 0.031, 0.016 and 0.008 mg/mL in 100% MeOH. To facilitate comparison with published data, CA concentrations in this study were converted back to a frozen fruit basis and presented as mg/100 g frozen fruit.

---

2.4. Statistical analysis

Analysis of variance (ANOVA) was performed using the general linear model procedure (PROC GLM) with random effects using SAS software (version 9.4; SAS Institute, Cary, NC, 2012). Mean separations were performed using LSMEANS statement and Tukey’s honest significant test. For the commercial cultivars or breeding selections (replicated clones) the statistical model used was y_ijkl=µ + G_i + Y_j + P_k + R(Y)_(j)l + GY_ij + ε_(ijkl); where y=response from the experimental unit, µ=overall mean, G=genotype (cultivar or clone), Y=year, P=plant, R(Y)=replication within year, GY=genotype×year interaction effect, ε=experimental random error. For the breeding populations (with single plant genotypes within a population) the model used was y_ijk=µ + G_i + Y_j + P(G)_(i)k + GY_ij + ε_(ijk); where y=response from the experimental unit, µ=overall mean, G=cross, Y=year, P(G)=plant within cross, GY=cross×year interaction effect, ε=random error. The PROC MEANS statement was used to compute genotype mean, standard deviation, and range within the 2 years (2010 and 2011) separately.

---

3. Results and Discussion

---

3.1. Chlorogenic acid concentration

In this study, chlorogenic acid (CA) was assayed as the major phenolic acid in blueberry fruit, where it was eluted at 13.3 minutes after injection of samples into the HPLC-DAD (Figure 1). This was in agreement with previous reports which indicated that CA was the major phenolic acid found in blueberry ^[1,2,14]. The CA concentrations observed in this study for the commercial cultivars were comparable and in agreement with previous reports ^[2,14,15]. Blueberry ploidy level, number of plants, and descriptive statistics for the chlorogenic acid concentrations in the cultivars, clones, and breeding populations are presented in Table 1. Consistent and comparable CA content was observed from year to year, with minor deviations in the case of the breeding populations. While 4×cultivars showed an average of 49 mg/100 g FW, the CA average for the 6×cultivars was 102 mg/100 g FW (Table 2). Yadkin, Powderblue and Ira contained the highest CA concentrations among the cultivars, at 139, 132, and 126 mg/100 g FW, respectively. These three cultivars differed significantly from the 4×cultivars and some of the 6×cultivars; namely Montgomery, Onslow, and Premier. The significant variation for CA among blueberry genotypes that we observed in this study was consistent with previous reports ^{[16,17,18,19,20]}. The clones and breeding populations, with a broader gene pool, contained a mix of significantly higher or much lower CA content depending on the genetic background. The CA content ranged from 33-107 mg/ 100 g in breeding selections (Table 2). The clones NC 4398, SA-4:2, and SHF2B1-20:21 contained the highest CA concentrations among breeding selections (i.e., 77, 107, and 107 mg/100 g FW, respectively). With the expansion of the genetic background to include a larger number of species compared to the breeding selections, the CA content was significantly higher in some crosses, but ranged from 46-137 mg/100g FW in 4×crosses. For 6×crosses, the CA mean for certain crosses (i.e., NC 1223דColumbus”which contained 50% germplasm from the cultivated species V. virgatum and 50% from the wild V. virgatum) was significantly higher (156 mg/100 g), with some plants within this cross reaching 258 mg/100 g over two years of the evaluation. Detailed pedigree information for all plants evaluated in this study was previously reported ^[8]. However, when Columbus (6×) was crossed with two different breeding lines (NC 81-10-2 and NJ 89-158-24), no improvement in CA occurred.

---

3.2. Analysis of variance (ANOVA)

ANOVArevealed that genotype was the main source of variation for CA in the commercial cultivars, breeding selections (clones), and breeding populations (Table 3). While there was a year effect on the CA concentration, it was small compared to the genotype effect. Other experimental factors including individual plants, replications, and genotype×year interactions did not show any significant effect on CA accumulation in the commercial cultivars. A similar trend was observed with the breeding selections genotypes (Table 3). The genotypic variation (calculated as a percentage of the MS associated with genotype over the total MS for the statistical model) constituted 64% and 50% of the total variation in CA in the cultivars and clones, respectively. In the breeding populations, since the F₁ heterozygous plants were segregating for all of the polyphenolic phytochemicals measured in this study including CA, significant variation was observed within populations, and in the genotype×year interaction. However, the genotype effect constituted 72% of the total variation observed over the two years of this study, which can have significant implications for blueberry breeding efforts.

---

3.3. CA correlations

Two types of correlations were examined with CA in this study; correlation between CA and ploidy level, and correlation between CA and anthocyanin (ANC) content. Anthocyanin ranges in these blueberry genotypes were previously reported ^[8]. A strong positive correlation was observed between ploidy level and CA content (r=0.75) in the commercial cultivars. In the segregating populations, the correlation was not as evident.

A significant moderate correlation was observed between CA content in the blueberry materials used in this study, and their ANC content. This is particularly true with all ANC classes except for the acylated anthocyanins (Table 4). CA showed positive and significant correlation with total anthocyanins (r=0.5) in the clones and breeding populations but was lower (r=0.3) in the commercial cultivars. This positive correlation with ANC was expected, since CA and ANC share the same biosynthetic pathway, as derivatives from L-phenylalanine (Figure 2). These correlation trends are of particular interest in that it may facilitate selection for improved blueberry phytochemistry, using CA as a phenotypic marker that is easy, fast, and inexpensive to measure using traditional HPLC methods. When selecting for higher CA, ANC may indirectly be elevated at the same time. This can play a significant role in blueberry breeding, since quantitatively assaying ANC with a complex profile is prohibitively expensive and time-intensive to perform on the large number of plants required for breeding programs. Data obtained in this study showed that significant genetic variability exists among blueberry breeding material established in NC State University’s breeding program that can be used effectively to improve blueberry plants for CA and at the same time, indirectly improve ANC concentration. This is particularly important since breeding blueberry can be a long and demanding process that can take 10 to 20 years from the original cross to cultivar release ^[21]. Expanding the gene pool for blueberry resulted in enhanced CA content that may contribute to the blueberry phytochemical value and human health benefits.

---

Acknowledgments

The authors are grateful for funding provided by the UNC General Administration Special Allocation for Collaborative Research at the NCRC. The authors also wish to thank the North Carolina Department of Agriculture, Piedmont Research Station in Salisbury, North Carolina for field maintenance and technical help with experiments.

---

Conflict of interest

The authors declare no conflict of interest in this publication.

---