Accumulation of Carboxylate and Aromatic Fluorophores by a Pest-Resistant Sweet Sorghum [Sorghum bicolor (L.) Moench] Genotype.

The sugary juice from sweet sorghum [Sorghum bicolor (L.) Moench] stalks can be used to produce edible syrup, biofuels, or bio-based chemical feedstock. The current cultivars are highly susceptible to damage from sugarcane aphids [Melanaphis sacchari (Zehntner)], but development of new cultivars is hindered by a lack of rapid analytical methods to screen for juice quality traits. The mechanism of aphid resistance/tolerance is also largely unknown, though the importance of defense phytochemicals has been suggested. The purpose of this study was to develop low-cost methods sensitive to fluorescent fingerprints in sweet sorghum juice, which is a complex mixture of saccharides, carboxylates, polyphenols, and metal ions. Of primary juice components, tryptophan and trans-aconitic acid were the highest intensity contributors to the overall fluorescence and UV/visible absorbance, respectively, while tyrosine and polyphenols contributed to a less extent. In a test of 24 sweet sorghum cultivars, tryptophan and tyrosine contents were the highest in the aphid-susceptible hybrid N109A x Chinese, while sucrose, trans-aconitic acid, and polyphenols were the highest in the resistant line No. 5 Gambela. This suggests that the accumulation of carboxylate (trans-aconitic acid) and polyphenolic secondary products in No. 5 Gambela may contribute to its aphid resistance, thus allowing it to maintain sucrose production. Rapid detection of these chemical signatures could be used to prescreen the breeding material for potential resistance and juice quality traits, without analytical separation required for metabolomics.

Introduction

Sweet sorghum [Sorghum bicolor (L.) Moench] is traditionally cultivated for the edible syrup made from the sugary juice that is extracted from its stalks. This juice could also be used to produce chemical feedstock for the production of biofuels and other bio-based products. However, the development of improved sweet sorghum cultivars has been given little attention recently. The composition and quality of the juice are particularly important, but there are few methods available to rapidly screen breeding material for such traits. In addition, the sugarcane aphid [Melanaphis sacchari (Zehntner)] has recently emerged as a significant pest of sorghum in the United States. All of the commonly grown sweet cultivars are susceptible and few chemical control options exist. Thus, resistance to this pest is also of critical importance in the development of new sweet sorghum cultivars.

Low-cost spectrophotometric methods could help to expedite plant breeding by rapidly screening for various chemical traits. Fluorescence is a promising, rapid, low-cost, yet accurate “omics” fingerprinting technique for target analytes to address the needs for “advanced analytics for managing the food and agricultural system”.[1] Naturally occurring fluorescent molecules are widely used as native tracers to monitor food[2] and environmental[3] quality, without adding fluorescent dyes.[4] Examples of natural fluorophores include aromatic amino acids, riboflavin, chlorophyll, and vitamin A.[2,5] These aromatic molecules have conjugated double bonds that provide high molar absorption coefficient and quantum yield.[6] However, samples of agricultural, environmental, and food origins contain a complex mixture of fluorophores having different characteristic peak positions and intensities.[2,5] As a result, fluorescence peaks are often visually assigned to “bulk structures” based on literature comparisons. Examples of commonly observed bulk structures include “humic-like” and “microbial decomposition product” in environmental samples[5,7] and “fluorescent oxidation products” and “Maillard reaction products” for food samples.[2] In many cases, such peak assignments to “bulk structures” are made without examining the authentic standards for additional possible contributing structures.

We previously observed tryptophan-like, tyrosine-like, and “aromatic” fluorophores in the stalk juice of sweet sorghum.[8] The “aromatic” bulk structure was attributable to electron-donating polyphenols.[9] It was observed that sweet sorghum cultivars with high polyphenolic contents in the stalk juice sustained less damage from sugarcane aphid.[10] The aphid-resistant control, No. 5 Gambela (PI 257599 in Table S1 of Supporting Information),[18] was enriched with polyphenols and also contained a high amount of another secondary product, trans-aconitic acid.[10] Collectively, high polyphenol and trans-aconitic acid concentrations in phloem sap could serve as defense phytochemicals or phytohormonal signals upon leaf damage by aphids, as suggested in the literature.[11−14] There are different categories of resistance against aphids: antibiosis, antixenosis, and tolerance.[15] Redox-active (polyphenols) and metal chelating (trans-aconitic acid) secondary products could play a role in all categories as the allelochemical (antibiosis), deterrent (antixenosis), and antioxidant (tolerance).[16] Our present study was designed to examine the direct relationships between these target analytes (polyphenols and trans-aconitic acid) and host plant resistance. To achieve this goal with a field-deployable[17] spectrophotometric measurement, we first re-examined the bulk “aromatic” fingerprint by developing sensitive fluorescence techniques validated by authentic standards.[9,19] Results were then used to understand the relationships between the secondary products and response to sugarcane aphids, where the cultivar No. 5 Gambela was used as the aphid-resistant control.

Results and Discussion

Comparison of EEM and Synchronous Spectra against Authentic Standards

Figure presents solid-phase reflective fluorescence spectra for authentic standards: as-received pure chemical powders of model amino acids, carboxylate, and model polyphenol. These structures were previously observed in sweet sorghum juice.[8] The full excitation–emission mode (EEM) scan (left panels in Figure ) is expected to provide detailed spectral features, whereas the synchronous mode (right panels employing 60 nm delay) enables a shorter time of analysis affording simpler spectra. Depending on the fluorescence intensity of a molecule, slit width was varied from 1, 2.5, to 5 nm, while keeping all other parameters constant. Slit (bandpass) width affects both the signal-to-noise ratio and resolution of collected spectra. Larger slit width decreases the noise, while making minor peaks less visible, and vice versa. Tryptophan had the highest fluorescence intensity (≈8000 maximum intensity in Figure e) at the narrowest EEM slit widths (1 nm). Oppositely, quercetin required the widest slit widths (5 nm) to achieve sufficiently high fluorescence intensity (≈500 in Figure h). In the EEM mode (Figure a–d), the characteristic peak position shifted toward higher excitation and emission wavelengths in the following order: amino acids (tryptophan and tyrosine), carboxylate (trans-aconitic acid), to a model polyphenol (quercetin). Such peak shifts (toward higher EEM wavelengths) are known to occur proportionally to the degree of conjugation and aromaticity.[2,6] Similarly, the primary synchronous peak (having maximum intensity) of authentic standards shifted toward longer wavelengths form tyrosine, tryptophan, trans-aconitic acid, to quercetin (Figure e–h). In conclusion, synchronous scan affords a snapshot of characteristic peaks with reduced scan time compared to EEM. In addition, subsequent statistical separation of peaks (by PARAFAC to resolve overlapping EEM peaks) is not necessary for the synchronous mode. Both EEM and synchronous modes showed the following structure–wavelength relationships of the characteristic peak positions: aromatic amine < conjugated carboxylate < polyphenol.

Figure 1

Solid-state reflective fluorescence EEM (left) and synchronous (60 nm delay, right) spectra of amino acids (a,b,e,f), carboxylate (c,g), and polyphenolic (d,h) authentic standards.

Based on Figure , tryptophan is expected to be the highest intensity fluorophore in juice. A sample containing the highest amino acids (Section ) and low concentrations of other fluorophores (trans-aconitic acid and phenolics) was selected to identify the protein peaks in Figure : N109A x Chinese planted in April. The synchronous spectrum (Figure c) indicates that tryptophan is the primary fluorophore in this juice sample. Tyrosine has a minor (lower intensity) contribution relative to tryptophan, and the peak of trans-aconitic acid is outside the wavelength range of juice.

Figure 2

Solution-phase penetrative fluorescence EEM (a,b,d) and synchronous (c, 90 nm delay) spectra of diluted (20-fold) juice in protein-rich cultivar (N109A × Chinese/April) with 2.5 nm EEM slits. This juice sample has low trans-aconitic acid content of 3.7 mM. In (b,d), intensities of amino acids were normalized to match juice. Synchronous spectra of juice samples in penetrative (with 20-fold dilution) and reflective (without dilution) modes are comparable (Section SII, Supporting Information).

In EEM, there are two peaks at Em wavelengths of 310 and 345 nm (Figure a). For a direct comparison of EEM peak positions, a 3D plot of juice (red mesh) is compared with amino acid standards (clear mesh) in Figure b,d. The 3D plots indicate the contributions of both tryptophan and tyrosine. However, because peaks of two amino acids overlap, their contributions cannot be distinguished in the 3D plots.

Next, a juice sample with high trans-aconitic acid content (24 mM without dilution) was investigated in Figure : No. 5 Gambela (May planting). A high emission peak at 430 nm, which was absent in a low trans-aconitic acid cultivar (Figure a) is visible in No. 5 Gambela (Figure a). In Figure b, emission spectra (red lines) were extracted from the peaks of the EEM contour plot (at 355 and 430 nm emission wavelengths of Figure a). trans-Aconitic acid (blue line in Figure b) is the primary contributor to the 430 nm emission spectrum of No. 5 Gambela (dashed red line in Figure b), where the contributions of amino acids (black and green lines in Figure b) were negligible.

Figure 3

New peak (430 nm Em) is visible in No. 5 Gambela (May planting) cultivar with high (24 mM) trans-aconitic acid content. Emission spectra of No. 5 Gambela indicates the contribution of trans-aconitic acid (blue line) on the new (430 nm Em) peak (b). Contributions of tryptophan (green) and tyrosine (black) to the peak (430 nm Em) are minor in both EEM (b) and synchronous (c) modes. Instrumental parameters in (c): 2.5 nm slits for juice and trans-aconitic acid, 1 nm slits for tyrosine and tryptophan, and 60 nm delay time for all authentic standards.

In the synchronous spectra of No. 5 Gambela, longer delay wavelength (90 nm, dashed red line in Figure c) enhanced the secondary peak at 300–360 nm, relative to a shorter delay (60 nm, solid red line). Two overlapping peaks in the synchronous spectrum (red lines in Figure c) were resolved into two independent peaks in the emission spectra extracted from EEM (red lines in Figure b). The secondary peak is enhanced in No. 5 Gambela, relative to the cultivar low in trans-aconitic acid (Figure ). Subsequent sections will employ the longer (90 nm) delay time to investigate nonprotein fluorophores observable in the secondary peak. Overall, EEM is more sensitive to low intensity peaks (e.g., tyrosine in Figure ). Synchronous mode offered a higher peak resolution than the emission spectra extracted from EEM (Figure b,c).

Cultivar Effects on the Chromophores Detectable by Fluorescence and UV/Visible Spectrophotometry

As observed in the previous section, visual inspection alone does not distinguish different fluorophores in a complex mixture of chromophores composing the sweet sorghum juice. Visual comparison is particularly problematic for structures with overlapping wavelength ranges, such as tryptophan and tyrosine. Therefore, the following statistical analyses were undertaken. First, overlapping EEM peaks were resolved by PARAFAC. Then, fluorophore concentrations (estimated as the PARAFAC contributions) were evaluated for cultivar effects (p < 0.05), where three field replicates were employed to test for reproducibility. Figure a–c presents EEM/PARAFAC fingerprints for diluted (20-fold) juice (2.5 nm EEM slits). Three fingerprints were attributed to tryptophan-like (Figure a), tyrosine-like (Figure b), and aromatic structure (Figure c), similar to juice samples from previous years (2015–2016).[8,19] The subsequent section will explore the contribution of trans-aconitic acid and polyphenols on the “aromatic” PARAFAC fingerprint. The concentration of each fluorophore was estimated as (1) PARAFAC contributions (tryptophan, tyrosine, and aromatic) and (2) integrated areas of large (250–300 nm primarily attributable to amino acids) and small (300–350 nm) peaks obtained in the synchronous mode with 90 nm delay. Cultivar (Table S1 of Supporting Information) and planting month effects were separately evaluated for these two techniques in Table .

Figure 4

Three EEM/PARAFAC fingerprints obtained for diluted (20-fold, a–c) juice in the penetration mode, and bagasse powder in the reflective mode (d–f) from 24 sweet sorghum cultivars in 2017. All spectra were collected with 2.5 nm EEM slits.

Table 1

Fluorescence Properties of Sweet Sorghum Juice and Bagasse Harvested at the Hard-Dough Stage in 2017a

							significant (<0.05) p value
variable	n	mean	SD	min	max	non-zero	cultivar	plantingb
Synchronous Peak Areas of Juice (90 nm Delay, 20-Fold Dilution) and Bagasse Powder (60 nm Delay)
juice (250-300 nm peak)	174	20147	7691	541	82177	174	0.008 (N109A x Chinese > N111A x Isidomba, Isidomba)	<0.00001 (↓-)
juice (300–350 nm)	173	597	859	83	9958	173	0.03 (No. 5 Gambela > all except N111A x Dale, N109A x Dale, Atlas, N111A x Isidomba, Isidomba, N110A x Dale, Dale)*	0.007 (↑)
bagasse (250–600 nm)	202	44111	18376	5055	94193	202		0.01 (↑-)
Absolute and % Contribution of EEM–PARAFAC Fingerprints for Juice (20-Fold Dilution) and Bagasse
juice 1 (tryptophan-like)	175	5814	2061	1467	24 500	175	0.002 (N109A x Chinese > N111A x Isidomba, Isidomba)	<0.00001 (↓-)
juice 2 (tyrosine-like)	175	2379	768	1204	8485	175	0.007 (N109A x Chinese > N109B, N111A x N98)	0.0002 (-↑)
juice 3 (aromatic)	175	886	482	276	3434	175	<0.00001 (No. 5 Gambela > all others;* Dale > N111B, N111A x N98)	<0.00001 (↑↑)
juice sumc	175	9079	2713	3342	35 876	175	0.007 (N109A x Chinese > N111A x Isidomba)	0.001 (↑-)
juice % 1	175	64	8	44	78	175	<0.00001 (Dale, Isidomba, No. 5 Gambelad < all others)	<0.0001 (↓↓)
juice % 2	175	27	6	13	48	175	<0.00001 (Isidombae, Dalef > all others; N111A x Dale > No. 5 Gambela, N109B)	<0.00001 (↑↑)
juice % 3	175	10	5	3	36	175	<0.00001 (No. 5 Gambela > all others;* Isidomba > N111A x Chinese, N111B, N111A x N98, N109A x Chinese; Dale > N111A x Chinese, N111B, N109A x Chinese)	<0.00001 (↑↑)
bagasse 1 (mid-EEM)	189	13 798	6547	1457	38 948	189		0.01 (↑-)
bagasse 2 (aromatic)	189	2220	7225	0	75 882	145
bagasse 3 (least aromatic)	189	2961	5770	0	75 882	174		0.04g
bagasse % 1	189	83	15	26	100	189	0.04g
bagasse % 2	189	6	8	0	61	145	0.05g
bagasse % 3	189	11	8	0	47	174		0.03 (↑)

Only Significant P-values (<0.05) are presented for the cultivar (24 varieties in Table S1) and planting month (April, May, June) effects with post hoc Tukey’s HSD test. For each dependent variable (synchronous peak areas and EEM/PARAFAC contributions), sample number (n), mean, standard deviation (SD), minimum and maximum values, and number of non-zero values are provided. Cultivar × planting Interaction was observed only for normalized (%) contribution of Juice PARAFAC factor 1 in Figure a (p = 0.05)

Arrows indicate time trend (p < 0.05 by post hoc Tukey): increase (↑), decrease (↓), or no change (−) from April to May (first arrow), and from May to June (second arrow); one arrow indicates significant difference only between April and June.

Sum of absolute contributions 1–3.

For No. 5 Gambela, except N110A x Chinese, N111A x Dale, N109A xDale, N111A x Isidomba, Isidomba, N111A x Atlas, N110A x Dale, Dale, N110A x Atlas, and N110A x Isidomba.

For Isidomba, except N111A x Dale, N110A x Dale, and Dale.

For Dale, except N111A x Dale, N109A xDale, N111A x Isidomba, Isidomba, N110A x Dale, N110A x Atlas, and N110A x Isidomba.

No significant difference by Tukey. *Maximum in No. 5 Gambela (p < 0.05 by Tukey).

Table shows significant (p < 0.05) cultivar and planting month effects on the primary (250–300 nm) and secondary (300–350 nm) synchronous peaks of diluted (20-fold) juice with 90 nm delay (dashed red lines in Figures c, and 3c). The primary synchronous peak (250–300 nm) attributable to amino acids (Figure ) showed higher contribution on N109A x Chinese than N111A x Isidomba or Isidomba and decreased from April to May planting. The same cultivar and planting effects were observed for the tryptophan-like PARAFAC fingerprint (Figure a) with a lower p value of cultivar effects (0.002) than the synchronous method (p = 0.008). The minor synchronous peak of juice (300–350 nm) contributed primarily to No. 5 Gambela and increased from April to June planting; the same trend was observed for the “aromatic” PARAFAC fingerprint of juice (Figure c) with a lower p value of cultivar effects [<0.00001 for both absolute and % contributions of juice 3 (aromatic)]. Tyrosine-like maximized in N109A x Chinese, similarly to the tryptophan-like PARAFAC fingerprint. Collectively, EEM–PARAFAC was a more sensitive method to classify cultivars (lower p values) and detect fluorophore (tyrosine was distinguished from tryptophan) than the synchronous method. The sum of absolute contributions for all PARAFAC factors (1–3) showed a similar cultivar dependence as tryptophan, which is the highest intensity fluorophore in juice (Figure ). When each fingerprint was normalized to the sum, the cultivar trend indicated the accumulation of the aromatic fingerprint (juice %3) in No. 5 Gambela and among the lowest contents of protein fingerprints (%1 and %2) in No. 5 Gambela.

Figure d–f presents PARAFAC fingerprints for bagasse powder. While mid-EEM (Figure d) and least aromatic (Figure f) fingerprints were previously observed in dry bagasse (2015–2016 samples with 2.5 nm slits)[8] and leaf (2015 samples with 5 nm slits)[10] powder, the 2017 bagasse sample showed a new high aromaticity fingerprint (Figure e). The emission wavelength of this high aromaticity fingerprint (≈540 nm in Figure e) reached that of riboflavin,[2] as observed in the 2015–2016 bagasse samples at higher excitation wavelength (430 nm),[8] as opposed in 270 nm in Figure e. Therefore, a lower excitation wavelength was required to achieve the high emission wavelength (540 nm) in 2017 than previous years (2015–2016). Riboflavin is a highly fluorescent molecule with excitation maxima at 270, 370, and 450 nm and an emission maximum at 525–531 nm.[2] Riboflavin is a strong reducing agent and plays a role in the light-induced oxidation of food (e.g., dairy) products.[2] In contrast to juice, bagasse fluorescence did not differentiate cultivars by Tukey’s honestly significant difference (HSD) in Table . Representative synchronous spectra of bagasse powder are provided in Section SIII of the Supporting Information. The synchronous peak of bagasse (obtained at 250–600 nm with 60 nm delay) increased from April to May planting, similar to the mid-EEM fingerprint of bagasse PARAFAC (Figure d). Normalized (%) contribution of the least aromatic fingerprint (Figure f) increased from April to June.

To understand the relative contributions of trans-aconitic acid and polyphenols on the “aromatic” PARAFAC fingerprint, Figure presents UV/visible spectra of trans-aconitic acid, sugars (sucrose, glucose, and fructose), and carboxylates (oxalate, citrate, and trans-aconitate) previously observed in sweet sorghum juice.[8] Because of high absorbance, trans-aconitic acid was diluted to 0.1 g/L, while concentrations of other authentic standards were 0.5 g/L in Figure a. Compared to sugars and other carboxylates, trans-aconitic acid showed higher absorbance at longer wavelengths (Figure a). In Figure b, trans-aconitic acid was directly compared with representative juice containing 5–209 mg/L trans-aconitic acid after dilution (20-fold). Figure indicates an enhanced 255 nm peak as a function of the trans-aconitic acid content in juice, from N111B/April (5 mg/L trans-aconitic acid), N110x Atlas/April (24 mg/L), No. 5 Gambela/April (174 mg/L), to No. 5 Gambela/May (209 mg/L). However, higher absorbance (across the spectral range) was observed in No. 5 Gambela/May (209 mg/L trans-aconitic acid) and No. 5 Gambela/April (174 mg/L) than 500 mg/L trans-aconitic acid standard, indicating contributions from additional chromophores, for example, polyphenols, in juice. In addition, a higher wavelength peak at 310–350 nm (and baseline shifts toward 400 nm) in those juice samples (red and green lines in Figure b) are not observable in trans-aconitic acid (cyan line in Figure b). Those observations indicate that polyphenolic structures comprise chromophores in sweet sorghum juice, in addition to trans-aconitic acid, as observed previously.[9]

Figure 5

UV/visible spectra of authentic standards (a) and representative juice samples with varying trans-aconitic acid contents (b). In (a), concentrations are 0.5 g/L, except for trans-aconitic acid (0.1 g/L). In (b), trans-aconitic acid concentrations (of 1–20 diluted samples) are given in parentheses. All spectra are color-coded, and blank-subtracted.

Cultivar Effects on the Juice Composition

Table shows the effects of cultivar and planting month on the concentrations of trans-aconitic acid and other organic carbon products. trans-Aconitic acid concentration was significantly higher in No. 5 Gambela (20.6 ± 1.4 mM as mean ± standard error across planting months) than all other cultivars (except N109A x Dale with the second highest concentration of 13.8 ± 1.5 mM), and the concentration consistently increased toward later planting months. Similar to trans-aconitic acid, the sucrose content was the highest in No. 5 Gambela (77.6 ± 7.3 g L–1), Dale (76.9 ± 8.5 g L–1), and Isidomba (57.3 ± 6.3 g L–1); later planting increased sucrose contents. This trend in sucrose contents leads to analogous cultivar and planting month effects of total sugar (sucrose, glucose, and fructose) contents in Table : highest concentrations in No. 5 Gambela (116.7 ± 9.5 g/L), Dale (129.3 ± 11.2 g L–1), and Isidomba (103.1 ± 8.3 g L–1). Consequently, total organic carbon (TOC) was the highest in Dale (46.2 ± 4.2 g C L–1) and No. 5 Gambela (40.6 ± 3.6 g C L–1) and increased for later plantings. Similar trends for cultivar, planting month, and interactions (Figure a) were observed for Brix (estimate of g sucrose/100 g solution) in Table .

Table 2

Chemical Composition of Sweet Sorghum Juice Corresponding to Table .

							significant (<0.05) p value
variable	n	mean	SD	min	max	non-zero	cultivar	planting	interaction
Concentrations of Sugars and Organic Acids
glucose (g/L)	175	21	11	1.4	60	175	0.0002 (Dale > N110A x N98)	<0.0001 (↑↑)
fructose (g/L)	175	19	9	3.8	56	175	0.00008 (N110A x N98 < N111A x Dale, N111A x Isidomba, N111A x Atlas, N110A x Dale, N110A x Atlas; N109A xN98 < N110A x Dale)	<0.00001 (↑↑)
sucrose (g/L)	175	31	31	0.0	112	129	<0.00001 (No. 5 Gambela > all except N109A x Dale, Isidomba, Dale, N109A x Isidomba; Dale > all othersd; Isidomba > N110A x Chinese, N111A x Atlas, Chinese, N111A x Chinese, 110B, N111A x N98, N111B, N109A xChinese, N110A x Atlas, N110A x Isidomba; N109A x Dale > N111A x Atlas, Chinese, N111A x Chinese, N110B, N111A x N98, N111B, N110A x Atlas)	<0.0001 (↑↑)	0.04
total sugar (g/L)	175	70	43	5.2	157	175	<0.00001 (Dale > all others;e No. 5 Gambela > all others;e Isidomba > Chinese, N111A x Chinese, N111A x N98, N110B, N111B, N109A x Chinese, N109B)	<0.0001 (↑↑)	0.01
trans-aconitic acid (mM)	175	10	7	0.6	28	175	<0.00001 (No. 5 Gambela > all except N109A x Dale; N109A x Dale > Chinese, N111A x Chinese, N110B, N111A x N98, N111B, N109A x Chinese; N109A x Isidomba > N111A x Chinese, Chinese, N111A x N98, N110B, N111B, N109A x Chinese; Dale > Chinese, N111A x Chinese; N111B)	<0.0001 (↑↑)	0.002
cis-aconitate (mM)	175	15	14	1.4	96	175		<0.00001 (-↑)
citrate (mM)	175	2.7	5.5	0.0	73	172
oxalate (mM)	175	1.2	2.9	0.0	15	27		<0.00001 (-↑)
pH, Electric Conductivity (EC), Brix, and Total Organic Carbon/Nitrogen (TOC/N)
pH	174	5.7	0.3	5.0	6.4	174		0.001 (↓)	0.04c
EC (mS/cm)	173	8.4	2.4	3.9	16.3	173	<0.00001 (Isidomba, Dale < N98, N109A x Dale, N109A x Atlas, N110A x N98, N109A x N98, N111A x N98, N109A x Isidomba, N109A x Chinese, N109B; N109A x Isidomba, N109B > N110A x Chinese, N111A x Isidomba, N110A x Dale, N110A x Atlas, N110A x Isidomba)	<0.0001 (-↓)
Brix	174	8.2	4.2	1.1	17.2	174	<0.00001 (Dale > all others;f No. 5 Gambela, Isidomba > all others;f,g N109A x Dale > Chinese, N111A x N98, N111B, N109B; N110A x Dale, N111A x Isidomba > N111A x N98)	<0.0001 (↑↑)	0.002
TOC (gC/L)	175	24	17	0.0	60	161	<0.00001 (Dale,a No. 5 Gambelaa,b > all others; Isidomba > Chinese, N111A x Chinese, N111A x N98, N110B, N109A x Chinese, N111B)	<0.0001 (↑↑)	0.007
TN (gN/L)	175	0.71	0.23	0.09	1.79	175	0.04c	<0.00001 (-↓)

Except No. 5 Gambela, N111A x Dale, N109A x Dale, Atlas, N111A x Isidomba, Isidomba, N110A x Dale, N109A x Isidomba.

Not significantly different from N109A x Atlas, N110A x Atlas, and N110A x Isidomba.

No significant difference by Tukey.

Except No. 5 Gambela, N98, N109A x Dale, N111A x Isidomba, N110A x N98, N109A x N98, N110A x Dale, N109A x Isidomba, and Isidomba.

Except No. 5 Gambela, N111A x Dale, N109A x Dale, Atlas, N111A x Isidomba, Isidomba, N110A x Dale, and N109A x Isidomba (N110A x Atlas and N110A x Isidomba are not significantly different from No. 5 Gambela).

Except No. 5 Gambela, N109A x Dale, N111A x Isidomba, Isidomba, and N110A x Dale.

Not significantly different from N111A x Dale, Atlas (and N110A x Atlas for No. 5 Gambela).

Figure 6

Significant (p < 0.05) cultivar × planting date interactions observed for Brix (a), TOC (b), trans-aconitic acid (c), and % contribution of tryptophan fingerprint (d) variables (Tables –2).

In Table , electric conductivity (EC), pH, and total nitrogen decreased by later planting month. In addition, EC was the lowest in Isidomba and Dale, which accumulated among the highest amounts of total sugar in Table . The influence of ionic strength (estimated by EC) and pH on fluorescence intensity is well-described in the literature.[21,22,31] The fluorescence intensity of tryptophan and tyrosine sharply increases near pKa1 and then decreases near pKa2.[22] The ionizable groups (phenol for tyrosine, indole for tryptophan) of those amino acids engage in the cation−π interaction with their own aromatic ring.[22] Cation−π interaction (in addition to hydrophobic, hydrogen bonding, and ion-pair interactions) impacts the protein conformation, pKa, and fluorescence intensity of aromatic amino acids.[21,23,24] For sweet sorghum juice samples in this study, there was no cultivar-dependent pH change (Table ) to impact the interpretation of fluorescence signatures in Section .

Figure provides a detail for the cultivar × planting month interactions observed for the following variables: Brix, TOC, trans-aconitic acid, and juice %1 contributions (Tables and 2). Overall, Figure indicates consistently high trans-aconitic acid concentration, Brix, and TOC in No. 5 Gambela (and Dale to a lesser extent), regardless of planting month; other cultivars reached the same level only in the June planting. For trans-aconitic acid, Brix, and TOC, significant interactions were observed between cultivars of April and June plantings (Figure a–c). For example, N109A × N98 had among the highest trans-aconitic acid content in June (21.1 ± 2.0 mM) and accumulated among the lowest concentration in April (2.2 ± 2.0 mM), relative to the 23 other cultivars (Figure c). A contrasting interaction trend with higher p value was observed for the tryptophan-like PARAFAC contribution (%1 in Figure d). Both planting month and cultivar trends for tryptophan (Figure d) were reversed from the organic carbon variables (Figure a–c).

Possible Relationship between Fluorophores and Sugarcane Aphid Tolerance

Among the 24 cultivars tested, No. 5 Gambela sustained the least amount of visible damage from sugarcane aphids. Dale was found to be more susceptible than No. 5 Gambela but was still less damaged than most other entries. These two cultivars were among the highest in juice concentration of trans-aconitic acid. The relative marginal effect of damage, averaged over all three planting dates, was plotted against the average trans-aconitic acid concentration of stalk juice in Figure . This plot reveals a strong negative correlation (r = −0.873, p < 0.0001) between these two variables. A similar correlation was observed within the April and May plantings (r = −0.786 and −0.637, respectively) but was not observed in the June planting (r = −0.085) due to low aphid infestation in the latest planting (not shown). This suggests a possible role of trans-aconitic acid in tolerance to sugarcane aphid, though it is probably not the only factor involved.

Figure 7

Relative marginal effects of sugarcane aphid damage for each of 24 cultivars averaged across three planting dates (April, May, and June) at Tifton, GA in 2017 (y-axis) vs average trans-aconitic acid concentration in stalk juice (x-axis).

Conclusions

The purpose of fluorescence technique is to estimate the relative fraction of target analyte in a large set of samples, without analytical separation required for metabolomics. Similar peak assignments were obtained using synchronous and EEM/PARAFAC techniques; however, the low fluorescent structure, tyrosine, was only observable using EEM/PARAFAC. Tryptophan had the highest overall contribution (sum of all absolute PARAFAC contributions) to juice samples because of its highest fluorescence intensity (Figure ). Cultivar No. 5 Gambela accumulated the highest amount of a fluorophore consisting of trans-aconitic acid and polyphenols (22 ± 1% and 2354 ± 115 absolute contributions of juice fingerprint 3), which increased as a function of planting month, and among the lowest amount of protein-like structures (21 ± 1% of juice fingerprint 2). Cultivar N109A × Chinese accumulated the tryptophan-like structure with a decreasing trend as a function of planting month. Higher stem total sugar concentrations at physiological maturity with later planting (April, May, to June) is consistent with the 2015 planting year at Tifton due to heavy infestation of sugarcane aphids in the early plantings.[8] The opposite trend was observed in another study in the absence of aphid infestation, where higher total sugar was observed in earlier planting from May to July at various sites in Arizona.[29] Although a correlation does not validate a cause-and-effect relationship, the highly negative correlation between aphid damage and trans-aconitic acid concentration in the juice could be used as a rapid chemical sensing technique to prescreen breeding material for potential resistance along with selection for other juice quality traits.

Experimental Section

Field Experiment

The field experiment was conducted in Tifton, GA in 2017. The experimental design was a split-plot with three replications, where planting dates (April, May, and June) were the main plot factor and 24 sweet sorghum cultivars (Table S1) were randomly assigned to subplots. Sugarcane aphid damage was rated biweekly on six occasions. Damage was rated on a scale of 1–9, where 1 represents no damage and 9 represents dead plants, similar to the scale described by Sharma et al.[25] Each subplot was harvested when it reached the hard-dough stage of maturity (when the stem sugar content typically peaks). Three representative stalks were harvested from each subplot, panicles and leaves were removed, and juice was extracted from the stems by passing twice through a portable three-roller mill (Sor-Cane Porta-Press, McClune, Reynolds, GA). Juice samples were immediately frozen after measuring the soluble solids concentration (Brix) using a digital refractometer (Refracto 30GS, Mettler-Toledo, Columbus, OH). The bagasse portion was dried to completion at 60 °C, ground in a Wiley mill (Thomas Scientific, Swedesboro, NJ), and sieved (<2 mm). Additional details about the field experiment and chemical analysis of juice samples are given in Section SI of the Supporting Information and were described in detail previously.[8,9,19]

Reflective and Penetrative Fluorophotometry in EEM and Synchronous Modes

Penetrative fluorophotometry employs a 90° angle between the excitation light and detector, and the Beer–Lambert law follows for dilute aqueous samples.[2] However, scattering, quenching, and inner-filtering[20] become significant enough to violate the Beer–Lambert law for concentrated, turbid/colloidal liquid samples and for solid samples.[2] For such samples, a front-faced cell holder is used for the reflective measurement of emission (only from the sample surface) to minimize the influence of nonfluorescence disturbance.[2,21]

Reflective fluorescence EEM of solid bagasse powder (<2 mm) and selected undiluted, unfiltered juice samples were obtained using an F-7000 spectrofluorometer (Hitachi, San Jose, CA) at 220–500 nm excitation and 280–730 nm emission wavelengths in 3 nm intervals; 2.5 nm excitation and emission slits; auto response time; and 2400 nm min–1 scan rate. A solid sample holder was used to allow the light beam irradiation at 30° and reflection at −60°, along the center line of the quartz window; this design prevents the reflected light beam from reaching the light detector.[21] An empty powder sample cell with a quartz window (for bagasse powder) or distilled, deionized water (DDW) in the front-surface cuvette (for juice) was used as the blank and was subtracted from each sample to remove the lower intensity Raman scattering.[22] Penetrative spectral collection for diluted (20-fold with DDW) and filtered (0.45 μm) juice samples used a 10 mm square cell with DDW as the blank, using the same spectral parameters described above.

Additional EEM regions dominated by Rayleigh and Raman peaks and the region without fluorescence were removed, and then, parallel factor (PARAFAC) models[26] were constructed with non-negativity constraint using MATLAB version 8.6.0.267246 (R2015b; Mathworks, Natick, MA) with PLS toolbox version 8.6.2 (Eigenvector Research, Manson, WA). As described in detail elsewhere,[26] PARAFAC models three-way data (samples, excitation wavelengths, and emission wavelengths) by minimizing the sum of squares of the residuals. On the basis of residual/leverage analysis, core consistency diagnostic scores of 2–7 component models and comparison with prior reports,[8,10] three-component models (core consistency: 90 for juice, 57 for bagasse) were selected to interpret the PARAFAC results.

The EEM and synchronous scans[27,30] are available both for the penetrative measurement of dilute aqueous samples and the reflective measurement of solid samples. The EEM offers detailed spectral features by scanning the entire emission (Em) wavelength range for each excitation (Ex) wavelength. A synchronous scan is used to decrease the measurement time and to improve the resolution of characteristic spectral bands.[27] In the synchronous scan, Ex and Em wavelengths are simultaneously varied, while keeping a constant wavelength interval (delay time) between Ex and Em; luminescence intensity depends on both Ex and Em wavelengths.[27] Wavelength scan in the synchronous mode was conducted for bagasse powder and diluted (20-fold) juice samples without blank subtraction, using 60 nm (for bagasse; 280 nm emission wavelength) and 90 nm (for juice; 310 nm emission wavelength) delays at 220–730 nm excitation wavelength; 2.5 nm excitation and emission slits; auto response time; and 1200 nm min–1 scan speed. Areas of synchronous peaks were calculated by the trapezoidal integration at 250–730 nm using OriginPro 2019 (OriginLab, Northampton, MA).

Statistical Analysis

For chemical parameters, the effects of cultivar, planting month, and their interaction were examined by factorial ANOVA using Statistica version 12 (Statsoft, Tulsa, OK) at a significance level of p < 0.05. Type VI sums of squares were used to test the effective hypothesis for unbalanced observations. If significant differences existed, post hoc comparisons were conducted using Tukey’s HSD test. Sugarcane aphid damage ratings were converted to ranks using Proc Sort and Proc Rank in SAS v. 9.4 (SAS Institute, Cary, NC), followed by repeated-measures ANOVA on the ranks in Proc Mixed, as described by Shah and Madden.[28] The obtained midrank values were then converted to relative marginal effects.[28]