Is There an Alternative Pathway for Starch Synthesis? - NCBI

Plant Physiol. (1992) 100, 560-564 0032-0889/92/1 00/0560/05/$01 .00/0

Is There

an

Received for publication February 18, 1992

Accepted May 15,

1992

Alternative Pathway for Starch Synthesis?' Thomas W. Okita

Institute of Biological Chemistry, Washington State University, Pullman, Washington ABSTRACT

15, 17), indicating that the basic enzymology of starch synthesis is the same in amyloplasts as in chloroplasts. The flow of carbon and the exchange of metabolites between the amyloplast and cytosol in developing sink organs, however, are not identical to the processes established for leaves (1, 6,

In leaf tissue, carbon enters starch via the gluconeogenesis pathway where D-glycerate 3-phosphate formed from CO2 fixation is converted into hexose monophosphates within the chloroplast stroma. In starch-containing sink organs, evidence has been obtained indicating that the flow of carbon into starch follows a different pathway whereby hexose monophosphates formed from sucrose are transported into the amyloplast, a plastid specialized in starch accumulation. In both chloroplasts and amyloplasts, the formation of ADPglucose, the substrate for starch synthase, is controlled by the activity of ADPglucose pyrophosphorylase, a key regulatory enzyme of starch synthesis localized in the plastid. Recently, an alternative pathway of starch synthesis has been proposed in which ADPglucose is synthesized from sucrose and transported directly into the plastid compartment, where it is used for starch synthesis. On the basis of the biochemical phenotypes exhibited by various plant mutants with defined genetic lesions, it is concluded that ADPglucose pyrophosphorylase is essential for starch synthesis, whereas the alternative pathway has only a minor role in this process.

7, 9, 17, 19). An alternative pathway of starch synthesis recently has been proposed based on the capacity of both chloroplasts and amyloplasts to transport ADPGlc (16 and refs. cited therein). In this review, I first summarize the latest developments in starch biosynthesis and then discuss whether the operation of this proposed alternative pathway of starch synthesis is compatible with our present knowledge of the biochemistry and genetics of starch biosynthesis.

STARCH FORMATION OCCURS VIA ADPGIc PYROPHOSPHORYLASE Extensive evidence indicates that the biochemical events leading to the formation of ADPGlc and its subsequent utilization for starch synthesis are restricted to the chloroplasts (reviewed in refs. 15 and 17). The more recent isolation and study of mutants defective in carbon metabolism also support this view. Caspar et al. (4) obtained a null mutant of Arabidopsis thaliana for the chloroplastic PGM that was highly defective in starch synthesis, accumulating less than 2% of the normal levels. The starchless phenotype of this null plastidic PGM mutant supports the prevailing view that the bulk of, if not all, carbon flow into starch occurs via the gluconeogenesis pathway within the plastid compartment. Kruckeberg et al. (11) examined mutant plants of Clarkia xantiana containing reduced activities of the cytosolic PGI (64%, 36%, and 18% of wild-type levels) or chloroplastic PGI (75% and 50% of wild type) to assess the effect of enzyme levels on carbon fluxes toward starch and sucrose synthesis under saturating and limiting light conditions. Decreased levels of the plastid enzyme had very little influence on starch and sucrose synthesis in low light. In saturating light, however, starch synthesis was suppressed, whereas little effect on sucrose synthesis was observed. Conversely, reduction in the levels of the cytosolic enzyme caused an increase in starch synthesis with a corresponding decrease in sucrose synthesis; this response was more evident under low light intensity than under saturating conditions. These observations with the plastid mutant lines reinforce the view that the events leading to carbon flow into starch are restricted to the plastid. Moreover, when sucrose synthesis is depressed

The events that lead to the flow of carbon into starch and their regulation have been well established for chloroplasts (reviewed in ref. 17). In contrast, less is known about the biochemical events that lead to starch synthesis in the amyloplasts of developing sink organs (1). This organelle, containing one or more starch granules, is delimited by a double envelope and is usually devoid of intracellular membranes. Because the amyloplast is dependent on the cytoplasm for both energy and carbon, the biochemistry of this organelle is likely to be distinct from that exhibited by the autotrophic (ATP-generating, C02-fixing) chloroplasts of leaf tissue. Recent studies have shown that the allosterically regulated ADPGlc2 pyrophosphorylase, as well other enzymes involved in starch synthesis, are localized in the amyloplasts (1, 10, l Supported in part by Department of Energy grant DE-FG0687ER13699, The Rockefeller Foundation, and Project 0590, College of Agriculture and Home Economics, Washington State University, Pullman, WA 99164. 2 Abbreviations: ADPGlc, ADPglucose; 3-PGA, D-glycerate 3phosphate; PGI, phosphoglucoisomerase; PGM, phosphoglucomutase; SS-ADPGlc, sucrose synthase-dependent synthesis of ADPGlc; UDPGlc, UDPglucose; Glc 1-P, glucose 1-phosphate; Glc 6-P, glucose 6-phosphate.

560

THE FLOW OF CARBON INTO STARCH

in the cytosolic mutants, more of the carbon is directed away from the cytosol and rerouted toward the synthesis of starch in the chloroplast. Direct evidence has been obtained showing that most, if not all, of the ADPGlc formed is controlled by the action of ADPGlc pyrophosphorylase. Lin et al. (12) isolated two Arabidopsis lines mutated at the Adgl and Adg2 loci that were defective in starch synthesis and ADPGlc pyrophosphorylase activity. Plants of the adgl line, which accumulated very little leaf starch, were devoid of both the large and small subunits of the ADPGlc pyrophosphorylase as viewed by immunoblot analysis, whereas plants of the line adg2, which accumulated 40% as much starch, were deficient in the large subunit of ADPGlc pyrophosphorylase. Therefore, the absence or depression in the levels of starch synthesis can be attributed to a direct causal relationship between defects in the expression of the structural genes for ADPGlc pyrophosphorylase and the concomitant lower amounts of enzyme activity. The rate of CO2 incorporation into starch by isolated chloroplasts is directly correlated with 3-PGA levels and inversely correlated with Pi levels (17). This is consistent with the in vitro evidence that ADPGlc synthesis is controlled by the activation and inhibition of ADPGlc pyrophosphorylase via these metabolites. Recently, a starch-deficient mutant of Chlamydomonas reinhardtii contained a defective ADPGlc pyrophosphorylase that was less responsive to allosteric activation by 3-PGA and inhibition by Pi. This provides direct in vivo evidence that the allosteric activation by 3-PGA of chloroplastic ADPGlc pyrophosphorylase is essential for maximum starch synthesis (2). A direct role for the allosteric regulation of the amyloplast enzyme in starch synthesis has yet to be resolved. However, Hnilo and Okita (8) have shown that when tuber slices were incubated in the presence of mannose, an effective sequestration agent of intracellular Pi, the incorporation of "4C-sucrose into starch was enhanced by 50%. These results suggest that the activity of the amyloplast enzyme is also modulated by intracellular Pi levels. Genetic studies also indicate a major role for ADPGlc pyrophosphorylase in starch synthesis of nonphotosynthetic developing sink organs. Maize mutations at two unlinked loci, Shrunken-2 (Sh2) and Brittle-2 (Bt2), result in a 60% reduction in starch levels in the endosperm with corresponding decreases in ADPGlc pyrophosphorylase activities of 66% and 63%, respectively (17). Recent biochemical and molecular studies (3, 15, 17) have shown that Sh2 and Bt2 encode ADPGlc pyrophosphorylase subunits of 54 kD and 51 kD, respectively, and that both subunits are required for maximum enzyme activity and starch synthesis. ADPGlc pyrophosphorylase also seems to have a major role in starch synthesis in pea embryos. Mature seeds recessive at the Rb locus have reduced levels of starch and elevated lipid and sucrose levels and contained less than 8% of the normal ADPGlc pyrophosphorylase levels (18). Although the nature of this genetic defect has not been determined at the molecular level, it is likely that the reduction of starch in the rb mutant pea lines is the result of a defect in ADPGlc pyrophosphorylase activity and not in any other major enzyme activity involved in starch synthesis. In potato tuber, a 'reverse genetics' approach was used to evaluate the role of ADPGlc pyrophosphorylase in starch

561

metabolism. Muller-Rober et al. (13) transformed potato with 'antisense' constructs for the small subunit of the heterotetrameric tuber ADPGlc pyrophosphorylase (14). These workers observed the almost complete absence of starch formation in the developing tubers of these transgenic plants and thereby clearly demonstrated an essential role for ADPGlc pyrophosphorylase in starch synthesis. THE FLOW OF CARBON INTO STARCH

In actively photosynthesizing leaves, the events leading to carbon flow into starch are restricted to the chloroplasts (reviewed in ref. 17). The 3-PGA formed during CO2 fixation is readily transformed into hexose monophosphates via gluconeogenesis within the stroma where they can serve as substrates for ADPGlc formation (Fig. 1). In young developing leaves that serve as sinks, sucrose is first broken down by an alkaline invertase or by the successive action of sucrose synthase and UDPGlc pyrophosphorylase in the cytoplasm Hexose pathway

ATP

Glc 6-P

do

Gic 1-P

-

Gic 6-P

.

Gic 1-P

IT

UDPG

PP

fructose

ADPG

sucrose

starch

UDP

cytosol

amyloplast SS-ADPG pathway

ADPG

-

-ADPG

fructose

starch sucrose

ADP

Figure 1. The flow of carbon into starch: two hypotheses. The upper panel depicts the transport of hexose monophosphates into amyloplasts and its conversion into starch; the bottom panel depicts the SS-ADPGlc pathway. In the upper panel, Glc 1-P is the preferred metabolite transported in wheat endosperm amyloplasts (1), whereas Glc 6-P is specifically transported in pea embryo (7) and root amyloplasts (6). In both instances, ADPGlc pyrophosphorylase is required for the formation of ADPGlc, the substrate for starch synthase. In contrast, the SS-ADPGlc pathway utilizes the cytosoliclocalized sucrose synthase for the formation of ADPGlc, which is subsequently transported into the plastid for starch synthesis.

562

OKITA

(1). The hexose sugars are then metabolized via glycolysis into three carbon intermediates that are transported via the Pi translocator (6) into the chloroplasts where they can be metabolized further for starch synthesis. Although several studies (reviewed in refs. 15 and 17) have suggested the transport of C3 intermediates into isolated amyloplasts, this does not appear to be the main pathway by which carbon enters starch synthesis. Keeling et al. (9) examined the extent of redistribution of 13C between carbons 1 and 6 in the starch glucosyl moiety when developing wheat endosperm were fed [1_-3C]- or [6-13C]glucose or -fructose and found that there was only partial redistribution (12-20%) of '3C into starch between the Cl and C6 atoms (9). Because the redistribution of label in the glucose moiety of starch would be expected to be more extensive if carbon flow into starch occurred by the C3 pathway via triosephosphate isomerase, Keeling et al. (9) suggested that hexose monophosphates, and not triosephosphates, are the most likely candidates for entry into the amyloplast. The observation of direct import of hexose units into amyloplasts of potato tubers, fava beans (19), maize endosperm, and suspension cells of Chenopodium rubrum (5) led to similar conclusions. Several studies (1, 6, 7) have demonstrated the direct transport of hexose monophosphates as well as other metabolites into isolated amyloplasts and plastids from nonphotosynthetic tissue. Tyson and ap Rees (reviewed in ref. 1) observed that only Glc 1-P served as an effective precursor for starch synthesis by intact wheat endosperm amyloplasts and that this labeling of starch was dependent on the degree of intactness of this organelle. A somewhat different specificity of transport was exhibited by pea amyloplasts (7), which readily took up and incorporated Glc 6-P, but not Glc 1-P, into starch at rates comparable with those measured in vivo (7). The incorporation of 14C label from Glc 6-P was dependent on the presence of ATP and on the integrity of the amyloplasts. In contrast, an amyloplast preparation from developing maize endosperm preferred dihydroxyacetone phosphate for uptake and incorporation into starch, although Glc 6-P and fructose 1,6bisphosphate were also transported, albeit at lower levels (reviewed in ref. 10). Heldt et al. (6) have proposed that amyloplasts and other plastids of nonphotosynthetic tissue contain a modified Pi translocator with a relatively broad metabolite specificity as inferred from the transport capabilities displayed by plastids from pea root. These plastids contain a Pi translocator similar to the type observed in chloroplasts in their ability to transport Pi, dihydroxyacetone phosphate, and 3-PGA in a counter-exchange mode, but the root-specific Pi translocator is also able to transport Glc 6-P but not Glc 1-P. Because pea root plastids also appear to lack a plastid fructose 1,6-bisphosphatase, Heldt et al. (6) reasoned that the transport of Glc 6-P would not only allow entry of carbon into starch but would also provide a source of reducing power via the oxidative pentose phosphate pathway required for nitrite reduction. In light of the above discussion, it appears that the flow of carbon from the cytoplasm into starch accumulated by the amyloplasts does not involve C3 intermediates; instead, carbon is directed into starch in a more direct route (Fig. 1). Hexose monophosphates formed from the successive cleav-

Plant Physiol. Vol. 100, 1992

ages of sucrose and UDPGlc (1) are directly transported into the amyloplasts, where they are converted into starch via the combined action of ADPGlc pyrophosphorylase and starch synthase. THE ALTERNATIVE PATHWAY OF STARCH SYNTHESIS

An alternative pathway of carbon flow into starch has been proposed by Akazawa et al. (16 and refs. cited therein). These authors (16) have shown that both chloroplasts and amyloplasts contain an ADP/ATP translocator that also appeared to be capable of transporting radiolabeled ADPGlc, but not UDPGlc, with a percentage of the radioactivity incorporated into starch. The transport of ADPGlc and the subsequent incorporation of the glucosyl moiety into starch led these authors to suggest the direct synthesis of ADPGlc from sucrose and subsequent incorporation into starch. In this hypothesis, ADPGlc is synthesized directly from sucrose via the action of a cytosolic sucrose synthase and then transported across the amyloplast membranes where it could be utilized by starch synthase, a process labeled as the SSADPGlc pathway (Fig. 1). The proposed SS-ADPGlc pathway has the added advantage over the commonly accepted routes of carbon flow into starch in being more energy efficient and is consistent with three lines of evidence. First, sucrose synthase can use ADP as the glucosyl acceptor to form ADPGlc (16). Second, a mutation in the major maize endosperm-specific form of sucrose synthase, Sh 1, results in a decrease in overall starch levels. Third, the SS-ADPGlc pathway is consistent with the asymmetric distribution of carbon in the glucose moiety of sucrose, starch, and phosphate esters when green leaves are exposed to "4CO2 under photosynthetic conditions or are fed [Cl-11C]glucose, or when rice grains are radiolabeled with [(U-'4C)glucose, [6-3H-fructose]-sucrose (see 16 and refs. cited for a more complete discussion). It should be pointed out, however, that these studies only indirectly support the SS-ADPGlc pathway; moreover, they cannot discriminate between the various hypotheses proposed for the flow of carbon into starch, i.e. they are also not inconsistent with the consensus pathways of starch synthesis. An important distinction between the SS-ADPGlc hypothesis and the consensus pathways of starch synthesis is the subcellular location of ADPGlc formation that is required for starch synthesis. Evidence obtained from more recent studies, discussed above and elaborated below with regard to this alternative pathway, demonstrate that the sugar nucleotide utilized by starch synthase is synthesized in the plastid. If the SS-ADPGlc pathway were a major route, starch synthesis would be dependent on sucrose synthesis and ADPGlc pyrophosphorylase would have a minor role in ADPGlc formation and, hence, in starch production (Fig. 1). These predictions are not supported by the results obtained from the biochemical-genetic analyses of plant mutants. As discussed earlier, Clarkia mutants defective in the cytosolic PGI accumulated less sucrose under limiting light conditions (11). If ADPGlc were formed from sucrose via the action of sucrose synthase (Fig. 1), then starch levels should also be depressed. Instead, increased levels of starch were observed, indicating that the restriction of carbon into sucrose in the

THE FLOW OF CARBON INTO STARCH

cytosol caused a rerouting of this carbon into starch in the plastid (11). Results of the study of mutants defective in chloroplast PGI are also inconsistent with the SS-ADPGlc pathway, which predicts that a defect in plastid PGI would have no significant effect on starch synthesis. Instead, reduced starch but not sucrose levels were observed in the plastid PGI mutants, indicating that the ADPGlc formation required for starch synthesis was restricted to the chloroplasts (11). The most compelling evidence for a direct role of ADPGlc pyrophosphorylase in ADPGlc formation is also derived from mutant analysis. The Arabidopsis mutants adgl and adg2 (12, 17), the maize endosperm mutants sh2 and bt2 (3, 15, 17), the transgenic potato plants harboring antisense DNA constructs to an ADPGlc pyrophosphorylase gene (13), and a Chlamydomonas mutant (2) have depressed levels of starch and deficient levels of ADPGlc pyrophosphorylase activities due to specific losses of one or both of the ADPGlc pyrophosphorylase subunits or reduction in allosteric activation by 3-PGA. Likewise, a direct correlation was also observed between the levels of starch and ADPGlc pyrophosphorylase enzyme activity and not between starch and sucrose synthase activity in the pea embryo mutant rb (18). These studies demonstrate direct causal relationships between decreased levels in gene products, of enzyme activities, and of starch levels, and therefore substantiate the view that ADPGlc pyrophosphorylase is absolutely required for starch synthesis. Therefore, one can conclude that, on the basis of the results of these studies (2, 3, 12, 13, 15, 17), the ADPGlc used for starch synthesis is formed in the plastid via the activity of ADPGlc pyrophosphorylase and that the SS-ADPGlc pathway plays little, if any, role in this process.

hexose monophosphates across the amyloplast membranes? With regard to the latter, the permeability properties of the isolated amyloplast are completely dependent on the degree of intactness, but present methods to assess amyloplast integrity rely on metabolic activities and enzyme latency studies (1) and on cytological examination (16). These criteria, however, are insufficient to assess the extent of membrane damage and cytoplasmic contamination that can occur during amyloplast isolation (see ref. 1 for more discussion). In many respects, the current definition of amyloplast intactness is at a stage comparable with the time preceding the development of cellular fractionation techniques for the successful isolation of intact functional chloroplasts (20). Once it was recognized that the ability to fix CO2 was dependent on chloroplast intactness, 'Class A' chloroplasts were routinely obtained (20). A similar biochemical test to assess the intactness of the amyloplast must be implemented before the transport properties of this organelle can be accurately studied. In the case of pea embryo amyloplasts (7) and pea root plastids (6), the uptake of metabolites and their incorporation into starch and as a source of reducing power for nitrite reduction, respectively, may be useful criteria to establish the degree of intactness and, hence, facilitate efforts that will lead to a better understanding of the biochemistry of this specialized, starchcontaining organelle. ACKNOWLEDGMENTS I thank Professors Frank Loewus and Gerald Edwards and Dr. Mirta Sivak for their helpful discussions and editorial suggestions on this minireview. I apologize to the many workers in the area of carbon metabolism whose publications were not referenced in the manuscript. The omission was not due to an oversight but rather dictated by space limitations.

CONCLUSIONS

Overwhelming biochemical evidence gathered in the past years has documented the events leading to starch production in chloroplasts. Carbon flow into starch is mediated by gluconeogenesis, and one of the pivotal reactions, ADPGlc formation, is controlled by the allosteric activity of ADPGlc pyrophosphorylase (15, 17). Although several studies have suggested the presence of a similar pathway in developing sink organs, sufficient evidence has been gathered indicating that most of the carbon is routed into starch by the direct transport and utilization of hexose monophosphates (1, 6, 7). Likewise in several developing sink organs, evidence has been obtained from biochemical and genetic studies (1, 2-4, 8, 11-13, 17, 18) that conclusively demonstrates that ADPGlc pyrophosphorylase activity is absolutely essential for starch synthesis. Although the alternative SS-ADPGlc pathway is a potential source of ADPGlc, there is no evidence that directly supports the formation of ADPGlc in the cytoplasm and its incorporation into starch in leaf tissue and in developing sink organs. On the contrary, results of biochemical-genetic studies (4, 11, and refs. cited in 15, 17) discount the utilization of the SS-ADPGlc pathway as a major route leading to starch synthesis. Questions that remain unresolved include to what extent does ADPGlc synthesis occur in the cytoplasm and what are the rates of transport of ADPGlc, triosephosphates, and

563

1.

2.

3. 4.

5.

6.

7. 8.

LITERATURE CITED ap Rees T, Entwistle G (1989) Entry into the amyloplasts of carbon for starch synthesis. In CD Boyer, JC Shannon, and RC Hardison, eds, Physiology, Biochemistry, and Genetics of Nongreen Plastids. American Society of Plant Physiologists, Rockville, MD, pp 49-62 Ball S, Marianne T, Dirick L, Fresnoy M, Delrue B, Decq A (1991) A Chlamydomonas reinhardtii low-starch mutant is defective for 3-phosphglycerate activation and orthophosphate inhibition of ADP-glucose pyrophosphorylase. Planta 185: 17-26 Bhave MR, Lawrence S, Barton C, Hannah LC (1990) Identification and molecular characterization of Shrunken-2 cDNA clones of maize. Plant Cell 2: 581-588 Caspar T, Huber SC, Somerville C (1986) Alterations in growth, photosynthesis and respiration in a starch mutant of Arabidopsis thaliana (L.) Heynh deficient in chloroplast phosphoglucomutase activity. Plant Physiol 79: 11-17 Hatzfeld W-D, Stitt M (1990) A study of the rate of recycling of triose phosphates in heterotrophic Chenopodium rubrum cells, potato tubers, and maize endosperm. Planta 180: 198-204 Heldt HW, Flugee U-I, Borchert S (1991) Diversity of specificity and function of phosphate translocators in various plastids. Plant Physiol 95: 341-343 Hill LM, Smith AM (1991) Evidence that glucose 6-phosphate is imported as the substrate for starch synthesis by the plastids of developing embryos. Planta 185: 91-96 Hnilo J, Okita TW (1989) Mannose feeding and its effect on starch synthesis in developing potato tuber discs. Plant Cell Physiol 30: 1007-1010

564

9. Keeling PL, Wood JR, Tyson RH, Bridges IG (1988) Starch biosynthesis in developing wheat grain. Plant Physiol 87: 311-319 10. Kim WT, Franceschi VR, Okita TW, Robinson N, Morell M, Preiss J (1989) Immunocytochemical localization of ADPglucose pyrophosphorylase in developing potato tuber cells. Plant Physiol 91: 217-220 11. Kruckeberg AL, Neuhaus HE, Feil R, Gottlieb LD, Stitt M (1989) Decreased-activity mutants of phosphoglucose isomerase in the cytosol and chloroplast of Clarkia xantiana. Biochem J 261: 457-467 12. Lin T-P, Caspar T, Somerville CR, Preiss J (1988) A starch deficient mutant of Arabidopsis thaliana with low ADPglucose pyrophosphorylase activity lacks one of the two subunits of the enzyme. Plant Physiol 88: 1175-1181 13. Muller-Rober B, Sonnewald U, Willmitzer L (1992) Inhibition of the ADP-glucose pyrophosphorylase in transgenic potatoes leads to sugar-storing tubers and influences tuber formation and expression of tuber storage protein genes. EMBO J 11: 1229-1238 14. Okita TW, Nakata P, Anderson JM, Sowokinos J, Morell M, Preiss J (1990) The subunit structure of potato tuber ADPglucose pyrophosphorylase. Plant Physiol 93: 785-790

OKITA

Plant Physiol. Vol. 100, 1992 15. Okita TW, Nakata PA, Ball K, Smith-White BJ, Preiss J (1992) Enhancement of plant productivity by manipulation of ADPglucose pyrophosphorylase. In JP Gustafson, ed, Gene Conservation and Exploitation, Stadler Genetics Symposium, Plenum Press, New York (in press) 16. Pozueta-Romeo J, Ardila F, Akazawa T (1991) ADP-glucose transport by the chloroplast adenylate translocator is linked to starch biosynthesis. Plant Physiol 97: 1565-1572 17. Preiss J (1991) Biology and molecular biology of starch synthesis and regulation. In BJ Miflin, ed, Oxford Surveys of Plant Molecular and Cellular Biology, Vol 7, Oxford University Press, Oxford, pp 59-114 18. Smith AM, Bettey M, Bedford ID (1989) Evidence that the rb locus alters the starch content of developing pea embryos through an effect on ADP glucose pyrophosphorylase. Plant Physiol 89: 1279-1284 19. Viola R, Davies HV, Chudeck AR (1991) Pathways of starch and sucrose biosynthesis in developing tubers of potato (Solanum tuberosum L.) and seeds of faba bean (Vicia faba L.): elucidation by 13C-NMR spectroscopy. Planta 183: 202-208 20. Walker D (1975) Plastids and intracellular transport. In A Prison, MH Zimmermann, eds, Encyclopedia of Plant Physiology, New Series, Vol 3. Springer-Verlag, Berlin and New York, pp 85-136