Griffithsin also effectively inhibits transmission of HSV-2 , HCV , SARS-CoV , Ebola , and possibly other viruses yet to be studied. Importantly, Griffithsin appears devoid of cellular toxicity that is associated with other lectins. O’Keefe et al. conducted studies with explants of macaque and rabbit vaginal tissues ex vivo and showed that Griffithsin did not induce changes in the levels of cytokines or chemokines, nor did it alter lymphocyte levels in human cervical tissue nor elicit inflammatory responses in rabbit tissue . The combination of extremely wide viral target range and demonstrated preclinical safety makes Griffithsin potentially useful as a prophylactic and/or therapeutic in multiple and diverse antiviral indications. The potential indications for Griffithsin as a human prophylactic or therapeutic include its use as an active pharmaceutical ingredient in vaginal and rectal microbicides. In spite of the value shown by pre-exposure prophylaxis drugs to prevent HIV transmission, issues of cost, side effects, the potential for development of viral resistance through chronic use of antiretrovirals as prevention modalities, and access to PrEP drugs by under resourced populations remain. These unmet needs could be met by the availability of affordable, safe and effective “on demand” antivirals, especially with Griffithsin as the API and its potential to control co-transmitted viruses such as HIV-1, HSV-2 and HCV during intercourse. Adoption of Griffithsin as a new biologic drug, especially in cost-constrained products such as microbicides, is predicated on the feasibility of a scalable manufacturing process that can supply market-relevant volumes of the API at an acceptable cost of goods sold . Previously, we showed that recombinant Griffithsin can be expressed and isolated with high efficiency using transient gene expression in green plants . Although the process described can be further optimized,growing bags the achieved pilot-scale expression yields of >0.5 g Griffithsin per kg of fresh green biomass , recovery efficiencies of 60–90% overall, and Griffithsin purity of >99% of total soluble protein are already impressive.

In this study, we developed a technoeconomic model for Griffithsin manufacturing using a plant-based system with the goal of estimating API manufacturing cost and determined the factors that have the greatest impact on COGS. The output of our study should serve as a basis for additional process improvements, selection of a commercial-scale manufacturer, and should assist in the identification of future product targets for cost-sensitive markets such as prophylactic microbicides as well as those for less cost-constrained therapeutic indications. Technoeconomic modeling was performed with the widely used SuperPro Designer modeling software . The main analysis in this study was conducted using data available from pilot-scale manufacturing of Griffithsin in Nicotiana benthamiana plants using tobacco mosaic virus -induced transient gene expression, and assuming that manufacturing would take place in an existing and fully equipped state-of-the-art plant-based bio-manufacturing facility. Modeling costs based on existing resources of a contract manufacturing organization instead of a “greenfield” build of a new facility was seen as the most likely scenario for launch of a new product. Our reasoning was that dedicated infrastructure could be built subsequently depending on market demand for the drug. As a result, we did not estimate capital equipment or total capital investment costs, and neglected depreciation, insurance, local taxes and factory expenses in the manufacturing operating cost analysis as these investments would have been made by the CMO. Our analysis assumed a 20% net profit margin/fee assessed by the CMO and this figure was added to the production cost of the product to arrive at the final total product cost. In addition to the techno economic analysis, an Environmental Health and Safety Assessment of the designed process was conducted using the method described by Biwer and Heinzle to evaluate the environmental, health and safety impact of Griffithsin manufacturing using the plant-based system, with the goal of assessing the sustainability of the process. The techno economic modeling for this study was performed using SuperPro Designer , Version 9.5 , a software tool for process simulation and flow sheet development that performs mass and energy balances, equipment sizing, batch scheduling/debottle necking, capital investment and operating cost analysis, and profitability analysis. This software has been used to estimate cost of goods in a variety of process industries including pharmaceuticals produced by fermentation and plant-made pharmaceuticals .

It is particularly useful at the early, conceptual plant design stage where detailed engineering designs are not available or warranted. SuperPro was chosen because it has built-in process models and an equipment cost database for typical unit operations used in the biotechnology industry, such as bioreactors, tangential flow ultrafiltration and diafiltration, chromatography, grinding or homogenization, and centrifugation. There are some specific unit operations and processes used in this study that are currently not included in SuperPro, such as indoor plant cultivation, transplantation, plant harvesting and screw press/disintegrator. Such unit operations were addressed through the “Generic Box” feature of the application. Unless otherwise noted, the maintenance costs of major equipment, unit operation-specific labor requirements and costs , pure components, stock mixtures, heat transfer agents, power and consumables used in the analysis were determined using the SuperPro built-in equipment cost model and default data banks. Additional case study specific design parameters were selected based on experimental data from journal articles, patent literature, the authors’ laboratories, interviews with scientists and technologists conducting the work cited, technical specification sheets or correlations, heuristics, or assumptions commonly used in the biotechnology and/or agricultural industry.Process flow and unit operations were derived from published methods and unpublished results obtained by the authors and collaborators who have participated in the development and scale-up of the process described and in the development of Griffithsin products. On the basis of this information, the SuperPro software was used to select and size equipment for each of the unit operations to achieve the desired production target , simulate the operations by performing material and energy balances, and specify and schedule all operations taking place within each piece of equipment to calculate material inputs and outputs and process times. Costs for raw materials, utilities, consumables, labor, laboratory QA/QC, waste disposal and equipment maintenance were then used to determine annual operating costs, and per-unit mass or per-dose costs . The main case study model was based on an existing plant based manufacturing facility, operating in batch mode, and excluded new capital investments and other facility dependent costs, except for equipment maintenance costs, which were included.

For the downstream portion of the Griffithsin manufacturing process, an annual available operating time of 7,920 h for the facility was used with indoor-grown Nicotiana benthamiana plants. Operating time was based on Holtz et al. for a similar facility, which was designed with overlapping utility capacity and in which the largest single utility unit can be down for maintenance and/or repairs and the utility loads can be maintained with redundant equipment. Likewise, per Nandi et al. it was assumed that the plants would be grown continuously throughout the year . Land costs, upfront R&D, upfront royalties, and regulatory/certification costs were neglected in the model as these costs can vary widely. Griffithsin protein can be produced in plants in a number of ways. These include stable expression in recombinant plants; inducible expression in transgenic plants; transient expression induced directly by tobacco mosaic virus replicons; or via agrobacterial vectors introduced into the plants via vacuum assisted, or surfactant-assisted, infiltration . Relative to stable transgenic plants, the advantages of speed of prototyping, manufacturing flexibility,nursery grow bag and ease of indoor scale-up are clearly differentiating features of transient systems and explain why this approach has been widely adopted in the manufacture of many plant-made pharmaceuticals . In our base-case analysis, we modeled expression of Griffithsin using TMV induction described in Fuqua et al. and results from 3 pilot-scale manufacturing runs because these batches provided the most extensive and complete data set; however, this process has been corroborated in 6 additional manufacturing runs at pilot-scale or larger.icotiana benthamiana host plants are generated from seed and propagated indoors under controlled environmental conditions until sufficient biomass is obtained for inoculation with the TMV vector carrying the Griffithsin gene. The process is summarized as follows. An N. benthamiana Master Seed Bank is generated from seeds obtained from the U.S. Department of Agriculture Repository. For bio-manufacturing, seeds from the TW- 16 line are obtained in bulk and stored securely. The Master Seed Bank is qualified for germination rate , freedom from disease, and genetic uniformity, and stored in sealed containers under temperature-controlled conditions . If the seed batch passes release tests, it becomes the Production Seed Batch and is used in the designated production run . Seedlings are allowed to grow for 21 days under controlled environmental conditions . At this stage, the seedlings are transplanted to accommodate their larger size and moved to another growth room to await inoculation, as described in the following sections.

The API extraction procedure modeled is per Holtz et al. except that a 1:1 ratio of biomass:buffer is used. Briefly, the aerial parts of the plants containing accumulated Griffithsin are mechanically inverted and cut with a mechanical cutter. The harvested biomass is collected in baskets for transport to the extraction suite, to initiate downstream processing. The harvested biomass fresh weight is determined to calculate the volume of extraction buffer to be added, typically at a rate of 1 kg biomass FW:1 L buffer mix . The pH is adjusted to 4.0 and the mixture is heated to 55◦C for 15 min to help precipitate major host plant proteins. The heated mixture is passively cooled and filtered to yield a crude extract. The crude extract is stirred overnight at 4◦C in the presence of bentonite and MgCl2. This procedure helps remove TMV coat protein , which at this step represents the largest protein impurity in the extract. The suspension is filtered to remove aggregated TMV CP, yielding a clarified and partially purified API-containing solution and is then sterile- filtered . In-process controls are applied throughout downstream processing unit operations to determine reagent volumes and assess yield and quality at key steps. To adequately meet the projected initial annual market demand for a rectal micro-bicidal formulation in the United States, approximately 6.67 million doses of Griffithsin API at 3 mg/dose would be needed. This translates into a production rate of 20 kg of purified Griffithsin API per year. The manufacturing facility to produce the required 20 kg of API per year was assumed to segregate production operations into two broad categories; namely, upstream production and downstream recovery and purification. To accommodate a large number of plants, the facility uses a vertical cultivation design with integrated irrigation and runoff collection system. Each rack is compatible with an integrated transportation infrastructure to move each tray to the next phase of the growth cycle. The upstream portion of the facility houses unit operations for N. benthamiana propagation, inoculation with TMV vector, and Griffithsin protein expression and accumulation. These processes begin with seeding and end when the biomass is taken to harvest. The downstream portion of the facility begins at harvest and continues through purification of the Griffithsin DS. Upstream processing is assumed compliant with good agricultural practices , whereas downstream processing is subject to FDA current good manufacturing practice . The general layout of the upstream growth rooms was adapted from Holtz et al. , and includes one germination chamber for seeds, one pre-inoculation room for biomass growth, and an isolated post-inoculation chamber where N. benthamiana inoculated with TMV expresses and accumulates Griffithsin. Plants are arrayed in equally sized trays under light-emitting diode light systems tuned to the optimized photosynthetic absorbance spectrum of N. benthamianaand are continuously illuminated. The plants are rooted in rock wool cubes held in the trays by polystyrene foam floats and perfused with a nutrient solution . Hydroponic irrigation is on a 12-h cycle and is accomplished via nutrient film technique . We modeled a hydroponic system because the nutrient solution is recycled; hence, water is conserved, and fertilizer runoff is reduced although not eliminated. The mass of nutrient solution taken up by the plants, the cost of the nutrient solution per liter, and the mass of residual nutrient solution that goes to the wastewater treatment system are shown in Supplementary Table 1 in Supplementary Material.

The United States and Mexico appeared close to agreement on a program to legalize farm and other workers before September 11, 2001. However, after the war on terror was declared, the momentum for a new meant that more tons of vegetables were produced from the same acreage, while acreage of fruits and nuts rose from 2 million acres in 1990 to 2.4 million acres in 2000, a 19% increase over the 1990s. Many FVH commodities are labor intensive, with labor accounting for 15% to 35% of production costs. Most of the workers employed on FVH farms are immigrants from Mexico, and a significant percentage are believed to be unauthorized . In recent years, several proposals have aimed to reduce unauthorized worker employment in agriculture . In September 2001, Mexican President Vincente Fox called for a U.S.-Mexico labor migration agreement so that “there are no Mexicans who have not entered this country [U.S.] leguest-worker program and the legalization of immigrants already in the country slowed. In summer 2003, there were several new proposals for a migration agreement with Mexico to legalize the status of currently unauthorized workers and allow some to earn immigrant status by working and paying taxes in the United States. There is little agreement, however, on what impacts such a program would have on California’s farm labor market. We used a unique database to examine farm employment trends in California agriculture. The data suggests that: about three individuals are employed for each year-round equivalent job, helping to explain low farm worker earnings; there was a shift in the 1990s from crop farmers hiring workers directly to farmers hiring via farm labor contractors ; and there is considerable potential to improve farm labor market efficiency,growing bags by using a smaller total workforce with each worker employed more hours and achieving higher earnings.

California employers who pay $100 or more in quarterly wages are required to obtain an unemployment insurance reporting number from the California Employment Development Department . The EDD then assigns each employer or reporting unit a four-digit Standard Industrial Classification or, since 2001, a six-digit North American Industry Classification System code that reflects the employer’s major activity . Major activities are grouped in increasing levels of detail; for example, agriculture, forestry and fisheries are classified as a major industrial sector and, within this sector, SIC 01 is assigned to crops, 017 to fruits and nuts and 0172 to grapes. We defined “farm workers” as unique Social Security numbers reported by farm employers to the EDD, and then summed their California jobs and earnings. This enabled us to answer questions such as how many farm and non-farm jobs were associated with a particular SSN or individual in 1 year, and in which commodity or county a person had maximum earnings. We adjusted the raw data before doing the analysis. Farm employers have reported their employees and earnings each quarter since 1978, when near universal UI coverage was extended to agriculture. Although it is sometimes alleged that farm employers, especially FLCs, do not report all their workers or earnings, there is no evidence that under reporting of employees or earnings is more common in agriculture than in other industries that hire large numbers of seasonal workers, such as construction. We excluded from the analysis SSNs reported by 50 or more employers in 1 year . We also excluded wage records or jobs that had less than $1 in earnings and jobs, or that reported earnings of more than $75,000 in one quarter. These adjustments eliminated from the analysis 2,750 SSNs, 62,571 wage records or jobs and $803 million in earnings. These exclusions were about 0.25%, 2.7% and 6.1% of the totals, respectively, and are documented more fully in Khan et al. .

There is no single explanation for the outlier data we excluded. In some cases, several workers may share one SSN, while in others our suspicion that a SSN had “too many” jobs may represent data-entry errors. During the 1990s, the Social Security Administration cleaned up SSNs, including threatening to fine and reject tax payments from employers with too many mismatches between SSNs and the names associated with those SSNs, which should have reduced the number of SSNs reported by employers. We think the rising number of SSNs reflectsmore individuals employed in agriculture, not more noise in the data.Agricultural employment can be measured in three major ways: at a point in time, as an average over time or by counting the total number of individuals employed over some period of time. In the non-farm labor market the three employment concepts yield similar results. If 100 workers are employed during each month and there is no worker turnover from month to month, then point in time, average and total employment is 100. However, agricultural employment during the six summer months may be 150, versus 50 during the six winter months, meaning that point, average and total employment counts differ. We began with all SSNs reported by agricultural employers , summed the jobs and earnings of these SSNs within each SIC code, and assigned each SSN to the four-digit SIC code in which the worker had the highest earnings. This means that a SSN reported by a grape employer as well as by an FLC would be considered a grape worker if his highest-earning job was in grapes. The number of individuals or unique SSNs reported by California agricultural employers has been stable over the past decade — 907,166 in 1991, 966,593 in 1996 and 1,086,563 in 2001 .Farm workers had a total of 1.5 million farm jobs in 1991, 1.7 million in 1996 and 1.8 million in 2001. One-quarter also had at least one non-farm job — about 407,000 workers were both farm and non-farm workers in 1991, 453,000 in 1996 and 697,000 in 2001 . The total California earnings of persons employed in agriculture were $11.1 billion in 1991, $12.0 billion in 1996 and $15.8 billion in 2001 . The share of total earnings for farm workers from agricultural employers was 77% in 1991, 77% in 1996 and 71% in 2001, indicating that in the late 1990s, farm workers tended to increase their supplemental earnings via non-agricultural jobs.

Average earnings per job were highest in livestock, $13,800 per job in 2001. There was little difference between average earnings per job in agricultural services and crops . Average earnings per job were higher for the non-farm jobs of agriculture workers than for agriculture jobs .In 2001, California’s farm workers held 2.5 million jobs, including 1.8 million jobs with agricultural employers. These agricultural jobs included 630,000 in crops, 69,000 in livestock and 1.1 million in agricultural services. The agricultural services sector includes both farm and non-farm activities, such as veterinary and lawn and garden services; FLCs accounted for 70% of the employees reported by farm agricultural services. Fruits and nuts accounted for 53% of the crop jobs, dairy for 39% of the livestock jobs and FLCs for 58% of the agricultural services jobs. The major change between 1991 and 2001 was the drop of 54,000 jobs in crop production and increase of 313,000 jobs in agricultural services. We placed SSNs in the detailed commodity or SIC code that reflected the maximum reported earnings for the worker,nursery grow bag and considered workers to be primarily employed in the SIC with maximum earnings. In 2001, there were 877,000 primary farm workers, and they included 322,000 reported by crop employers, 50,000 reported by livestock employers and 504,000 reported by agricultural service employers. Fruit and nut employers accounted for 47% of the crop-reported workers, dairy for 40% of the livestock-reported workers and FLCs for 44% of the agricultural services–reported workers. The major change between 1991 and 2001 was the increase in number of SSNs with their primary job in agriculture — from 758,000 to 877,000. There was a slight drop in the number of workers reported by crop employers, a slight increase in livestock workers and a sharp 135,000 increase in agricultural services workers, anchored by a 59,000 increase in workers reported by FLCs in 2001. Most farm workers had only one job. In 2001, 53% of the SSNs were reported by only one employer to the EDD, 26% were reported twice, 12% three times, 5% four times and 4% five or more times. During the 1990s, about 65% of farm workers were reported by one agricultural employer only, 17% to 21% by two agricultural employers, 5% by at least two agricultural employers and one non-farm employer, and 9% to 12% by one farm and one non-farm employer. In the three-digit SIC codes representing more detailed commodity sectors, 60% to 83% of the employees had only one job. For example, in 2001 79% of the employees reported by dairy farms had one dairy farm job, while 7% also had a second agricultural job — 3% had a dairy job, a second farm job and a non-farm job, and 11% had a non-farm job in addition to the dairy job. About two-thirds of the employees of FLCs and farm management companies had only jobs with one such employer; 22% had another farm job; 6% had an FLC job, another farm job and a non-farm job; and 6% had a non-farm job in addition to the FLC job. Even more detailed four-digit SIC codes showed the same pattern: the commodities or SICs most likely to offer year-round jobs such as dairies and mushrooms had 70% to 80% of employees working only in that commodity, while commodities or SICs offering more seasonal jobs, such as deciduous tree fruits and FLCs, had 53% to 63% of employees working only in that commodity.

At the four-digit, SIC-code level, the five largest SICs accounted for about 45% of the agricultural wages reported.Agricultural employers paid a total of $11 billion in wages in 2001, an average of $10,200 per worker . Earnings were highest for the 64,000 workers primarily employed in livestock; they averaged $14,800, followed by those primarily employed by crop employers and those employed by agricultural farm services, custom harvesters and FLCs . There was considerable variation in earnings among workers in agricultural farm services: workers in soil preparation services averaged $21,100 in 2001, versus $12,700 for crop preparation services for market and $4,400 for FLC employees. The average earnings of primarily farm workers varied significantly, even within detailed four-digit SIC codes — in most cases, the standard deviation exceeded the mean wage . Median earnings were generally less than mean earnings, reflecting that higher wage supervisors and farm managers pulled up the mean. If the workers in detailed commodities are ranked from lowest-to-highest paid, the lowest 25% of earners in an SIC category generally earned less than $4,000 a year. For example, among workers primarily employed in vegetables and melons in 2001 , the first quartile or 25th percentile of annual earnings was $3,000. This reflects relatively few hours of work — if these workers earned the state’s minimum wage of $6.25 an hour in 2001, they worked 480 hours. The 25th percentile earnings cutoff was lowest for those employed primarily by FLCs, only $634, suggesting that FLC employees receiving the minimum wage worked 101 hours. The highest 25th percentile mark was in mushrooms , $9,491, which reflects 1,519 hours at minimum wage. The 75th percentile marks the highest earnings that a non-supervisory worker could normally expect to achieve — 75% of workers reported earning less than this amount and 25% earned more. The 75th percentile varied widely by commodity: $6,172 for those primarily employed by FLCs, $10,572 for those in grapes and $29,465 for those in mushrooms.The number of individuals and jobs reported by agricultural employers increased in the 1990s, reflecting increased production of labor-intensive fruit and vegetable crops and, the data suggests, more farm workers each worked a fewer number of hours. With the state’s minimum wage at $6.25 per hour after Jan. 1, 2001 , the earnings reported by employers suggest that most farm workers are employed fewer than 1,000 hours per year . FLCs increased their market share in the 1990s, but dependence on them varied by commodity. For example, FLCs rather than citrus growers reported many citrus workers, while dairy employers reported most dairy workers. FLCs are associated with low earnings, which suggests few hours of work — the median earnings reported by FLCs for their employees in 2001 were $2,650, or 400 hours if workers earned the state’s $6.25 minimum wage. California’s farm labor market has large numbers of workers searching for seasonal jobs; FLCs are matching an increasing share of these workers with jobs, resulting in lower earnings for FLC employees. Workers who avoid FLCs experience higher earnings in agriculture or in the non-farm labor market. If FLCs are most likely to hire recently arrived and unauthorized workers, as the National Agricultural Worker Survey suggests, FLCs serve as a port of entry for immigrant farm workers. The impact of guest workers, legalization and earned legalization will depend on the details of any new program. If the status quo continues, the percentage of unauthorized workers is likely to rise. Alternatively, if there were a legalization program, farm workers might more quickly exit the farm workforce. However, an earned legalization program could slow this exit if workers were required to continue working in agriculture to earn full legal status. The next step in this analysis is to examine the mobility of individual farm workers over time and geography, examining where workers migrate during 1 year and patterns of entrance to and exit from the farm workforce . Do farm workers who increase their earnings by moving to non-farm jobs stay in non-farm jobs, or do they sometimes return to agriculture? Answers to these questions will help to determine the trajectory of the farm labor market.