Environ. Sci. Technol. 2003, 37, 5254-5261
Use of Multiwavelength Transmission Spectroscopy for the Characterization of Cryptosporidium parvum Oocysts: Quantitative Interpretation MICHAEL R. CALLAHAN,† JOAN B. ROSE,‡ AND L U I S G A R C IÄ A - R U B I O * , † College of Marine Science, University of South Florida, St. Petersburg, Florida 33701, and Department of Fisheries and Wildlife, Michigan State University, East Lansing, Michigan 48824
Combined scattering and absorption properties of suspended particles can be obtained as a function of wavelength by measuring the complete ultraviolet-visible (UV-vis) spectrum. This research reports on the quantitative interpretation of measured UV-vis spectra of Cryptosporidium parvum oocyst suspensions obtained from several commercial sources and evaluated using two different purification techniques. The reproducibility of the measured spectral data was assessed, and the quantitative interpretation of the oocyst spectra in terms of the particle size and the chemical composition of the particles are reported herein. The interpretation model of the spectra is based on light scattering theory, spectral deconvolution techniques, and on the approximation of the wavelength-dependent optical properties of the basic constituents of living organisms. A characteristic set of optical properties for C. parvum oocysts has been determined as a function of wavelength and used for the quantitative interpretation of UV-vis spectra. The results from the spectral deconvolution show quantitative differences among oocyst preparations. These results represent the first step in establishing a set of critical parameters (e.g., oocyst size and chemical composition) necessary for the detection and identification of C. parvum oocysts in water using spectroscopy.
Introduction Cryptosporidium parvum is an enteric protozoan known to cause cryptosporidiosis (acute gastroenteritis) in humans and other animals. Transmission of crytosporidiosis is via the fecal-oral route and is commonly disseminated through water contaminated with the disease agent (Cryptosporidium oocysts). The resistant nature of the oocyst life stage allows Cryptosporidium (spp.) to survive in water for months and renders the organism impervious to conventional disinfection and chemotherapeutic treatments (1). Oocysts are regularly detected in surface waters used for drinking purposes (2-5), and have been implicated as the etiologic agent in at least * Corresponding author phone: (727)553-1246; fax: (727)553-1189; e-mail:
[email protected]. † University of South Florida. ‡ Michigan State University. 5254
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 22, 2003
75 waterborne outbreaks of cryptosporidiosis worldwide (4, 5). The characterization of the oocyst has been suggested as a first step in the potential development of approaches used to reduce the water-borne transmission of cryptosporidiosis (1). A newly developed technique for the characterization of biological suspensions is mulitwavelength transmission spectroscopy (6-11). Analysis of multiwavelength transmission spectra is generally performed across the ultravioletvisible (UV-vis) light range of the electromagnetic spectrum, where the absorption and scattering properties of a particle can be expected to have a predictable influence over the observed spectral pattern (6-16). Quantitative characterization is achieved through interpretation models based on light scattering theory, spectral deconvolution techniques (8), and on the approximation of the wavelength-dependent optical properties of the basic constituents of living organisms (611). In addition, by obtaining spectral information over a range of optical wavelengths, multiwavelength spectroscopy offers significant advantages over other optical characterization techniques (e.g., single wavelength laser-light scattering techniques and nephelometry techniques) in that rapid particle enumeration, classification, and detection are possible even in multicomponent particle suspensions (8). C. parvum oocyst preparations from three U.S. vendors were characterized using multiwavelength transmission spectroscopy techniques. Spectral patterns among oocysts suspensions from the preparation of a single vendor were not significantly different, while spectral patterns on a vendorto-vendor comparison were found to be distinct. In this paper, the measured spectra have been evaluated via modeling techniques to quantify the spectral differences observed among commercial preparations for both vendor-purified and IMS(immunomagnetic separation)-purified oocyst samples. Oocyst optical properties, particle size distributions, particle concentrations, and oocyst nucleic acid concentrations were estimated using the interpretation model described herein. The modeling technique presented represents an initial step in establishing the natural variability in the spectral features of C. parvum oocyst populations. The consistent agreement found between model calculated and measured oocyst spectral patterns suggest that future development of light scattering and absorption models may be used to characterize, enumerate, and potentially detect oocysts in natural waters using very simple measurements of multiwavelength transmission.
Experimental Section Interpretation Model. The proposed model used to interpret C. parvum oocyst spectra is based on Mie theory (17) and on recent developments in the interpretation of spectral data (6-11), where the volume of the microorganisms is expressed in terms of an equivalent sphere. As part of the model’s interpretative regime, the complex physical and chemical structures of the microorganisms are approximated by partitioning the structural constituents into three main groups or populations. Each of the three groups are characterized by a set of scattering and absorption components which include (i) the body of the microorganism and its characteristic dimensions, (ii) a characteristic body dimension representing the average internal structure and corresponding chemical composition, and (iii) the absorption and scattering due to other well-defined structures such as inclusions and chromosomes. The observed spectrum is a weighted sum of the total scattering and absorption components from the three groups. Chromophoric groups such 10.1021/es034366y CCC: $25.00
2003 American Chemical Society Published on Web 10/22/2003
as nucleic acids can be considered to be part of a particular scattering element (e.g., the body of the organism, nucleus, chromosomes, etc.) or considered as a scattering element by themselves. Under these approximations, the turbidity spectrum of a microorganism can be written in terms of the three distinct populations (6, 7, 9-11):
τ(λ0) ) π Npl [(x1 4
() ∫Q
(x2
∫Q ∞
0
∞
0
1ext(m(λ0),D)D
2ext(m(λ0),D)D
(x3
∫Q ∞
0
2
2
f1(D) dD)macrostructure +
f2(D) dD)internal structure +
3ext(m(λ0),D)D
2
f3(D) dD)chromophore] (1)
where τ(λ0) is the turbidity measured at a given wavelength λ0, fi(D) is the normalized particle size distribution for each scattering population, D is the spherical equivalent particle diameter, Qi,ext(λ0,D) corresponds to the extinction efficiency of the ith particle population, Np is the total number of particles per unit volume distributed between the number of scattering elements or populations, l is the optical path length and xi (i ) 1-3) corresponds to the number fraction for each population such that 3
∑x ) 1
(2)
i
i)1
Assuming volume additivity, the total concentration can be readily calculated in terms of the concentration of each population (ci): N
ctotal )
∑c
(3)
i
i)1
The complex refractive index mi(λ0) is a function of the chemical composition and can be calculated as a weighted sum of the contributions from the chromophores within each population: N
∑ mi(λ0) )
N
ωijnij
j)1
i +
n0
∑ω k
ij ij
j)1
n0
(4)
where ωij is mass fraction of jth chromophore contained in the ith population, ni and ki correspond to the real and imaginary refractive indices of each population, and N is the total number of chromophoric groups. Adding the scattering contributions represented by eqs 1 and 2 closes the total mass balance for each chromophoric group. The diameter in eq 1 can be calculated from the closest geometrical approximation to the shape of the microorganisms. Note that within these approximations it is possible to consider additional groups or structures such an endospores, nucleotides, sporozoites, and others. Oocyst Preparations. C. parvum oocysts were obtained from three U.S. vendors [Waterborne Inc. (New Orleans, LA), Parasitology Research Labs/LLC DyNAgenics (Neosho, MO), and Pleasant Hill Farms (Troy, ID)]. All C. parvum oocyst preparations were initially obtained from dairy calf isolates. The Waterborne (WTB) oocyst isolate came from a local strain collected from a Louisiana calf. Oocysts from Pleasant Hill Farm (PHF) and Parasitology Research Laboratories (PRL) are of the Ames, IA, isolate. Waterborne and Parasitology Research Labs maintained their respective isolates by passage of the C. parvum oocysts through mice. Pleasant Hill Farms maintained their isolate by passage of the oocysts through newborn calves.
Separation of the oocysts from the host fecal debris was performed using density gradients to suspend or overlay the oocysts above the host waste material. The suspended oocysts were then collected at the density gradient interface. Waterborne Laboratory used a solution of percoll-sucrose to achieve an oocyst overlay (18). Pleasant Hill Farms used ethyl ether to separate the lipid material in the feces, followed by an oocyst overlay using Sheathers solution (19). Parasitology Research Laboratory overlaid oocysts using a sodium chloride density gradient (K. Swabby, personal communication). Each vendor preparation was split into six separate samples. Half of the samples (three per vendor) were spectrally measured after washing (3×) for 2 min at 10500g by centrifugation (Microfuge Beckman) in the desired spectral buffer (referred to as non-IMS samples). The other half of the samples (three per vendor) were measured after processing through an advanced purification step known as immunomagnetic separation (referred to as IMS samples). Following IMS, the oocysts were washed (3×) by centrifugation in the spectral buffer (as described above). Spectroscopy Measurements. Two types of spectroscopy measurements were conducted: transmission measurements to obtain the UV-vis transmission spectra of C. parvum oocyst suspensions and absorbance measurements with an integrating sphere to obtain estimates of the absorption component of the complex refractive index. The transmission measurements were recorded on a Hewlett-Packard 8452A diode array spectrophotometer (1-cm path length) according to procedures previously described (8). Multiwavelength Absorbance Measurements. Measurements of absorbance were taken using a Perkin-Elmer Lambda 18 UV-vis scanning spectrophotometer equipped with an integrating sphere (Labsphere, New Hampshire). Absorption measurements were taken using a 1-cm path length quartz cuvette (Starna; Atascdero, CA), which was rinsed in tap water (3x) and dionized water (3x) prior to being filled with an aliquot of sample. The cuvette was then inverted (10x), and its exterior surface was wiped dry using a Kimwipe. The cuvette was then placed in the holding chamber of the integrating sphere with the same orientation for each spectral read. For measurement purposes, the absorbance (A) of all suspensions was obtained per wavelength as a ratio of transmitted light intensity (A ) log Io/I) using the suspending medium with (I) and without (Io) the suspended oocysts. Spectral readings were taken at wavelengths between 190 and 1100 nm with a resolution of 1 nm. Particle Sizes and Particle Counts. The particle sizes and the particle counts were estimated using the UV-vis interpretation model (eqs 1-4). Corroborative particle size measurements and particle counts were taken with an ocular micrometer, an impedance-based hematology analyzer (model 9010+, Serono Baker; Allentown, PA), and a Z-2 Coulter Counter according to procedures described elsewhere (8, 20). Optical Properties. The optical properties necessary for the implementation of eqs 1-4 were divided into two categories: (i) the average optical properties pertaining to the scattering groups approximating the oocyst macrostructure and the internal structure and (ii) the optical properties of known chromophoric groups within the oocysts. Both sets of optical properties were estimated using the methods described by Alupoaei et al. (6, 7, 9-11). The optical properties for the macrostructure and the internal structure were estimated with a two-parameter Cauchy approximation of the real refractive index (the absorption coefficient being negligible). The refractive index equation for the macroVOL. 37, NO. 22, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5255
structure is given by
n(λ0) ) 1.380 +
1747 λ20
(5)
5900 λ20
(6)
and for the microstructure:
n(λ0) ) 1.550 +
The second category of optical properties corresponds to chromophoric groups that can be incorporated into defined scattering elements or as single scattering elements (i.e., DNA, RNA, etc.). To approximate the optical properties of the nucleic acids and free nucleotides, the values reported by Alupoaei et al. (6, 11) have been used. The refractive index of water n0(λ0) as a function of wavelength in eq 4 was calculated from the correlation reported by Thorma¨hlen et al. (21). Nucleic Acid Estimates. Nucleic acid estimates of C. parvum oocysts were achieved by (i) spectrally deconvoluting the absorption component of C. parvum spectra, (ii) acquiring estimates of the absorbance for the deconvoluted absorbance band at 260 nm, and (iii) using the A260 estimates in eq 7:
[NA] )
[
]
(A260)0.1176 (A260)0.8824 + Np-1 DNA RNA
(7)
where [NA] represents the concentration of nucleic acid per oocysts, Np is the number of oocysts per unit volume, and A260 is the absorbance values at 260 nm. For this calculation, the extinction coefficient () of DNA and RNA was approximated to be 0.02 cm2/µg (22) and 0.025 cm2/ µg (23), respectively. The constants, 0.1176 and 0.8824, represent the respective ratios of DNA and RNA to total nucleic acid estimated for C. parvum oocysts (24).
FIGURE 1. Average (n ) 15) spectral patterns for C. parvum oocyst vendors treated using non-IMS (a) and IMS (b) procedures. Error bars around the spectra represent the 95% confidence interval around replicate spectral measurement per vendor.
Results A set of spectral data was collected for three separate oocyst vendor preparations purified using either non-IMS or IMS protocols. Representative vendor spectra are shown for replicate measurements (n ) 15) of oocyst suspensions treated using non-IMS (Figure 1a) and IMS (Figure 1b) protocols. In general, no statistical difference was observed for replicate sample measurements while spectral patterns compared across vendors and across treatments were found to be statistically significant. The oocyst spectra were analyzed using the interpretation model (eq 1). As discussed in Mattley et al. (8) and Alupoaei et al. (6, 11), the quantitative interpretation of the spectra involves calculating a transmission spectrum using the interpretation model and the estimated optical properties. A close match between the calculated spectrum (eq 1) and the measured spectrum compared at each wavelength validates the adequacy of the interpretation model and estimated optical properties. The residual sum of squares estimates (RSSQN) from the normalized transmission spectra were used to quantify the match between the model predicted scattering spectra and measured C. parvum vendor spectra (Table 1). For the non-IMS oocyst treatments, RSSQN estimates between modeled and measured spectra ranged between 0.44 × 10-2 and 1.49 × 10-1. For the IMS-treated oocysts, RSSQN estimates ranged between 0.55 × 10-2 and 5.51 × 10-2. The low RSSQN values provide evidence that the interpretive model and estimated optical properties used in the model adequately approximate a variety of oocyst preparations. Values used to approximate the optical properties of the nucleic acids and free nucleotides (6, 9-11) have been 5256
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 22, 2003
compared with the absorbance values for an oocyst suspension measured with an integrating sphere (Figure 2). Notice that the spectra show similar features, particularly in the spectral region corresponding to the nucleotide absorption (250-300 nm). This result suggests that the approximations used to calculate the scattering contributions from the total nucleotide concentration are adequate. Evaluation of Estimated Optical Properties and Interpretative Model. To test the algorithm of the interpretation model and to make an estimate regarding the reliability of the optical properties determined for C. parvum oocysts, the interpretation model (eqs 1-6) was used to calculate estimates of the average particle size and count from measured non-IMS and IMS oocyst spectra. The model calculated estimates were compared to particle size and concentration estimates for the non-IMS and IMS oocyst preparations. The average model-derived oocyst size estimate for each of the composite non-IMS and IMS vendor preparations (n ) 5 spectra) are shown in Table 1. For the nonIMS spectra, average particle size estimates were 4.92, 4.54, and 4.91 µm for the PRL, WTB, and PHF oocyst preparations, respectively. For the IMS spectra (n ) 15), the average modelderived particle size estimates for PRL, WTB, and PHF oocysts were 3.96, 4.37, and 4.20 µm, respectively. The mean average size estimates determined by the model for the three IMS- and non-IMS-treated vendor oocyst spectra were 4.17 and 4.78, respectively, and are in agreement with the 4-6 µm size range used by Bennett et al. (25) as one of the criteria for identifying potential oocysts. The validity of using the single average diameter size to describe a sample has been shown for blood platelets (8) and for vegetative
TABLE 1. Model Interpreted Estimates from Average Spectra per Experiment and Vendora vendor
treatment
PRL
IMS non-IMS
WTB
IMS non-IMS
PHF
IMS non-IMS
sample
Dc (µm)
Do (µm)
Vf ο
fNA
CNA (×10-12 g/cell)
RSSQN
1 2 3 1 2 3 1 2 3 1 2 3 1 2 1 2
3.91 3.92 4.06 5.01 4.86 4.90 4.31 4.33 4.48 4.49 4.57 4.56 4.20 4.20 4.95 4.86
0.099 0.099 0.086 0.102 0.094 0.101 0.084 0.086 0.086 0.079 0.092 0.146 0.100 0.094 0.218 0.091
0.268 0.265 0.291 0.228 0.245 0.230 0.254 0.258 0.314 0.201 0.165 0.085 0.143 0.159 0.147 0.163
0.101 0.087 0.069 0.094 0.063 0.106 0.072 0.085 0.041 0.023 0.046 0.088 0.042 0.037 0.111 0.094
0.973 0.865 0.704 1.497 0.976 1.599 0.815 0.983 0.637 0.230 0.396 0.390 0.249 0.241 1.036 0.975
3.85 × 10-2 3.75 × 10-2 5.51 × 10-2 1.49 × 10-1 3.54 × 10-2 1.41 × 10-1 4.82 × 10-2 4.23 × 10-2 4.76 × 10-2 0.44 × 10-2 1.13 × 10-2 1.04 × 10-2 0.55 × 10-2 0.73 × 10-2 4.53 × 10-2 2.99 × 10-2
a Coefficient of variation estimates for D , D , V , and C c o fo NA are 1.5, 5.0, 3.0, and 10%, respectively (7). Dc is the average particle diameter of oocyst macrostructure. Do is the average size estimate of internal scattering structures. Vfο is the volume fraction of internal scattering structures. fNA is the volume fraction of nucleic acids in internal scattering structures. CNA is the concentration of nucleic acid (g/oocyst). RSSQN is the residual sum of squares from the normalized spectra.
FIGURE 2. Absorption component of whole C. parvum oocysts (solid line) measured with an integrating sphere as compared to the absorbance spectrum of free nucleotides (dotted line) calculated using the approximation reported by Alupoaei et al. (7). cells and spores (6, 7, 9-11). Although this validation has not yet been shown for samples of C. parvum oocysts, Figure 3 shows that, for both non-IMS- and IMS-treated oocysts, the mean average particle size determined for all three vendors is within the total oocyst particle size distribution for the three vendor preparations measured using an ocular micrometer. The mean average model-derived particle size estimate obtained for the three non-IMS-treated vendor spectra was within 14.3% of the average particle size for the three vendors measured using the ocular micrometer. For the three IMS-treated oocyst vendor spectra, the mean average particle size estimate from the model was within -1.70% of the measured average. In addition, the mean average particle size calculated from the interpretive model for five non-IMS-treated samples from a single oocyst preparation was within the 95% confidence interval of the average oocyst size measured using an impedance-based particle counter (model Z-2, Coulter). Less than 2% difference was observed between the average particle size from the interpretation model and the Coulter Counter size estimates (Figure 4). The interpretation model (eq 1) also provides estimates of the particle concentration from spectra that have not been normalized. The results from the interpretation model were compared to concentration estimates measured using an
FIGURE 3. Average particle size distribution for all oocyst vendors determined from ocular micrometer estimates (a) compared with the size distribution of the mean (n ) 45) average particle size calculated from the interpretative model for all non-IMS (b) and IMS (c) treated samples. impedance-based Coulter Counter for five samples taken from a non-IMS-treated oocyst preparation. Oocyst concentration estimates for each of the five non-IMS oocyst samples using the interpretation model were within close approximation (
