text
stringlengths 1
6.27k
| id
int64 0
3.07M
| raw_id
stringlengths 2
9
| shard_id
int64 0
0
| num_shards
int64 16
16
|
|---|---|---|---|---|
reading processes.Instead, because researchers agree that direction of attention and cognitive procedures in reading influence fixations (Paterson et al. 2011;Paulson 2005;Miellet et al. 2009), it was hoped that readers might recall their thought processes associated with regressions, pauses, refixations and swift progress.Many readers did in fact recall thinking at points where their eye movement records reflected these features.For example, on noting his regression on the second word of the phrase 'isitsha esinezimbali' (a dish with flowers), a reader said he had first misread 'esinezimbali' as 'esinamazambane' (with potatoes) because he had associated 'isitsha' (dish) with food. Findings Although all participants had demonstrated their competence, eye movement records showed differences amongst them.In Figures 1-3, vertical sections of graph lines show the duration of each fixation and horizontal sections http://www.rw.org.zaOpen Access show movement to the left or right within a line of text and return sweeps to the left to begin a new line. Figure 1 shows the steady progress of an adept reader through almost three lines of text in approximately 6.5 seconds.The brief vertical line sections show the six or seven fixations he made per line, and the short horizontal line sections show his saccades as his point of focus moved through each line of text.There is one brief regression.Observation of his recorded point of focus moving over the text confirmed that he read many words in a single fixation and passed over some others without fixating on them, indicating a high degree of automaticity. Figure 2 and Figure 3 show the slower progress of two
| 1,200 |
55002387
| 0 | 16 |
slightly slower readers.Figure 2 shows very small movements of the point of focus to the right, indicating short saccades, suggesting a short span of recognition, with repeated fixations on some words. Figure 3 shows a similar rate of progress, but here the track shows long fixations at line beginnings and frequent regressions, as the reader 'reversed' in the text to have a second (or even third) look at words already passed.Figure 2 and Figure 3 show less instant word recognition than Figure 1. Textual features and automaticity As shown in Table 1, sentence length differed markedly across the texts, and reading speed decreased as sentence length increased.The Pearson Product Moment Correlation test showed a moderate, yet significant negative correlation between reading speed and sentence length (r = −0.413,n = 40, p = 0.01). RefixaƟon Readers' rate of fixations increased moderately but significantly as sentence length increased (r = 0.399, p < 0.05). Their rate of regressions also increased significantly with sentence length (r = 0.440, p < 0.01). Sentence length increases across the texts, with an average of 5.3 words per sentence in Text 1, 6.6 in Text 2, 10 in Text 3 and 16.7 in Text 4, showing that Text 3 and Text 4 (judged to be the more difficult of the four texts) were read more slowly, with more fixations and regressions than Texts 1 and 2. This could also be influenced by other less measurable differences, such as unfamiliar vocabulary or other complexities such as the need to infer. Readers' comments as they
| 1,201 |
55002387
| 0 | 16 |
watched the replay of the recorded movement of their focus reveal their perceptions of their own reading process at points where their progress slowed: SL: You see, you slowed down a lot here? NN: Yes I remember. SL: What was going on in your head? NN: 'Ezingaphambiyana nokugawula izihlahla, I think it is the same as the other one before this one.I was trying to understand, to read it and understand it at the same time … because it is talking about amadodana and now it's izihlahla so I am trying to put all of this thing together'. Illustrating Paulson's (2005) observation that readers regress to problem areas, in this case unexpected information, one reader spoke of her surprise upon reading of war between whites in South Africa (the Boer War): TM: I was not sure about the sentence because, really I never heard anything about the impi between amaNgisi namaBhunu.I only know the war between Blacks and Boers so … Mmh? Okay then to me it was something new. SL: So that is why you stuck there because you were thinking … Similarly, another reader exemplifies the observation of Paterson et al. (2011) that readers tend to regress on the second of related words in a sentence, perhaps because their similarity inhibits recognition (here the words are visually similar, though not related): SL: And you also regress there, on kwakuyiloluhlobo. CM: Ok there, I am not sure if it was this … ubuhholohholo because when I see that I thought there was something familiar. Average
| 1,202 |
55002387
| 0 | 16 |
word length did not differ much between the texts (Text 1: 7.65 letters, Text 2: 7.47 letters, Text 3: 8.31, Text 4: 7.49).Unsurprisingly, in view of this small difference, there was no significant correlation between word length and reading speed (r = −0.118). In the 400 words of the combined isiZulu texts used here, 62 three-letter strings occur 10 times or more and 11 fourletter strings occur more than 10 times (Decrypting Text 2013).Analysis of 5000 words of text from Isolezwe (http:// www.iol.co.za/isolezwe), a popular isiZulu newspaper, showed that in 400 words of text in this newspaper, 30 three-letter strings are likely to occur 10 times or more and 4 four-letter strings are likely to occur more than 10 times (Decrypting Text 2013).In comparison, analysis of 5000 words of text from Mercury (http://www.iol.co.za/ mercury), a popular English newspaper, showed that in 400 words, only 4 three-letter strings (the, and, -ing and -ent) occur 10 times or more, and no four-letter strings occur more than 10 times (Decrypting Text 2013).This comparison demonstrates frequent repetition of letter strings in isiZulu. Similarity between word forms hinders easy distinguishability between words in any language (Abadzi 2011), and the high recurrence of the same letter strings may hinder the development of automaticity.In the example above where a reader regressed on isitsha esinezimbali (dish with flowers) (Text2), after misreading it as isitsha esinamazambane (dish with potatoes) the reader's mistake showed signs of automaticity because he had predicted a word referring to food, and on sight of a long word containing the prefix
| 1,203 |
55002387
| 0 | 16 |
esina/e-followed by a stem containing z and mba, assumed the stem to be amazambane, a miscuing error because of the similarity between word forms.Almost immediately after this regression, the reader had trouble with the same letter combination again, regressing on ezimbalwa (several) after misreading it as izimbazo (axes). There is a strong, highly significant correlation (r = 0.743, p < 0.01) between saccade length and reading rate.A high reading rate is strongly associated with automaticity, and these figures show that the readers whose saccades are long are those who read swiftly.This is apparent in the graphs of readers' eye movements as shown above. Less strong, but still significant, was the negative correlation (r = −0.348,p < 0.05) between saccade length and longer sentences including unfamiliar vocabulary since Texts 3 and 4 included words that readers found obscure: SL: Ngezikhadlana -is that a construction you are familiar with?MS: Yes, izikhala is [means] spaces.So izikhadlana is smaller spaces 4 -but a lot of the language I am not used to. There was no significant correlation (r = 0.053) between the duration of readers' fixations and sentence length, indicating that readers did not increase the duration of their gaze when coping with longer sentences or words they found obscure.This indicates that slower reading rates on texts with longer sentences and obscure words were because of the higher number of fixations (including regressions) made by readers, rather than longer fixations. The graph in Figure 4 compares sentence length, reading rate, fixations (reduced by a factor of 10 to allow graphic
| 1,204 |
55002387
| 0 | 16 |
comparison) and regressions across the four texts.The trend line shows how reading rate decreased as sentence length increased and with it the rate of fixations and regressions. Word recognition Word recognition might not always coincide exactly with a reader's shifts of focus from word to word, but the number of fixations on a word and the duration of fixations are used by researchers to reflect the ease of word recognition (Rayner et al. 2007).Thus, the average number of fixations on a word by different readers and their average duration of fixation indicate how easily a word is recognised.On this basis, each 4.The reference is to a description of gaps in doorways allowing cold wind into houses.word in each of the four texts was coded according to the number of readers who skipped it, fixated on it once, refixated on it or regressed to it.These data were summarised to show which words appeared to have been instantly recognised in 80% or more of the recordings and which words appeared to have required cognitive work to decipher in 80% or more of the recordings. Instantly recognised (automaticised) words Of the words, 24.5% (98 of 400) appear to have been instantly recognised in 80% or more of the recordings.They are listed in Appendix 2. Average word length across the four texts used is 7.73 -significantly longer than the average word length amongst Europe's languages: 4.6 letters in English, 4.7 in Danish, 4.9 in Swedish, 5.6 in German and 7 in Finnish (Björnsson 1983).Average length of words processed in a
| 1,205 |
55002387
| 0 | 16 |
single brief fixation here is 6.5 letters, indicating that, as in other languages (Paterson et al. 2011), shorter words are more likely to be instantly recognised.It is predictable that short, unagglutinated, high-frequency words, such as nje (just), khona (there) and futhi (again), will be instantly recognised by readers, and possibly skipped.Here, only 3 of the 400 words (< 1%) were skipped in at least half of the recordings.These were uma 'if' or 'when', indlu 'house', and yomuzi 'of the home' and were skipped where their predictability was high.This is dramatically less than the 25%-30% of the words skipped by readers of English (Rayner 2009;White 2008), suggesting that word recognition differs substantially between isiZulu and English. Most instantly recognised words consisted of not more than two morphemes.Where words with more than two morphemes were instantly recognised, they were highfrequency word forms and/or were highly predictable such as um/fo/wa/bo in the context of '… wabonakala uDuda umfowabo omncane kaNkalimba…' ('…could be seen Duda, Nkalimba's little brother …'), where the capital letter of the first name and the context would have cued expectation of 'brother'.Thus, predictably, words with few morphemes appear to be recognised more easily than words with three or more morphemes.It is possible that familiar combinations of morphemes are instantly recognised, as in this instance: SZ-W: … I just discarded the first beginning part ezingama-part and just concentrated on the qhukwana part.I thought -What could that be? SL: Ok because those e-zi-nga-ma -they're all familiar parts of words?SZ-W: Yes Many of the words that were instantly recognised
| 1,206 |
55002387
| 0 | 16 |
at some points in the text were actively decoded at other points.This indicates that factors such as position, collocations, context and the process of meaning construction appear to influence recognition.Notably, no words at the start of a line of text were instantly recognised, even though some of them (uma, lapho and ukuthi) were instantly recognised elsewhere in the text.These refixations on words at the beginning of lines accords with the observation by Rayner et al. (2007) that fixations at line beginnings often involve corrective saccades, presumably made as readers ensure that they have found the right line.A reader's indefinite perception of this habitual process is clear in the instance below: Words requiring cognitive work to decipher In 80% or more of the recordings, 23.5% (94 of 400) were read with two or more fixations or at least one regression.These words are listed in Appendix 3.Only a few of them included uncommon vocabulary, so it appears that readers broke familiar words into their constituent parts to decipher them.Their average length is 11.1 letters (considerably longer than the overall average word length in these texts of 7.73 letters), and they consist of up to seven agglutinated morphemes, for example: ngi/nga/ka/li/hlangan/is/i, (I had not yet gathered together) kwa/ku/ngo/wo/ku/phatha (it was to take care of) a/yi/si/shiya/galo/lu/nye ('that were nine', or literally 'that which leaves digit of one'). In view of the length and amount of information in these multimorphemic words, it is unsurprising that nearly all the readers required more than one fixation to read them and predictable that readers
| 1,207 |
55002387
| 0 | 16 |
are likely to develop automaticised recognition of only those complex permutations that occur with unusually high frequency.The excerpt below illustrates the strategy of one of the swiftest readers, which appears to exemplify the reconstruction of the speech sounds and smallgrain size processing strategy described by Ziegler and Goswami (2005) in relation to transparent and consistent orthographies: SL: And how do you manage to unravel this because … there's one, two, three four, five six 'bits' [morphemes] in that word. How do you do it in your head? NN: I think I read it slowly like [mimicking] kwa-ku-ngo-woku-phatha then I understood it … I say the word, but silently. The need to process these complex words implies that this small-grain size processing strategy should be explicitly taught to learner readers of languages with orthographies similar to that of isiZulu. Readers' treatment of recurring words Interestingly, some words recognised immediately by nearly all skilled readers at some points in the text were read with multiple fixations and/or regressions at other points 5 .For example, the word khona (there) was instantly recognised in 100% of the recordings at three points in the text, but in only 60% of them at another point, where it refers to a pantry.Although the word khona is grammatically predictable at that point, a pantry might be an unexpected concept for some of these readers. Ubaba (father) and phela (here, an interjection like 'well') recur several times in the text, and were instantly recognised at some points but fixated on more than once or regressed
| 1,208 |
55002387
| 0 | 16 |
to at others.This suggests that factors other than the distinguishability and familiarity of particular words affect the ease with which they are recognised. Position on a line of text is salient, with a moderate but significant correlation (r = 0.399, p < 0.01) between being in the last half of a line of text and the likelihood of being instantly recognised. Seventeen of 22 instances of frequently recurring words that were recognised immediately in at least 70% of the recordings were in texts 1 and 2, which had less than two clauses per sentence on average.Only 5 frequently recurring words were instantly recognised in the more difficult texts 3 and 4, whose sentences averaged 2.7 and 4.2 clauses respectively. With regard to the position of words within a sentence, it appears that a relatively short high-frequency word is likely to be recognised with a single fixation (but not skipped) at the beginning of a sentence.It seems more likely to require multiple fixations when appearing further into a sentence because the number of readers instantly recognising a word is lower for words appearing near the end of sentences.Yet surprisingly, and in spite of the strong correlation found between lower reading rates and longer sentences, the correlation between a word's position in a sentence and the likelihood of it being instantly recognised does not reach significance (r = −0.083). Limitations The enquiry in the article is based on eye movements of the 10 swiftest readers of a sample of 33, recorded as they each read four texts.This narrow
| 1,209 |
55002387
| 0 | 16 |
selection ensured that only recordings demonstrating established automaticity were used, but more confidence could be placed in data from a larger sample.However, this exploratory study, the first of eye movements of competent readers of isiZulu, may be a useful starting point in the particular research question addressed. At 60× per second, the Visagraph equipment used has a lower sampling rate than some other systems, but its advantage is that it allows a close simulation of natural silent reading of continuous text, which was the particular focus of 5.For a table showing how words that recurred repeatedly in the texts were read, see Land, S., 2015, 'Reading isiZulu: Reading processes in an agglutinative language with a transparent orthography', PhD thesis, School of Arts, University of KwaZulu-Natal. this study.The effects of readers' awareness of being recorded as they read are unavoidable, and a limitation in all reading research involving readers' cooperation. Conclusions and implications The orthography of isiZulu is shaped by its agglutination, conjoined writing system, comparatively long, complex words and a high rate of recurring strings of particular letters.This makes it less conducive to automaticity than orthographies with short words and high heterogeneity amongst word forms.Data in the study suggest that readers of isiZulu rely on small-grain size orthographic units.In languages where readers do this, there may be less of a distinction between reading by phonological decoding and by automatic whole word recognition than there is in languages where readers rely on large-grain size orthographic units such as English.Perhaps a continuum starting with slow recognition of the
| 1,210 |
55002387
| 0 | 16 |
root and morpheme synthesis, progressing towards instant recognition and synthesis is a more plausible model for automaticity in these languages? Reading speeds of participants was highest when reading short sentences and familiar vocabulary.When reading longer sentences with less familiar vocabulary their reading rate decreased, whilst their rate of fixations and regressions increased significantly, but the duration of their fixations did not.Saccade length increased when reading text with familiar vocabulary, indicating a higher rate of automatic recognition. Although it is common across languages for reading rate to slow as textual complexity increases, the data in the study suggest that there are different word recognition processes in isiZulu and English.Almost 25% of words were recognised instantly, although less than 1% were skipped, suggesting that few isiZulu words have enough of a distinctive form to be predicted in the indistinct parafoveal view.Words that were instantly recognised tended to be shorter than average, suggesting that, as in other languages, short words are more easily automatically recognised.One implication of this for teaching reading in isiZulu is that predictive skills could be enhanced by ensuring that new readers recognise a core list of short words that do not agglutinate, such as conjunctives, and take cognisance of their semantic significance in the sentence.Another is that new readers should be given guided practice in spotting common word stems in complex words, as well as common combinations of morphemes. Readers fixated more than once on almost a quarter of the words in the texts.Predictably, these were longer (often comprising more than 11 letters) and more
| 1,211 |
55002387
| 0 | 16 |
complex than average, usually comprising three or more morphemes.Readers' clear recall of their thought processes at particular points of their reading suggests that, despite the subliminal nature of many elements of the reading process, the construction of mental representations of meanings are complex and powerful influences on how words are read, and that cognitive processing of words and even sentences sometimes lags behind perception of words in visual focus. Learner readers need to be enabled to disaggregate complex words into their constituent morphemes and to integrate their semantic value into their mental representation of the meaning of text as they read. FIGURE 1 :FIGURE 2 : FIGURE 1: Eye movements of a highly skilled reader. FIGURE 3 : FIGURE 3: Eye movements of a reader with repeated regressions. FIGURE 4 : FIGURE 4: Sentence length of texts and average scores. CM: I think it's -when I am starting the new line -and then I don't know if I was on the third or the second line, I'm not sure … SL: So, how do you make sure that you are on the right line? … CM: Like if I was on the third line or fourth line?SL: Yes.Do you know?CM: It only just happens. TABLE 1 : Measures of texts and average reading scores on each. Average reading rate wpm/letters per minute (lpm)/letters per second (lps) Average number of fixations Average number of regressions 1. Sengikhulile 100 words 765 letters 16 lines 19 sentences Average number of words per sentence: 5.3 Average number of clauses per
| 1,212 |
55002387
| 0 | 16 |
sentence: 1.4 159 wpm/ 1214 lpm /20 http://www.rw.org.zaOpenAccessTM: WHAT? amaNgisi namBhunu?Yes… I had to re-read the sentence to make sure if I was reading the right thing.
| 1,213 |
55002387
| 0 | 16 |
Coronary microcirculatory dysfunction in hypercholesterolemic patients with COVID-19: potential benefit from cholesterol-lowering treatment Abstract Patients with hypercholesterolemia often have coronary microvascular dysfunction (CMD). Viral infections, such as the SARS-CoV-2 infection, may also result in CMD. Three non-randomized studies have shown significant beneficial effects of statins on CMD in non-infected patients. Similarly, in SARS-CoV-2 - infected patients one beneficial mechanism of action of statins may be the amelioration of endothelial dysfunction, which is a major driver of CMD. Apart from statins, lipoprotein apheresis and PCSK9 inhibitors can also improve or even reverse CMD. The potential reversal of CMD by using effective cholesterol-lowering medications during and after COVID-19 infection, especially in hypercholesterolemic COVID-19 patients, is important. KEY MESSAGES Coronary microvascular dysfunction (CMD) is common in patients hospitalized with SARS-CoV-2 infection Three nonrandomized studies in non-infected patients are showing the beneficial effects of statin treatment on CMD Effective cholesterol-lowering medication during and after SARS-CoV-2 infection, especially in hypercholesterolemic COVID-19 patients, is of great significance Introduction Coronary microcirculation refers to the coronary vessels with a diameter of less than 500 µm which, in individuals without epicardial atherosclerosis, vasodilate during exercise and thereby augment myocardial perfusion [1]. Coronary microvascular dysfunction (CMD) may lead to impaired coronary flow reserve and a high index of coronary microcirculatory resistance [2][3][4]. Typical diseases that lead to myocardial ischemia without epicardial coronary atherosclerosis are hypertrophic cardiomyopathy, aortic stenosis, and myocarditis [5][6][7]. It has been shown that endothelial dysfunction is a key component of CMD pathogenesis [8][9][10]. By causing endothelial dysfunction, severe hypercholesterolemia can contribute to CMD,
| 1,214 |
258180659
| 0 | 16 |
and, in a swine model of familial hypercholesterolemia, it was demonstrated that CMD is a contributor to cardiac ischemia even before the development of critical coronary atherosclerotic cardiovascular disease (ASCVD) [9]. Importantly, several human studies have shown that severe hypercholesterolemia, such as in subjects with familial hypercholesterolemia, induces systemic endothelial dysfunction, which often includes CMD [11][12][13][14]. Discussion In addition to hypercholesterolemia, viral infections may also result in CMD. More than a decade ago CMD was reported in patients with acute respiratory distress syndrome and influenza A (H1N1) [15]. This year an echocardiographic coronary flow velocity reserve analysis revealed that CMD is common in patients hospitalized with SARS-CoV-2 infection [16]. In a very comprehensive study of patients who had recently recovered from COVID-19, patients with classic This is an open Access article distributed under the terms of the creative commons Attribution license (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The terms on which this article has been published allow the posting of the Accepted Manuscript in a repository by the author(s) or with their consent. hypertrophic cardiomyopathy and healthy controls were compared using vasodilator stress cardiovascular magnetic resonance examination for the evaluation of myocardial perfusion reserve (MPR) [17]. In this study significantly reduced MPR was found in the COVID-19 patients compared to healthy controls, and the reduced MPR observed was as severe as that found in the patients with hypertrophic cardiomyopathy. Drakos et al. [17] postulated that the reduced MPR was caused by both direct and
| 1,215 |
258180659
| 0 | 16 |
triggered mechanisms caused by the SARS-CoV-2 virus infection, which thereby resulted in CMD. This conclusion is supported by a recent review suggesting a variety of injury mechanisms causing endothelial dysfunction in the microcirculatory environments in COVID-19 patients [18]. As statins improve endothelial dysfunction and can reduce vascular inflammation, they were included in the list of secondary prevention strategies depicted in the recent European Society of Cardiology Working Group on Coronary Pathophysiology and Microcirculation position paper on "coronary microvascular dysfunction in cardiovascular disease" [1]. It is noteworthy that, in observational studies statins have been shown to reduce both the morbidity and mortality of patients with SARS-CoV-2 infection and that several similar randomized studies are currently underway [19][20][21]. One beneficial mechanism of the action of statins may be the reduction of CMD exacerbation caused by the SARS-CoV-2 infection. This opinion is supported by the findings obtained in several nonrandomized studies performed on non-infected patients and which are showing significant beneficial effects of treatment with statins on CMD [22]. The benefit of rosuvastatin was shown in patients with hypertension [23,24], the benefit of atorvastatin in patients with angina pectoris [25], and the benefit of simvastatin in patients with documented ASCVD [26]. Apart from statins, lipoprotein apheresis and PCSK9 inhibitors as cholesterol-lowering strategies can also reverse CMD [27,28]. The benefit of the PCSK9 inhibitor evolocumab was shown in a recent randomized study of 60 hospitalized patients with severe SARS-CoV-2 infection [29]. Within 30 days of observation, the patients treated with a 140 mg subcutaneous injection of evolocumab had lower mortality
| 1,216 |
258180659
| 0 | 16 |
and less need for intubation compared to the placebo group (23.3 vs. 53.3%; 95% CI −53.4 to − 6.59%). It has been shown earlier that in Dengue virus infection, a PCSK9 inhibitor could interfere with the production of IL-6, thereby reducing the inflammatory response [30]. In addition, the favorable mechanisms of PCSK9 inhibitors, especially in HeFH, could be argued to also result from an effective reduction in the levels of LDL-C and to a lesser extent in the reduction of Lp(a) levels, with a resultant improvement in endothelial function [31,32]. Regarding SARS-CoV-2-infected patients, the microvascular angiopathic consequences on the myocardium may lead to myocardial ischemia and dysfunction with grave consequences [33]. Therefore, it is of paramount importance to aim at preventing the development of CMD in COVID-19 patients. Conclusion Both severe hypercholesterolemia and SARS-CoV-2 infection cause endothelial dysfunction and the associated CMD. Therefore, it is appropriate to attempt to mitigate or reverse CMD in hypercholesterolemic COVID-19 patients by using effective cholesterol-lowering medications, particularly statins. This reminder is particularly important since many hypercholesterolemic patients, including those with familial hypercholesterolaemia in whom the hypercholesterolemia can be severe, remain undertreated or even untreated before, during, and after the infection. Data availability statement All data relevant to the study are included in the article. Author contributions AV: conceptualizing, drafting, writing and designing the first version. AV, PTK, FR: revising critically, editing to produce the final draft. All authors contributed and are accountable all aspects of the article and approved the version to be published. Disclosure statement No potential conflict of
| 1,217 |
258180659
| 0 | 16 |
interest was reported by the author(s). Funding AV has received consultancy fees from Amgen and Novartis. Associate Editor, Cardiology & Cardiovascular Disorders, Annals of Medicine. PTK has received consultancy fees, lecture honoraria, and/or travel fees from Amgen, Novartis, Raisio Group, and Sanofi. Section Editor, Cardiology & Cardiovascular Disorders, Annals of Medicine. FR has received research grants, honoraria, or consulting fees for professional input and/or lectures from Sanofi, Regeneron, Amgen, Novartis, and LIB Therapeutics.
| 1,218 |
258180659
| 0 | 16 |
Estimating trends in working life expectancy based on health insurance data from Germany – Challenges and advantages Against the backdrop of population aging and growing strain on pension systems, monitoring the development of Working Life Expectancy (WLE) is vital to assess whether the policies taken are effective. This is the first study investigating time trends and educational inequalities in WLE based on German health insurance data. The analyses are based on the data of the AOK Lower Saxony (N = 3,347,912) covering three time periods (2006-08, 2011-13, and 2016-18). WLE is defined as years spent in the labor force (i.e. in employment and unemployment) and was calculated for each age between 18 and 69 years for the three periods to depict changes over time using multistate life table analysis. Educational inequalities in 2011-13 are reported for two educational levels (8–11 years and 12–13 years of schooling). WLE increased in both sexes with increases being stronger among women. This holds irrespective of whether WLE at age 18 (35.8–38.3 years in men, 27.5–34.0 years in women) or the older working-age (e.g. at age 50 10.2–11.7 years in men, 7.8–10.5 years in men) is considered. Among women at all ages and men from their mid-20s onwards, WLE was higher among higher-educated individuals. Inequalities were most pronounced among women (e.g. Δ3.1 years in women, Δ1.3 years in men at age 50). The study supports previous research indicating that measures to extend working life are effective, but that noticeable inequalities in WLE exist. Health insurance data represent a valuable source for
| 1,219 |
251917982
| 0 | 16 |
such research that has so far remained untapped. The data provide a suitable basis to investigate trends and inequalities in WLE. Future research should build on the strengths of the data by broadening the research towards a more comprehensive analysis of the development of WLE from a health perspective. Introduction Increasing life expectancy and population ageing are putting growing strain on public pension systems in high-income countries. Therefore, in recent decades measures have been taken in many countries to ease this burden by prolonging working lives and increasing the ratio of the working to the non-working population (Anderson et al., 2019;Hytti & Valaste, 2009). Among them, Germany is one of the demographically oldest countries with a median age of 44.1 years for men and 47.6 years for women in 2019 (Bundesinstitut für Bevölkerungsforschung, 2022). Policies responded by gradually raising the regular retirement age from 65 to 67, which will be reached in 2031. Early retirement leads to lower pension payments, but is frequently used in Germany. Therefore, the actual retirement age due to old age was 64 years on average in 2021 (Deutsche Rentenversicherung, 2022). Against this background, monitoring the development of the length of working life over time is becoming increasingly important, as this can provide information on whether the political measures to extend working life are effective (Loichinger & Weber, 2016). Furthermore, the question arises as to how large the existing social inequalities in working life expectancy are and how they have developed over time. Studies investigating trends in the length of working life
| 1,220 |
251917982
| 0 | 16 |
usually draw conclusions from analyzing the development of working life expectancy (WLE). This indicator depicts the number of (remaining) years expected to be spend in economically active states (Hoem, 1977). Previous research differs with respect to the definition of WLE (years in paid employment vs. years in the labour force) and to the methods applied (i. e. multistate life tables vs. Sullivan method (Sullivan, 1971)). Therefore, direct comparisons between studies are often hampered especially if exact values of WLE should be compared. However, irrespective of these limitations previous European studies examining the development of WLE of the older working-age population consistently reported distinct increases in recent decades for both sexes, with increases being most pronounced in women (Leinonen et al., 2018;Loichinger & Weber, 2016;Nurminen, 2012;van der Noordt et al., 2019;Weber & Loichinger, 2020). This holds irrespective of whether WLE is defined based on active labor (e.g. Robroek et al., 2020)) or on labor force participation (i.e. years in employment and unemployment (e.g. (Eurostat, 2022a;Loichinger & Weber, 2016)), indicating that time trends are similar and general trends can be compared. This development was mainly driven by increased labor force participation at age 50 and older (Loichinger & Weber, 2016). Previous research has shown that WLE varies considerably by education (Loichinger & Weber, 2016;Robroek et al., 2020;van der Noordt et al., 2019;Weber & Loichinger, 2020) and occupational group (Kadefors et al., 2019;Leinonen et al., 2018;Schram et al., 2021), with individuals with lower socioeconomic status having lower WLE. Since current political measures primarily aim at increasing the labor force
| 1,221 |
251917982
| 0 | 16 |
participation in the elderly, most studies on WLE focus on the older population (mostly age 50+). However, against the background of prolonged training periods and increasing female employment across the entire range of working-age, it is also important to investigate changes in WLE at the beginning of the working biography as well. Previous research investigating time trends or social inequalities in WLE has been primarily based on survey data (e.g. Dudel et al., 2021;Loichinger & Weber, 2016;Nurminen, 2012;Weber & Loichinger, 2020)). For Germany, studies on trends in WLE reported substantial increases at age 15 (Eurostat, 2022) and 18 (Heller et al., 2022) as well as at age 50 and above (Heller et al., 2022;Loichinger & Weber, 2016;Weber & Loichinger, 2020) during the last two decades. So far, however, evidence on social inequalities in WLE in Germany is still very limited. This is due to the lack of official life tables by socio-economic characteristics in Germany, which are often used to calculate WLE based on survey data. To the best of our knowledge there is so far only one study which investigated time trends in paid full-time equivalent WLE at age 55 to 64 in Germany, ignoring the effect of mortality on WLE (Dudel et al., 2021). This study found that educational inequalities in WLE are substantial and that disparities have increased, at least in East Germany (Dudel et al., 2021). This raises the question of whether other data sources can also can be used to study trends and inequalities in WLE. For Germany, the use of
| 1,222 |
251917982
| 0 | 16 |
statutory health insurance data is conceivable since information on employment status and mortality are very well recorded and socioeconomic information such as education, occupation and income is also available. The aim of this paper is to describe the possibilities of using health insurance data to estimate the WLE and to analyze time trends and social inequalities in WLE based on this kind of data. In this study, WLE is defined as years spent in the labor force rather than in paid employment only in order to depict trends in years potentially spent in active labor. Given the ageing population, labor shortages are expected to worsen in the future. WLE defined as the years in the labor force provide insight into the potential years in active employment available as the labor market adapts to older workers. Based on this, our study contributes to the current research by addressing the following questions: • How did WLE develop over time? • Did the time trends in WLE differ by gender? • How large is the gap in working life expectancy between educational groups? In particular, it will be discussed how the general level of WLE as estimated from German health insurance data can be compared to the WLE calculated from German survey data. Furthermore, it will be considered how the time trends in WLE vary over time between the different data sources. Due to the limited number of studies on social inequalities in WLE in Germany, an additional focus is on the calculation of WLE by education. Here, the
| 1,223 |
251917982
| 0 | 16 |
advantages and disadvantages of using health insurance data for the calculation of social inequalities in WLE are addressed in detail. Data The analyses are based on the data of the statutory health insurance provider (AOK Lower Saxony, AOKN) of three time periods (2006-2008, 2011-2013, and 2016-2018). Statutory health insurance is an essential element of the social security system in Germany. Due to legal regulations restricting the access to private health insurance, about 90% of the German population are insured with a statutory health insurance provider (electronic report). The dataset contains records of approximately two million insured individuals aged 18 years and above per year (in total N = 3,347,912 in all three periods) and cover about a third of the population of Lower Saxony (AOK Niedersachsen, 2019), Germany. The data were collected for accounting purposes and contain longitudinal information on insurance histories, diagnoses, medical procedures and mortality, as well as on employment and unemployment periods and other socioeconomic characteristics, e.g. income and educational level. With respect to sex and age, distributions within the AOKN population is comparable to the total population of Lower Saxony and Germany (Jaunzeme et al., 2013) but differ in terms of occupational groups and educational level with higher levels being underrepresented (Epping et al., 2021). Table 1 displays some basic characteristics of the study population. Labor force definition In previous research, the labor force concept of the International Labor Force Organization (ILO) (Benes, 2018;European Commission, 2016) has been applied to calculate WLE (e.g. (Loichinger & Weber, 2016;Weber & Loichinger, 2020)). According to
| 1,224 |
251917982
| 0 | 16 |
this definition, all employed individuals are counted as part of the labor force as well as individuals who are employed but temporarily not working due to illness, parental leave, further training, etc. In addition, all unemployed individuals are assigned to the labor force if they are actively seeking employment and are quickly available to start a new job (Benes, 2018;European Commission, 2016). Accordingly, the labor force represents not only the employed, but also the population which would potentially be available to the labor market. In line with the ILO-definition, episodes of paid employment (including self-employment) and unemployment (i. e. receiving unemployment benefits) were defined as episodes contributing to the time an individual spent in the labor force. In contrast, episodes of retirement (e.g. due to old age or disability) or of family insurance (i.e. episodes without income during which spouses and children below age 23 can be co-insured free of additional fees with a working married partner or a parent) were assigned to the periods spend in the non-labor force. Further details on the assignment of the different episodes can be found in the online supplement (Online Resource 1, additional remarks 2). Whenever there were overlapping episodes of employment and unemployment, the person was defined to be economically active and thus the episode contributed to the lifespan spent in the labor force. Educational information Information on educational attainment is available for insured individuals who were ever employed within the observation period (in this case 2005 to 2018). For these analyses we used the highest schoolleaving qualification,
| 1,225 |
251917982
| 0 | 16 |
since school education is usually completed at age 18 to 19 and is therefore suitable for the analysis of WLE at age 18 and above. If education was missing for the current episode (e.g. due to unemployment or missing data), the educational information from the previous or the following episodes was assigned. Three educational groups can be distinguished: (1) graduation after 8-11 years of schooling (Volksschule, Hauptschule, Realschule or equivalent, low), (2) after 12-13 years of schooling (Abitur, Fachabitur or equivalent, high), and (3) no school-leaving qualification or unknown qualification (Table 1). Inequalities were analyzed by comparing WLE of the low-and the high-educated group. Since education is only coded for insured individuals who have ever been employed during the observation period, it is especially often lacking for individuals with long-term family insurance or unemployment episodes. Therefore, the total percentage of missing information on education is considerably higher in the non-labor force (57%) than in the employed individuals (27%). In order to reduce the share of missing values among the non-employed population, we transferred the educational information of the working spouse or parent to the non-working spouse or child aged 18 years and above with missing information. These strategy was based on the assumption of educational homogamy within married parters (Blackwell & Lichter, 2004;Muschik et al., 2015;Safieddine et al., 2020) and that the educational qualification and that the educational qualifications of parents and children are often comparable (Björklund & Salvanes, 2011;Black et al., 2005;Dickson et al., 2016;Sirin, 2005). Following these strategies, the proportion of missing values among
| 1,226 |
251917982
| 0 | 16 |
the non-labor force decreased by 16 percentage points. Individuals who still had missing information on education were excluded from the analyses of educational inequalities in WLE. However, due to the short time remaining in gainful employment at older working age, the educational information is often missing at high working age in the first period while it was often derived from earlier years for the elderly in last period. In a similar way, this applies to the very young working age in the last period. Due to this uneven distribution of missing values across age between periods, the analyses on educational inequalities in WLE are limited to the middle period 2011-2013 in which the proportion of missing values on education is lowest (Online Resource 1, Table A5). Statistical analyses First, the proportions of labor force participation by period and single-year age group were calculated to depict the general trend in the labor force participation over time and which age groups contributed strongest to this development. To illustrate educational differences across age the labor force proportions by education were calculated as well. Working life expectancy (WLE) was calculated based on multistate life table analysis. These life tables include two transit states (labor force, non-labor force) allowing for repeated transitions between them and one competing absorbing state (death). Between the three states, four transitions are possible: 1) non-labor force to labor force, 2) labor force to non-labor force, 3) non-labor force to death, and 4) labor force to death (Table 1). In order to describe the temporal development of
| 1,227 |
251917982
| 0 | 16 |
the transition risks between these three states, cox-proportional hazard models were estimated for each of the three periods. These models contain period as a categorical covariate, are stratified by sex and controlled for age (in single-year age groups). Moreover, age-specific transition rates were calculated. These rates were used as input for the multistate life table analysis which is based on matrix multiplication as described by Palloni (2001, pp. 256-272). WLE was calculated for the three periods 2006-2008, 2011-2013, and 2016-2018. The calculation of WLE is based on partial life expectancy for ages 18-69, since the proportion of individuals being in labor above that age is very low (<1%) in both sexes. The analyses were stratified by sex. WLE is reported at age 18, 50 and 60 to capture differences in trends over time across age. Finally, the differences in WLE are analyzed according to the highest school-leaving qualification for the period 2011-2013. Data management and the regression models were performed using Stata MP 14.2 (Stata Corp, 2015), the WLE were calculated using R 3.5.1 (RCore Team, 2015). Confidence intervals were obtained from 1000 bootstrap samples (with replacement). Time trends and educational differences in the labor force participation Between 2006-2008 and 2016-2018e, the total labor force proportion of the insurance population rose in men (72%-79%) and women (50%-66%). Among men, this increase was strongest among the older working-age population above the age of 50. While for women the overall level of the labor force proportion was lower, the extent of increases were much higher than in men
| 1,228 |
251917982
| 0 | 16 |
across the full age range. Similar to men, the labor force participation increased strongest at higher working ages (Fig. 1). The labor force proportion is higher among women with higher education in all ages (in total 69% vs. 62%). Among younger men, by contrast, the proportion is higher in individuals with lower education (40% vs. 17% at age 18), reflecting their earlier labor market entries (Fig. 2). Time trends in transition risks For both sexes, the rates of entering the labor force increased over time. This increase was stronger in women (HR 1.57, Ref. = period 1) than in men (HR 1.38, Ref. = period 1). In contrast, the risk of the Note: Percentages do not always add up to 100% due to rounding; Data source: AOK Lower Saxony health insurance data. a Values refer to the middle period 2011-2013 only (for more information, see section "educational information"). reverse transition from the labor force to the non-labor force increased slightly in men (HR 1.05), while it decreased slightly in women (HR 0.94). Among the non-labor force, the competing risk of death remained quite constant in men but increased in women while death risks decreased for economically active men and women (Table 2). Time trends and educational differences in working life expectancy In the insurance population, the number of economically active life years at age 18 increased from 35.8 years to 38.3 years for men and from 27.5 to 34.0 years for women. WLE also increased substantially at ages 50 (10.3-11.7 years in men, 7.8-10.5 years in
| 1,229 |
251917982
| 0 | 16 |
women) and 60 (2.8-3.7 years in men, 2.0-3.3 years in women). The stronger increase in WLE among women is also evident at higher working age (age 50: +2.7 years compared to +1.5 years in men) (Fig. 3). All increases in WLE over time are statistically significant (Online Resource 1, Table A1). Analyzing trends across the full range of working age, it becomes apparent that WLE increased substantially at every age up to the highest working ages (Online Resource 1, Fig. A2). Furthermore, WLE differs by educational group. For men, these inequalities vary with respect to age. While WLE at age 18 is higher among low-educated than among higher-educated men (Δ 2.5 years), it is lower at older working age (Δ 1.3 years at age 50 and Δ1.1 years at age 60). However, when considering the entire age range from 18 to 69 years, it becomes apparent that men with higher education consistently have a higher WLE from their mid-20s onwards than men with lower educational attainment of the same age (Fig. A3, Online Resource 1). This is because higher educated men stay longer in the labor market than men with low education (Fig. 2). Stronger inequalities were found among women, pointing in the same direction over the entire age range, with WLE being higher among higher-educated women than among women with low education (Δ +4.1 years at age 18, Δ +3.1 years at age 50, Δ +1.6 years at age 60) (Fig. 4, Fig. A3 Online Resource 1). Discussion This study is one of the few
| 1,230 |
251917982
| 0 | 16 |
studies analyzing time trends and educational differences in WLE in Germany. While previous research has been based on survey data, our study is the first one to use health insurance data. Due to the availability of detailed data on diagnoses and other medical information, health insurance data are a valuable data source to enhance research on the development of the length of working lives from a health perspective. Main findings The study shows clear increases in WLE over time for both sexes. These increases were observed in both younger and older working ages and are driven by increasing rates of labor market entries and growing labor force participation, especially in the elderly. Overall, the findings indicate stronger increases in labor force participation and WLE across the entire age range among women than among men. This finding is in accordance with the higher transition rates to labor force and the lower transition rates to non-labor force among women. Furthermore, the length of working lives differs by educational level. Younger men with low educational level had higher WLE than men with high educational attainment. This is due to earlier labor market entries among men with low education compared to younger men with higher education. From their mid-twenties onwards, however, WLE is higher among men with higher education. For women with higher education, WLE is higher regardless of the age group considered. While clear inequalities in WLE emerged in both sexes they were more pronounced among women reflecting that women with higher education enter the labor market more frequently
| 1,231 |
251917982
| 0 | 16 |
and stay economically active up to higher ages than lower-educated women. In addition, the higher level of labor force participation indicates that men tend to be economically active more often irrespective of education. Discussion in the context of previous research Previous studies analyzing time trends in WLE in Germany focused mostly on the higher working-age (Loichinger & Weber, 2016;Weber & Loichinger, 2020) and reported mainly partial WLE for specific age ranges (Dudel et al., 2021;Weber & Loichinger, 2020). In addition, there are differences either in the definition of the working population (Dudel et al., 2021) or in the time period studied. Therefore, findings cannot be compared directly across the entire period (Loichinger & Weber, 2016). Nevertheless, comparisons with previous research can provide some insight into whether general trends and educational inequalities in the insurance population differ from the findings reported for the total German population obtained from survey data. Comparing our results with the WLE at age 50 based on the ILO labor force concept, the overall level of WLE tends to be lower than the survey-based results (Heller et al., 2022;Loichinger & Weber, 2016). Similarly, the level of WLE at young working-age is lower in our study than those reported by Eurostat (Eurostat, 2022a) and another study based on the German Socio-economic Panel (Heller et al., 2022). This is in accordance with our expectation and may be explained by the differences in the statistical methods used (Sullivan versus multistate life approaches) and because individuals with low SES (and therefore with low WLE) are overrepresented compared
| 1,232 |
251917982
| 0 | 16 |
to the general German population. However, although the general level of WLE reported in this paper is lower, comparing the results with previous research shows that the data reflect time trends in WLE in the general German population well. This holds for time trends at younger (Eurostat, 2022a;Heller et al., 2022) and older working age (Dudel et al., 2021;Heller et al., 2022;Loichinger & Weber, 2016;Weber & Loichinger, 2020). In line with previous research, our study shows that WLE increased much stronger among women than among men (Eurostat, 2022a;Heller et al., 2022;Loichinger & Weber, 2016;Weber & Loichinger, 2020). The stronger increase in women led to a narrowing gender gap in WLE over time. However, as in earlier studies, the gap in WLE between men and women persisted and men still work considerably more years than women (Dudel et al., 2021;Eurostat, 2022a;Heller et al., 2022;Loichinger & Weber, 2016;Weber & Loichinger, 2020). To the best of our knowledge, there is only one previous study investigating social differences in WLE in Germany (Dudel et al., 2021). This study also found substantial disparities in WLE between educational groups. This is consistent with international research, which also reported large inequalities, especially at higher working ages (e.g. (Dudel & Myrskylä, 2017;Leinonen et al., 2018;Loichinger & Weber, 2016;Robroek et al., 2020;Schram et al., 2021;Weber & Loichinger, 2020)). While this previous study reported high educational inequalities at age 55-64 in both sexes (Dudel et al., 2021), our findings indicate much higher inequalities in WLE in women than in men. This difference between men and women
| 1,233 |
251917982
| 0 | 16 |
in inequality levels between the studies may be explained by methodological reasons, especially due to the differing definitions of working life (e.g. by using employment periods only instead of combining periods of employment and unemployment, ignoring the effect of mortality on WLE, or applying an adjustment of total WLE for working hours to measure full-time equivalent WLE). Since the continuity of employment biographies differs greatly between the SES groups, higher inequalities are to be expected if only the periods of active employment are included. Comparing our results with a recent study indicates that inequalities are significantly greater in terms of years in active employment than in terms of time spent in the labor force, especially for men (Dudel et al., 2021). While studies on WLE based on employment periods alone add to the current knowledge on trend and inequalities in life years spent in active labor (Dudel et al., 2021), the results are influenced by the current economic conditions and the "potential" WLE cannot be captured. Therefore, previous studies also differ in terms of content-related aspects. Our study complements to the current research by reporting educational differences in the length of life spent in the labor force. Since most research on social inequalities is focused on WLE among the elderly, the evidence is more limited when it comes to social inequalities in WLE at very young working ages (e.g. (Robroek et al., 2020)). The focus on paid employment in previous research may explain why higher WLE at younger working ages was found in both sexes in
| 1,234 |
251917982
| 0 | 16 |
the group of higher educated, while our results suggest that WLE at age 18 is higher among men with low educational attainment. This may be due that lower-educated men are more likely to enter the labor market through unemployment than men with higher education, which also contribute to WLE following the labor force concept. For women, our results are in line with previous research indicating longer working lives Table A4. Individuals with missing information on education were excluded Data source: AOK Lower Saxony health insurance data. for women with higher education irrespective of the age group considered (Robroek et al., 2020). Labor force concept Previous studies define WLE either as the number of years expected to be in the labor force or as years spent in paid work (Dudel et al., 2021;Leinonen et al., 2018;Loichinger & Weber, 2016;Parker et al., 2020aParker et al., , 2020bRobroek et al., 2020;Schram et al., 2021;Weber & Loichinger, 2020). While both approaches clearly contribute to the current state of research, the use of the labor force definition allows to describe potential years in employment. This study defines WLE as the number of years spent in the labor force, i.e. periods of employment and unemployment equally contribute to WLE. This definition is particularly suitable to address the question of whether working life has increased independently of the changing economic framework. . Other studies, however, are more interested in the years spent in paid employment, as increases in active employment directly contribute to economic output and ease the burden on pension funds. Accordingly,
| 1,235 |
251917982
| 0 | 16 |
both definitions contribute to the current state of research and it depends on the research question which definition is to be preferred. Another advantage of analyzing trends in WLE based on labor force participation is that this approach is less sensitive to short-term period effects which may lead to sudden but temporary drops in WLE, e.g. due to temporary sharp increases in unemployment rates due economic crises. Similar to many previous studies, our analyses are based on period life tables which are prone to such period effects. By explicitly including unemployment episodes, the impact of such effects can be reduced. Data characteristics relevant to calculate WLE The calculation of WLE is based on detailed data on labor force and educational attainment, which allowed us to analyze WLE based on different transition rates relevant to the length of working lives. Furthermore, information on mortality are provided in the same dataset. Therefore WLE could be analyzed for the general insurance population without excluding those who died within the study period and is therefore not conditional on surviving to a respective age. This is an advantage especially when WLE is reported for older ages, when deaths occur more frequently. Without taking mortality into account, WLE at general population level would be overestimated. In this study, labor force is measured following the ILO-definition (Benes, 2018;European Commission, 2016) as far as possible. However, this study differs in two aspects from this definition. First, while the ILO-definition is based on the employment status of a specific reference week received from surveys, labor
| 1,236 |
251917982
| 0 | 16 |
force participation in our study is measured based on secondary data. This means that the information on different (non-)employment episodes is recorded across the full time period in which the individual is insured with the health insurance provider and labor force histories within these periods can be followed straightforward. The continuity of the labor force information in the data is a strength of the study because it allows for a time-related assignment of employment status, which enabled us to calculate WLE based on multistate life tables. Multistate analyses are considered preferable to the Sullivan method (Sullivan, 1971) frequently used for cross-sectional data, because life expectancies can be estimated more precisely, especially when transition rates fluctuate over time (Jagger et al., 2014;Mathers, 1991). Second, the health insurance data do not provide information on whether the unemployed insured individuals actually intend to achieve gainful employment. In contrast to the ILO-definition, we could therefore not distinguish between persons who are actively seeking for employment and who are ready to take up an offered job within the next two weeks and those who are not. Therefore, WLE might be somewhat overestimated compared to the results that would have been obtained if the calculation of the WLE was based on the original ILO labor force definition. However, since the labor force definition is applied consistently across the full study period, we assume that the analyses on time trends in WLE are not affected. Educational inequalities in WLE The data contain information on the highest school-leaving qualification, which allowed us to calculate
| 1,237 |
251917982
| 0 | 16 |
WLE by two educational groups. While using three or more educational groups would allow for a deeper insight in inequalities in WLE, a more detailed classification could not be used due to data restrictions. The information on education is usually available for those who ever had an occupation within the study period, but it is lacking for those who were not (e.g. long-term unemployed and long-term non-labor force). Therefore, the educational information on working spouses or parents have been transferred to non-working family insured individuals with missing information on educational attainment. Beyond theoretical consideration, previous studies have shown that other strongly education-dependent characteristics, such as health status and health inequalities, can be well depicted based on transferred educational information (Muschik et al., 2015;Safieddine et al., 2020). Although the assignment may not be correct in every case, the above mentioned theoretical considerations and previous research (Muschik et al., 2015;Safieddine et al., 2020) suggest that this methodological approach can be used to estimate educational inequalities in the WLE. The robustness of our results is supported by the fact that the educational inequalities found in our study are overall consistent with the findings from previous research (Dudel et al., 2021). Although the analyses on educational inequalities are limited to the period with the lowest missing proportion (Table 1; Online Resource 1, Table A5), there is some selectivity, as persons with missing educational information had to be excluded from the analyses. The exclusion of these persons led to a higher WLE in the two education groups compared to the whole
| 1,238 |
251917982
| 0 | 16 |
insurance population (Fig. 4 compared to Fig. 3). This is probably due to the fact that persons with very low education, who often have interrupted employment histories, were disproportionately often excluded from the analyses due to missing information on education. Additional analyses are supporting this assumption, since WLE of individuals without educational information is much lower than in the total insurance population (Online Resource 1, Fig. A6). Furthermore, the data do not allow us to distinguish between persons without a school-leaving certificate and those with missing information, which means that individuals without a school-leaving certificate had also to be excluded from the analyses. This suggests that WLE among the loweducated group tends to be over-and educational inequalities in the WLE tend to be underestimated. However, by using strategies for assigning information on education, e.g. those of the parents or spouses, the effect of excluded individuals with missing information on education could be reduced (Online Resource 1, Fig. A7). Since the proportion of missing values is higher among women and low-educated individuals, the effect of assigning educational information on WLE is strongest for women and low-educated individuals (Online Resource 1, Fig. A8). For this study, the health insurance data covering 13 years were used, which reduced the share of individuals with missing information on education since it is usually available for individuals who were employed at some point during the observation period. Nonetheless, it must be noted that WLE is likely to be somewhat over-and educational inequalities in WLE to be somewhat under-estimated using the full dataset
| 1,239 |
251917982
| 0 | 16 |
of 13 years of observation as individuals without school-leaving qualification and low levels of education are likely to be excluded more often due to missing information on education. This issue is the larger, the shorter the period for which the data are available. Thus, the advantages of health insurance data must be weighed against these disadvantages especially if the data cover short observation periods. Future perspectives and practical implications In recent decades, great efforts have been made to reduce the impact of population ageing on social security systems and economy. These efforts were most apparent at older ages by current increases in the statutory retirement age. In Germany, the discussion about a further increase in the retirement age is mainly driven by two developments. First, as in other countries, rising life expectancy is leading to an increasing share of life spent outside the labor force. Second, the strong baby boomer generation in Germany is expected to reduce the number of employed individuals in the coming years and exacerbate the shortage of skilled workers (Geyer & Eberhard, 2020). Increasing WLE is considered an essential measure to mitigate the impact of these developments on the economy and society. Given the population ageing and increasing labor shortage, WLE defined as years spent in the labor force provides insight into the potential years in active employment. Not all years in the labor force are spent in active employment. However, similar to the development of life years in the labor force, increases were found in years spent in paid work (Dudel
| 1,240 |
251917982
| 0 | 16 |
et al., 2021), indicating that a meaningful proportion of the increase in years in the labor force is already spent in active employment. However, in order to counteract the consequences of further population ageing, it is vital to promote further adaptations of the labor market to the needs of older workers. In addition, the question arises as to whether the health of the ageing populations allows for a (further) extension of working life. This is especially true for countries where labor force participation in old age is already high, such as in Germany. So far, however, research about how healthy life years and healthy working life years develop and influence each other in ageing populations, and whether time trends differ between socio-economic groups, is still insufficient. From a social policy perspective, these questions are highly relevant, which underlines the importance of exploiting the potentials of existing data sources. Health insurance data represent a still little-used but valuable data source that can expand the previous research which is so far usually based on survey data. This applies especially to the potentials related to the combination of labor force participation with people's health status to study Health Expectancies and Healthy Working Life Expectancy (HWLE), e.g. with regard to specific diseases that are often associated with severe limitations or early retirement. This would be essential knowledge for assessing the possibilities and limits of a further extension of working lives (Boissonneault & Rios, 2021) what could not be investigated with other data sources. Conclusion The study shows that WLE has
| 1,241 |
251917982
| 0 | 16 |
increased for both genders, suggesting that the policy measures taken to extend working lives are effective. Although the gender gap in WLE has narrowed over time, men still spend considerably more years in the labor force than women. The potential for increasing WLE is particularly high among women with low education. Measures specifically aimed at increasing the employment rate of women with lower education may therefore be particularly promising in further increasing WLE at population level. Moreover, the study suggests that German health insurance data are an appropriate source to calculate educational inequalities in WLE as long as the data are available for a longer time period and the limitations are taken into account. Further research should build on the strengths of the data and investigate whether the observed increase in WLE is also associated with an increase in years in labor free of diseases that are known to be associated with reduced working ability and early exits from the labor market. Ethical statement Our study is based on claims data, i.e., on routinely collected data of a statutory health insurance provider. We confirm that all data are fully anonymized before we accessed them. The use of this sort of data for scientific purposes is regulated by federal law. The data protection officer of the Statutory Local Health Insurance of Lower Saxony (AOK Niedersachsen) has approved its use. Declaration of competing interest None. Data availability The authors do not have permission to share data.
| 1,242 |
251917982
| 0 | 16 |
Outcome of the AVID College Preparatory Program on Adolescent Health: A Randomized Trial This randomized trial tests whether an academic untracking program (Advancement via Individual Determination) leads to healthier social networks and reduced substance use among Southern California 9th grade students. Through near-daily exposure, schools have potential to shape adolescents' relationships, social norms, and social-emotional skills-1 factors strongly associated with health behaviors. 2,3 Educational interventions that alter these factors may have important spillover effects on health. However, this has been rarely studied. Academic tracking is a widely used strategy that groups students in classrooms according to prior academic performance. Critics argue tracking perpetuates structural disadvantage and racism by limiting access to educational opportunities for students of color and from lowincome families. 4 Although the educational merits of tracking are vigorously debated, 5-9 no known studies examine its health implications. Academic tracking might directly influence social networks by grouping students together with peers of similar academic performance and engagement. [10][11][12] Although potentially beneficial to highperforming students, tracking may reinforce school disengagement and risky health behaviors like substance use, violence, and delinquency among lowerperforming students. 13,14 Studies suggest teens tend to form friendships with peers based on similar levels of school engagement and risk behaviors. 15 Within schools, this process can be reinforced by placing similar students in the same classrooms. 16 This theory is strongly supported by evidence that adolescent health behaviors, including substance use, violence, and delinquency, are closely tied to behaviors and attitudes of individuals in their social network. 14 ,17-21 Academic tracking may determine
| 1,243 |
254819264
| 0 | 16 |
to which peers a student is exposed, further impacting sources of support, transmission of social norms, and access to and opportunities for risky health behaviors, and psychosocial wellbeing. 19,20,22,23 Advancement via Individual Determination (AVID) is a college preparatory program operating in nearly 5000 US high schools across 46 states (about 20% of all public high schools). 24 AVID targets students in the academic middle (earning B or C grades, on average) who would not typically be placed in high-performing, college preparatory academic tracks. 25 Through AVID, participating students are encouraged to enroll in college preparatory courses and as a result, AVID has been described as an academic untracking intervention. 6 Although prior studies examined AVID's educational outcomes, 26-31 there are no randomized trials of AVID in the United States and no studies examining its impact on health. To fill this gap, we conducted the first randomized trial of AVID in the United States to test whether AVID improved adolescent health. We hypothesized that students randomized to AVID would be exposed to more academically successful peers, resulting in more prosocial networks, and would have improved psychosocial wellbeing and reduced risky health behaviors. Because prior studies suggest the influence of schools on social networks and health behaviors may be stronger for boys versus girls, we aimed to test whether intervention effects vary by sex. 32-34 Finally, we examined whether social connections with AVID students were associated with risky peer networks and behaviors among high-performing students. Our initial study aimed to follow students through 11th grade, however the coronavirus
| 1,244 |
254819264
| 0 | 16 |
disease 2019 pandemicrelated school closures disrupted typical AVID delivery. Hence, we present our findings before COVIDrelated school closures, following student during their transition to high school, through the end of ninth grade. METHODS We conducted a multisite randomized trial testing whether AVID impacts adolescent health and wellbeing during the transition to high school among students attending schools in predominantly low-income, minority communities (Clinical Trials registration number NCT03059433). AVID Intervention AVID's secondary school program targets ninth through 12th grade students from demographic groups underrepresented in higher education who are performing in the "academic middle" (earning B or C average grades) and are less likely to be placed in and succeed in college preparatory coursework without additional academic and social support. 25 Students enter in ninth grade and are encouraged to remain in AVID through 12th grade. AVID students enroll in rigorous collegepreparatory courses, placing them in an academic track typically targeting higher-achieving students, and attend an AVID elective class during which teachers provide academic skills coaching, explain the college application process, and facilitate social-emotional skill development, including persistence in the face of challenges, problem solving, and coping skills. Finally, the program emphasizes the student and teacher relationship and cultivates a familylike atmosphere. 27 School Recruitment We partnered with a large urban school district in Southern California to recruit high schools (serving only grades 9-12) into the study. We invited schools that served lowincome minority families, had been certified by the national AVID office as achieving adequate program fidelity, and had more students who meet AVID eligibility criteria
| 1,245 |
254819264
| 0 | 16 |
than the AVID program had the capacity to serve. We sought to enroll 5 schools to achieve adequate power. Of the first 6 schools approached, 5 agreed to participate. School and participant recruitment took place over 2 consecutive school years (2017)(2018). Like the participating school district, participating schools served largely low-income Latinx students (Supplemental Table 2). Participant Recruitment At participating schools, AVID recruitment followed the school's typical practice, including presentations at feeder middle schools and student and parent meetings. Interested students completed an application and interview with the school's AVID coordinator. Eligibility for AVID participation included: eighth grade, grade point average (GPA) of 2.0 to 3.5, enrolling in ninth grade at a study school, student commitment to taking rigorous collegepreparatory courses, and parent permission for AVID participation. AVID-eligible students entered into a random admission lottery, regardless of study participation. Approximately twice as many students as each school had the capacity to serve entered the lottery (330 applicants for 138 AVID slots). Although study schools oversaw identification of students entering the lottery, the investigators conducted the lottery via a random number generator. A separate lottery was conducted for each school. Given the nature of the intervention, blinding was not feasible. Although students typically remain in AVID throughout high school, students were permitted to drop out of AVID over the course of the school year and open slots were filled, as per usual practice, on a first-come-first served basis, regardless of initial lottery result. These practices were agreed upon with each participating school before study initiation according to
| 1,246 |
254819264
| 0 | 16 |
the tenants of community-based participatory research. 35 All students entering the AVID lottery were eligible for study participation and received a study enrollment packet, including consent forms. Students returning a signed parental consent form and student assent form were enrolled in the study. Study participation had no bearing on AVID lottery results. Of the 2625 students matriculating into a study school, 330 entered the AVID lottery and 270 consented to participate in the study (participation rate 5 81.8%). Of those enrolled in the study, 124 "won" the lottery and were offered an AVID program spot (AVID group), whereas the remaining 146 participants were not offered an AVID spot (Control group) (Fig 1). Finally, we recruited a comparison group of high-performing incoming ninth grade students from the same schools at the same time as the AVID and Control groups. We identified these students (High performing group) by their grade point average of >3.5 during eighth grade. Of the 214 eligible students, 161 consented to participate in the study (participation rate 5 75.2%). Data Collection Students completed a baseline computer-assisted survey in school at the end of eighth grade or beginning of ninth grade (April-October), and follow-up interview at the end of ninth grade (May-June). At the time of survey administration, students were reminded that the study goal was to learn about schools, social networks, and substance use and all answers would remain confidential. There was no difference in survey completion or retention by study arm. Overall, 418 of the 431 initially enrolled students (117 AVID; 141 control;
| 1,247 |
254819264
| 0 | 16 |
160 high performing) completed the follow up survey (retention rate 5 97.0%). Of the 13 students lost to follow up, 12 switched schools, and 1 refused participation. Survey Measures Social Network: we assessed students' personal social networks using a standard procedure. 36 At baseline, students named 20 people (alters) in their network and answered questions describing each person. Alters could include friends or family. At follow up, students named 10 alters outside their family. At each wave, alters identified as "about my age" were considered peers. For each peer, students reported whether that peer is in AVID, has ever been drunk, and ever used marijuana. Participants also reported whether the peer "tries hard in school," "thinks it's important to do well in school," "thinks they should attend every class," "does not disrupt class," and "does not cause trouble." Peers having all of these characteristics were considered highly engaged in school. Psychosocial wellbeing: at baseline and follow up, students completed the Mental Health Inventory to assess general mental health (range 1-25, a 5 .80) 37 ; the Perceived Stress Scale (range 0-16, a 5 .61) 38 ; General Self-Efficacy Scale (range 8-40, a 5 .95) 39 ; Duckworth Grit Scale (range 13-40, a 5 .67) 40 ; and a 29-item school engagement scale from the High School Survey of Student Engagement (range 29-116, a 5 .96). 41 For these outcomes, higher scores indicate better mental health, more stress, and higher levels of self-efficacy, grit, and school engagement, respectively. Health risk behaviors: at baseline and follow up, using
| 1,248 |
254819264
| 0 | 16 |
measures from the Youth Risk Behavior Survey, 42 students reported their frequency of alcohol, marijuana, vaping device use, tobacco, or other drug use in the previous 12 months. We created dichotomous measures of any alcohol, any marijuana, any vaping, and any substance use in the prior 30 days and prior 12 months. Using measures from the National Longitudinal Study of Adolescent to Adult Health, 43 students reported their engagement in 8 delinquent behaviors in the previous 12 months (painting graffiti, damaging property, lying to a parent or guardian about where they had been or who they were with, stealing, running away from home, driving a car without permission, entering a house or building to steal something, using or threatening to use a weapon to get something from someone, or selling marijuana or other drugs) and if they had been in a physical fight in the last 12 months. Socio-demographic characteristics and intervention exposure: at baseline, students reported demographic (birthplace, home language, family structure, race and ethnicity) and parental characteristics (educational attainment, employment, birthplace), and whether they participated in AVID during middle school. At follow-up students reported whether they participated in AVID during the fall and spring semesters of ninth grade. Sex and grade point average came from eighth grade academic transcripts. Analytic Strategy T-test and x 2 analyses compared demographic characteristics and baseline health behaviors across groups. Intent-to-treat analyses tested whether intervention students had improved outcomes relative to control students. We used multilevel mixed effects models to account for clustering within schools and control for baseline
| 1,249 |
254819264
| 0 | 16 |
values of the outcome of interest. Psychosocial outcome measures were standardized on follow-up sample values with a mean of 0 and SD of 1. As prespecified, interaction terms tested whether effects were moderated by sex and, when significant, we estimated sexstratified models. Our prespecified primary outcomes on which the study was powered was the difference in proportion of prosocial peers in the network and difference in 30-day marijuana use between AVID and Control groups at the end of 11th grade. Although we present interim findings from the end of ninth grade, we maintained these primary outcomes. We also examined the odds of naming a prosocial peer in the network and past-12 month substance use. Finally, similar to other studies seeking to measure negative peer influence, 13,14,18 we tested whether naming an AVID student in the social network was associated with more risky networks and higher odds of substance use, violence, or delinquency for students in the high-performing comparison group. All outcome data were complete. RESULTS The sample is similar to low-income communities in Southern California with 82.8% identifying as Latinx (Table 1) and 78.9% reporting at least 1 parent born outside the United States. Just over half the sample (53.6%) had at least 1 parent who graduated high school and FIGURE 1 Recruitment and retention consort diagram. 22.0% participated in AVID during middle school. There were no significant differences in demographics, baseline health behaviors, social network, or psychosocial outcomes between AVID and control arms. However, compared with the AVID group, the high-performing group had significantly fewer
| 1,250 |
254819264
| 0 | 16 |
males, more Asian students, more students from immigrant families, and lower rates of fighting (12% vs 27%, P 5 .001) and marijuana use (2% vs 6%, P 5 .05). Of those who won the AVID lottery (AVID group), 81% participated in the program for at least 1 semester and 66% participated for both ninth grade semesters; 5% of the Control group participated in AVID for 1 semester and 5% for both semesters. Lottery "winners" who participated for both semesters were more likely to have a full-time working parent, less likely to report any delinquent behaviors at baseline, and had a higher eighth grade GPA compared with those who participated for just 1 semester (Supplemental Table 3). Table 4) revealed a greater proportion of peers who do not disrupt class (odds ratio [OR] 1.44, 95% confidence interval [CI] 1.11 to 1.86) but no difference in the proportion of substance using peers between AVID and control students (OR 0.95, 95% CI 0.73 to 1.28). However, the AVID Group had lower odds of naming a peer in their social network who has been drunk or used marijuana (OR 0.74, 95% CI 0.56 to 0.98) and higher odds of naming a peer who does not disrupt class (OR 1.23, 95% CI 1.07 to 1.41), was highly engaged in school (OR 1.73, 95% CI 1.11 to 2.70) and who was in AVID (OR 2.19, 95% CI 1.01 to 4.73), compared with the Control group. In addition (Fig 3), although there were no differences in 30-day substance use between groups, the AVID
| 1,251 |
254819264
| 0 | 16 |
group had lower odds of any substance use (OR 0.66, 95% CI 0.48 to 0.89) and any delinquent behaviors (OR 0.65, 95% CI 0.42 to 0.99) compared with the Control group. These effects did not vary by sex (interaction terms all >.05). Ad hoc analyses tested whether AVID was associated with lower odds of substance use, controlling for having a substance using peer, a highly engaged peer, or a peer in AVID (Supplemental Table 5), to explore whether intervention effects on substance use are explained by social network changes. Results suggest having a peer in AVID may account for some AVID effects on substance use. Intent-to-treat analyses (Fig 2, Supplemental For psychosocial wellbeing, intervention effects varied by sex (interaction terms <.05) for all outcomes except general mental health, hence we conducted sexstratified analyses. Although there were no differences between AVID and Control girls, for boys, the AVID group had lower levels of stress (b À.21, 95% CI À0.40 to À0.02) and higher self-efficacy (b .32, 95% CI 0.12 to 0.51), grit (b .28, 95% CI 0.04 to 0.52), and school engagement (b .23, 95% CI 0.05 to 0.41) (Fig 4 and boys. There were no effects on general mental health. As treated analyses examined outcomes among students who participated in 1 or 2 semesters of AVID relative to those who did not participate in AVID to check the robustness of our findings (Supplemental Table 7). Results show a similar pattern to intent-totreat analyses and improved outcomes for those who participated for 2 versus only 1 semester.
| 1,252 |
254819264
| 0 | 16 |
Finally, among high-performing students, naming an AVID-involved peer in the social network was associated with increased odds of naming a substance using peer in the network (OR 1.99, 95% CI 1.11 to 3.55 for marijuana use, OR 2.21, 95% CI 1.16 to 4.22 for marijuana or alcohol use), but was not associated with peer school-related behaviors nor self-reported health behaviors (Supplemental Table 8). DISCUSSION We found significant health benefits to adolescents randomized to AVID during the transition to high school, including more prosocial peer networks, decreased substance use, and, for boys, improved psychosocial wellbeing. These findings are notable because not only is this the first experimental study of AVID in the United States, but it demonstrates that academic interventions can have substantial spillover benefits to health. AVID expands access to rigorous courses for middle-performing students, thereby creating more connections between and among academically middle-and highperforming youth. In addition, AVID simultaneously improves health behaviors. We found that connections with AVID-related peers may partially explain the program's impact on substance use. Of note, this study focuses on the transition to high school, which may be a sensitive period when social FIGURE 3 Intervention effects on risky health behaviors. All models used intent-to-treat mixed-effects regression models with a random intercept for school and cluster-robust standard errors to account for clustering within schools, after adjusting for baseline values of the outcome of interest to test whether outcomes for students randomized to AVID differed from those randomized to the control group. Statistical significance is represented by a 95% confidence interval bar
| 1,253 |
254819264
| 0 | 16 |
that does not cross 1. FIGURE 2 Intervention effects on the odds of naming a peer in the social network with the following characteristics. All models used intent-to-treat mixed-effects regression models with a random intercept for school and cluster-robust standard errors to account for clustering within schools, after adjusting for baseline values of the outcome of interest to test whether outcomes for students randomized to AVID differed from those randomized to the control group. Statistical significance is represented by a 95% confidence interval bar that does not cross 1. networks are in flux and high-risk behaviors often emerge. Together, findings suggest the social network and health implications of academic tracking may be substantial and are critical to consider, particularly in light of critiques that low-income, Black, and Latinx students may be less likely to gain access to more advanced academic tracks. 44 Although not studied here, it is possible that low-performing and less engaged students may also benefit from increased access to rigorous college-preparatory courses. Applying AVID school-wide may be 1 strategy to accomplish this. Identifying the health effects of such an approach to more broadly reduce academic tracking can provide important insights into the public health implications of education tracking policies. Although multisite, all study schools were from the same district serving mostly low-income and Latinx students and findings are from only 1 school year. Future analyses are needed to test whether effects are generalizable and sustained over time and identify specific mechanisms through which AVID impacts substance use. Though all study schools met national
| 1,254 |
254819264
| 0 | 16 |
certification criteria, we did not directly observe or measure AVID implementation and cannot examine whether higher fidelity improves outcomes. We did not examine whether AVID achieves its primary objective of increasing college enrollment and persistence. Although we examine multiple outcomes, they are highly correlated. We rely on self-report substance use measures, though studies suggest high correlation with biologic testing, particularly when measured via computerized surveys employed here. 57 Blinding participants to study arm was not possible and we cannot rule out the possibility that AVID participants were more susceptible to social desirability bias. CONCLUSIONS Despite these limitations, this study has important implications for the role schools play in influencing adolescent health trajectories. First, AVID might be considered an evidence-based program that simultaneously prevents adolescent health risk behaviors and promotes wellbeing. More importantly, many of the FIGURE 4 Intervention effect on psychosocial wellbeing, stratified by sex. All models used intent-to-treat mixed effects models stratified by sex with a random intercept for school and cluster-robust standard errors to account for clustering within schools, after adjusting for baseline values of the outcome of interest to test whether outcomes for students randomized to AVID differed from those randomized to the control group. Statistical significance is represented by a 95% confidence interval bar that does not cross 1. All models used intent-to-treat mixed-effects regression models with a random intercept for school and cluster-robust standard errors to account for clustering within schools, after adjusting for baseline values of the outcome of interest to test whether outcomes for students randomized to AVID differed
| 1,255 |
254819264
| 0 | 16 |
from those randomized to the control group. *P values < .05. All models used intent-to-treat mixed-effects regression models with a random intercept for school and cluster-robust standard errors to account for clustering within schools, after adjusting for baseline values of the outcome of interest to test whether outcomes for students randomized to AVID differed from those randomized to the control group. *P values < .05. All models used mixed-effects regression models (linear and logistic) with a random intercept for school and cluster-robust standard errors to account for clustering within schools, after adjusting for baseline values of the outcome of interest. Models estimate whether the number of semesters of AVID participation is associated with each outcome, among AVID and Control study participants. *P values < .05. All models used intent-to-treat mixed-effects regression models (linear and logistic) with a random intercept for school and cluster-robust standard errors to account for clustering within schools, after adjusting for baseline values of the outcome of interest. Models estimate whether naming an AVID-related peer is associated with each outcome, among those in the high-performing group. P values less than .05 are bolded.
| 1,256 |
254819264
| 0 | 16 |
The Importance of Ethics in Modern Universities of Technology The twenty-first century will pose substantial and unprecedented challenges to modern societies. The world population is growing while societies are pursuing higher levels of global well-being. The rise of artificial intelligence (AI) and autonomous systems, increasing energy demands and related problems of climate change are only a few of the many major issues humanity is facing in this century. Universities of technology have an essential role to play in meeting these concerns by generating scientific knowledge, achieving technological breakthroughs, and educating scientists and engineers to think and work for the public good. The twenty-first century will pose substantial and unprecedented challenges to modern societies. The world population is growing while societies are pursuing higher levels of global well-being. The rise of artificial intelligence (AI) and autonomous systems, increasing energy demands and related problems of climate change are only a few of the many major issues humanity is facing in this century. Universities of technology have an essential role to play in meeting these concerns by generating scientific knowledge, achieving technological breakthroughs, and educating scientists and engineers to think and work for the public good. The aim of this Special Issue of Science and Engineering Ethics is to examine some of the ethical issues that arise for institutions of higher education in the field of engineering and applied science in meeting these challenges. In so doing, it highlights two specific areas. First, it considers the ethical issues that arise for institutions of higher education in the area of
| 1,257 |
209474634
| 0 | 16 |
engineering and applied science. Its focus is on specific issues at universities of technology, more specifically on the relationships both of individual academic researchers, and of institutions themselves, with industrial partners, commerce and innovation in for-profit organizations. The second area of focus is the matter of educating a new generation of engineers and scientists so that they will be equipped to deal with the future challenges that mankind faces, while also observing the highest moral standards of academic conduct and research integrity. In the arena of AI, events in 2019 have brought to light a number of cases that highlight increased sensitivity within and beyond the academic community to morally problematic crossover and interconnections between the corporate world and academia. For example, the director of the Massachusetts Institute of Technology 1 3 (MIT) Media Lab stepped down as a result of significant controversy regarding the acceptance of funding from Jeffrey Epstein, the disgraced financier who faced sex trafficking charges (Tracy and Hsu 2019). In addition, Google's AI ethics council (the Advanced Technology External Advisory Council), with members from academia, the corporate world, and think tanks, was dissolved immediately after the membership of an anti-LGBT advocate on the committee led to controversy among Google workers and in the media (Levin 2019). Moreover, because the stakes and commercial interests in AI research have become so significant, there are a variety of efforts underway to influence both the course of research into the ethics of AI, and the future of regulation and governance. Facebook has made a $7.5 million
| 1,258 |
209474634
| 0 | 16 |
(US) donation to the Technical University of Munich, establishing a new center devoted to the ethics of artificial intelligence (Kahn 2019). The private equity firm, Blackstone, has donated $180 million US dollars to Oxford University for both an ethics center, and for humanities and social sciences research into AI (Williams 2019;Reuters 2019). Further, the Wallenberg foundation is funding ethics and AI research in Sweden (WASP-HS 2019). This Special Issue has grown out of a workshop on 'Science and Integrity in the Modern University' that was held at Delft University of Technology in the Netherlands in 2013. 1 This workshop was organized in order to reflect on the so-called 'Valorization task' (i.e., the assignment to contribute effectively in addressing the societal challenges posed by technology) that universities, and more specifically universities of technology, are increasingly expected to assume as one of their core tasks. 2 The term 'valorization' refers to a process of facilitating knowledge transfer and, ideally, it creates benefits for society because scientific knowledge can then be translated into tangible results. While valorization is not necessarily a new phenomenon for universities of technology, the close and extensive collaboration between independent researchers working in academia and industry can raise intricate ethical questions. These questions need to be fully acknowledged and addressed. An ethics infrastructure is therefore indispensable when it comes to dealing with ethical issues that are specific for universities of technology. These issues are not only relevant to research, but they should also have a significant influence in shaping the education of future engineers. This
| 1,259 |
209474634
| 0 | 16 |
Special Issue is particularly aimed at considering both concerns for universities of technology such as the issues and potential conflicts associated with research funding, and also challenges associated with teaching broadlyoriented future scientists and engineers. This issue is organized such that each original research article is accompanied by a commentary. Comment pieces are not critiques of the arguments in the research article, nor are they summaries of the primary article. Rather, they are short essays 1 3 The Importance of Ethics in Modern Universities… on the same or a related topic aimed at providing the reader with an additional perspective. Further, this Special Issue has a very broad international viewpoint. Authors report on and reflect fascinating developments in the US, the Netherlands and Spain, but also in China and South Africa. Ethical Challenges in Research at Technical Universities Scientific research is expected to be in the service of society and certainly not to be at odds with the responsible pursuit of intelligent solutions to its problems. Among research funding agencies, it is commonly assumed that an important indicator for measuring social relevance is the willingness of industry to invest in research. In many countries, there is an upsurge in government policies aimed at encouraging academic-industry collaborations. Universities are expected to facilitate knowledge transfer to industry by means of systematic collaborations. Indeed, in some instances governmental funding schemes for independent research depend on co-funding by industry. The rationale behind these policies is that research in which industry is willing to invest is marketable and, hence, socially relevant.
| 1,260 |
209474634
| 0 | 16 |
Especially for universities of technology, a university's research income increasingly depends on these collaborations. This raises the question of how to design and implement institutional arrangements in order to anticipate and deal with potential conflicts of interest that might occur, and to address the effects these could have on the independence of scientific judgment. The emphasis on interactions between the academic domain and the market domain raises concerns that have been discussed by philosopher Michael Walzer, author of Spheres of Justice (Walzer 1983). He has argued that different spheres in society have their own normative logics. Many societies try to prevent the intermingling of the normative logics (e.g., expressed in the governance of institutions) of these separate spheres, that is, they attempt to ensure that criteria for the distribution of goods in one sphere are not used to allocate advantages in another sphere. In order to prevent this cross contamination, Western, democratic societies typically practice what Walzer calls 'the art of separation' of spheres. For example, in Western, liberal democracies constraints are put on what money can buy and it is a widely held view that appointments to political office should be kept separate from commercial considerations; that is, money should not be allowed to buy influence or power in the political sphere. There are norms and rules that govern the allocation of political responsibility, namely democratic elections. Further, eligibility for medical treatment should be kept separate from someone's status in the political sphere and therefore, priority on a waiting list for heart surgery should depend
| 1,261 |
209474634
| 0 | 16 |
on 'need', not on wealth or status. Walzer describes how some societies have blocked exchanges at the boundaries of social spheres so that family relationships cannot facilitate admission to a university, or that a university degree cannot influence eligibility to receive health care. Similarly, boundaries between the sphere of science and the market should be critically monitored so that financial gain does not compromise independent scientific judgment. Norms concerning proper conduct in science cannot be replaced by norms governing market behavior and the profit motive. The pursuit of scientific truth, or of a better understanding of the world, should in principle be kept separate from commercial benefits associated with it. Modern scientists run the risk of becoming the victim of role confusion and conflicts of interest similar to the company doctor who inhabits two worlds i.e., the doctor's clinical judgment runs the risk of being compromised or undermined by his allegiance to economic thinking within the company. The scientist in the age of commerce could similarly acquire some of the tragic features of the company doctor-normative confusion by design-unless the scientist's responsibilities and loyalties are clearly defined and separated and the interplay is made transparent. A university's ethics infrastructure should promote and facilitate this separation and avoid any semblance of conflicts of interest. Such infrastructure should at least include arrangements, rules and institutions to facilitate raising an ethical dilemma (e.g., a breach of integrity or a question about the use of human subjects in research) and to enable addressing this issue further. In 'Values in University-Industry
| 1,262 |
209474634
| 0 | 16 |
Collaborations: The Case of Academics Working at Universities of Technology', Rafaela Hillerbrand and Claudia Werker discuss the challenges of scholars working at universities and in industry (Hillerbrand and Werker 2019). They specifically focus on the role of an individual scholar, who may run into serious conflicts when working on joint university-industry projects. While universities aim to disseminate knowledge, industry aims to appropriate knowledge. This role confusion can lead to ethically problematic and complex situations and conflicts of interests. In his commentary, 'The Need for a Code of Conduct for Research Funders', Bert van Wee argues that the attention only on researchers in the ethics literature is insufficient: the focus needs to expand to include a code of conduct for funders of research ( van Wee 2019). Van Wee's commentary provides tangible recommendations such as 'policy relevant research should not be contracted and supervised by a client with an interest in the outcomes', and 'policy relevant research should always be examined by an independent institute'. In his contribution 'Institutional Conflicts of Interest in Academic Research', David Resnik extends discussions of conflicts of interest to the level of institutions (Resnik 2015). For example, institutional officials may have individual financial relationships that may inappropriately influence decision-making, and, together or separately, can give rise to an institutional conflict of interest (iCOI). In their commentary, 'Current Perspectives Regarding Institutional Conflict of Interest', Ann Nichols-Casebolt and Francis Macrina argue that academic institutions must develop strategies to remediate the unique challenges in iCOI, including clarifying the definition of iCOI and implementing a well-designed
| 1,263 |
209474634
| 0 | 16 |
electronic database for reporting and managing iCOI across multiple leadership constituencies (Nichols-Casebolt and Macrina 2015). Seumas Miller, in his contribution 'Whither the University?: Universities of Technology and the Problem of institutional Purpose', addresses the need to provide an appropriate normative conception of the modern university (Miller 2019). Such conception could help to admit differences between universities of technology and other universities. Building on the teleological normative theory of social institutions which implies that universities are to be considered organizations that provide collective goods by means of joint activity, Miller discusses the fundamental collective good(s) that universities of technology ought to provide. He argues that the absence of a normative conception is partially masked by the process of institutional evolution that has actually been taking place at universities. In her commentary, 'The Survival Imperative', Stephanie Bird delves further into the evolutionary requirement that a species pass on its 'survival knowledge' to the next generation (Bird 2019). Humans as a species tend to be too clever, powerful, ignorant and arrogant for our own good, and the good of the planet. It is essential that humanity and its societies determine how to more effectively teach future generations the key information they need to address their limitations and survive. The last two pieces in this part focus on fascinating new efforts in China and South Africa. In their paper, 'Ethics "Upfront": Generating an Organizational Framework for a New University of Technology', Penelope Engel-Hills, Christine Winberg and Arie Rip highlight an expectation in post-apartheid higher education in South Africa that technikons
| 1,264 |
209474634
| 0 | 16 |
(institutions similar to the British polytechnics) will be/should be converted to universities of technology (Engel-Hills et al. 2019). They discuss one of the new South African universities of technology as a case study and, more specifically, the opportunity to build a new university such that ethics could be placed 'upfront', rather than coming as an afterthought. This ethics upfront approach requires constructing an organizational framework that makes ethical issues integral to management and decision-making processes. In their commentary 'Development of Ethics Education in Science and Technology in Technical Universities in China', Qian Wang and Ping Yan introduce the specific situation and characteristics of ethics education in science and technology at Chinese technical universities (Wang and Yan 2019). China's ethics education in science and technology in China's five technical universities (also known as the 5TU) emphasizes the use of traditional ideological and cultural resources and practical cases. Teaching methods combine traditional Chinese ethics with non-Chinese experience and teaching methods, and aim at cultivating students' ability to solve ethical problems in the real world. Challenges in Teaching Ethics to Engineers and Scientists The second part of the special issue contains papers that mainly focus on teaching endeavors. In their piece 'Ethics Across the Curriculum: Prospects for Broader (and Deeper) Teaching and Learning in Research', Carl Mitcham and Elaine Englehardt assert that the movements to teach the responsible conduct of research (RCR) and engineering ethics at technological universities is not receiving enough scholarly attention; they argue that RCR should be seen as a part of the broader ethics across
| 1,265 |
209474634
| 0 | 16 |
the curriculum (EAC) movement that is receiving more scholarly attention (Mitcham and Englehardt 2016). The authors compare EAC initiatives at different universities, including the successful one at Utah Valley University that gave birth to EAC as a scholarly movement, and the one at the Colorado School of Mines that manifests continuing institutional resistance to EAC. In their commentary, 'Teaching Engineering Ethics to PhD Students: A Berkeley-Delft Initiative', Behnam Taebi and William Kastenberg draw a similar comparison between the University of California at Berkeley and Delft University of Technology (Taebi and Kastenberg 2016). The commentary highlights a variety of academic and institutional challenges at these two universities, when jointly teaching a graduate engineering ethics course first at UC Berkeley and later at Delft University. The authors argue that both a bottom-up approach at the level of the faculty and as a joint research and teaching effort, and a top-down approach that includes recognition by a University's administration and the top level of education management, are needed for successful and sustainable efforts to teach engineering ethics. Mary Sunderland in her 'Using Student Engagement to Relocate Ethics to the Core of the Engineering Curriculum' considers the core problem of perception with engineering ethics education: while ethics is meant to be a central component of today's engineering curriculum, it is often perceived as a marginal requirement to be fulfilled (Sunderland 2013). There is further a mismatch between the faculty's perceptions of ethics as emphasizing the nuances and complexity of engineering ethics, while students tend to perceive ethics as laws, rules,
| 1,266 |
209474634
| 0 | 16 |
and codes that must be memorized and reproduced. Sunderland describes a student engagement approach to pedagogy that includes students as active participants in curriculum design, which could help relocate ethics from the periphery to the core of the engineering curriculum. In his commentary 'Ethics and the UN Sustainable Development Goals: The Case for Comprehensive Engineering', Jeroen van den Hoven explores another important aspect in engineering curricula that should accommodate 'comprehensive engineering', as an approach that could help to accommodate ethical coherence, consilience of scientific disciplines, and cooperation between parties (van den Hoven 2016). Comprehensive engineering is key if engineers are to adequately and responsibly respond to the global problems that the world is facing, such as those formulated in the United Nations Sustainable Development Goals. Alejandra Boni, José Javier Sastre and Carola Calabuig, in their article 'Educating Engineers for the Public Good Through International Internships: Evidence from a Case Study at Universitat Politècnica de València', discuss a different approach to creating awareness among engineering students about their social responsibility through an internship program that places engineering students in countries of Latin America in order to expose them to the implications of being a professional in society in a different cultural and social context (Boni et al. 2015). An integral part of this program is a reflection on the dynamic relationship between technology and society by creating space before and during the internship, and upon the return of the students, to discuss and collectively reflect upon their lived experience. Colleen Murphy and Paolo Gardoni, in their commentary
| 1,267 |
209474634
| 0 | 16 |
'Understanding Engineers' Responsibilities: A Prerequisite to Designing Engineering Education', emphasize that all activities in engineering education, including study time abroad as well as internships, must be based on a comprehensive understanding of engineers' responsibilities (Murphy and Gardoni 2017). Globalization has implications for these responsibilities and international internships can play an important role in fostering the requisite moral imagination of engineering.
| 1,268 |
209474634
| 0 | 16 |
Structural insights into the transcription activation mechanism of the global regulator GlnR from actinobacteria Significance Two-component signal transduction systems play essential roles in cellular adaptation to complex environmental stimuli through a histidine kinase sensor and a response regulator (RR). In actinobacteria, an orphan RR of the OmpR/PhoB subfamily proteins called GlnR globally coordinates transcription of genes involved in nitrogen, carbon, and phosphate metabolisms. However, the underlying mechanism remains obscure. Here, using crystallography, cryo–electron microscopy, and biochemical assays, we demonstrate that GlnR activates transcription by interacting with DNA elements through cooperating with σ region 4 and RNAP β flap subunit. We also identify previously unobserved collaborative interfaces between four GlnR protomers and the promoter DNA or RNAP conserved domains. These interactions both retain the stability of the GlnR-dependent transcription activation complex (GlnR-TAC) and promote efficient transcription initiation. GlnR | crystal structure | cryo-EM structure | GlnR-dependent transcription activation complex (GlnR-TAC) | Mycobacterium tuberculosis Actinobacteria are a group of gram-positive, spore-forming, high GC-containing, largely filamentous bacteria, which are widely distributed in soil.They can produce an array of bioactive secondary metabolites which contribute to about half of clinically used antibiotics and important pharmaceutical agents (1)(2)(3)(4).Bacterial biosynthesis of these metabolites depends on the availability of carbon, phosphate, and nitrogen, especially nitrogen plays a pivotal role in bacterial survival and growth (5)(6)(7).To acquire more available nitrogen sources from the constantly changing environment, bacteria have developed a variety of complex transcription regulatory systems to sense the status of internal and external nitrogen by coordinating expression of genes involved in nitrogen metabolism (5,8).
| 1,269 |
261296163
| 0 | 16 |
Two-component systems (TCSs), typically consisting of a membrane-associated histidine kinase (HK) and a cognate intracellular response regulator (RR), are one of the most simple signal transduction systems (9)(10)(11).RRs are involved in the regulation of many processes, including metabolism, motility, quorum sensing, virulence, and antibiotic resistance (12)(13)(14)(15).The largest group of RRs identified in bacteria are the OmpR/PhoB subfamily proteins.It has been well documented that in low-GC-content enteric bacteria, the TCS NtrB-NtrC is used to regulate the expression of genes involved in nitrogen metabolism (16,17).While in high-GC-content actinobacteria, the metabolisms of nitrogen, carbon, and phosphate are controlled by GlnR, a global OmpR/PhoB subfamily protein.Since no cognate HK has been identified, GlnR is regarded as an orphan RR that acts independently of Asp phosphorylation (5,18).It has, however, been proposed that the protein is regulated by the posttranslational modifications of Ser/Thr/Tyr phosphorylation or acetylation (19).Comprising an N-terminal receiver domain (GlnR-REC) and a C-terminal DNA-binding domain (GlnR_DBD), GlnR and its homologs share high homology in sequence (Fig. 1 A, Top and SI Appendix, Fig. S1).Variations of GlnR are widely distributed in actinobacteria including the epidemic pathogen Mycobacterium tuberculosis (M.tuberculosis), the soil-dwelling saprophyte Mycobacterium smegmatis (M.smegmatis), the model organism Streptomyces coelicolor (S. coelicolor), the erythromycin-producing actinobacteria Saccharopolyspora erythraea (S. erythraea), and the rifamycin-producing industrial actinobacteria Amycolatopsis mediterranei (A.mediterranei) (2,5,7,8,18,20). Up to date, GlnR has been characterized as a pleiotropic transcription regulator for expression of ~1,000 genes (5,8,14,18).The protein can act as an activator or a repressor, mainly depending on its binding position in promoter DNA.The crystal structures of GlnR-RECs from M. tuberculosis
| 1,270 |
261296163
| 0 | 16 |
and A. mediterranei have been determined, and biochemical analyses have demonstrated that GlnR can form a functional dimer through its α4-β5-α5 interface from the GlnR-REC (20).Although the canonical GlnR-binding cis element was defined for more than 10 y (21), many studies failed to determine the crystal structures of full-length GlnR, GlnR_DBD, or GlnR_DBD-DNA binary complex.This difficulty may have resulted from the flexible loop connecting the N-terminal domain and the C-terminal domain of GlnR, which causes improper crystal packings. The conserved GlnR cis element encompasses two 22-bp GlnR boxes separated by six nucleotides.Each GlnR box contains two GlnR binding sites (a site and b site) (Fig. 1 A, Bottom).One GlnR box is composed of "a1 site" (gTnAc) and "b1 site" (GaAAC); the other GlnR box includes "a2 site" (gTnAc) and b2 site (GaAAC) (21,22).Recently, more and more GlnR target genes and their corresponding cis-elements have been identified, some of which are considered atypical.The amtB promoter of S. coelicolor was found to harbor three GlnR binding boxes consisting of a3-b3, a1-b1, and a2-b2 sites.The nas operon promoter of A. mediterranei also identified to contain the previously deduced 22-bp GlnR-binding consensus sequences with the a1-b1 and a2-b2 sites (22).However, in vitro biochemical and in vivo mutational assays showed that only three of the above four GlnR binding sites are essential for GlnR-dependent transcription activation (23), further highlighting the complicated nature of GlnR and its DNA-binding elements.Further investigations need to be performed to get a better understanding of the molecular mechanism. Transcription initiation is a key step for gene expression,
| 1,271 |
261296163
| 0 | 16 |
making it a predominant target for transcription regulation in bacteria.During this phase, it generally includes the assembly of the multi-subunit RNA polymerase (α 2 ββ′ωσ) holoenzyme.Then, the holoenzyme engages promoter DNA to form an RNA polymerasepromoter closed complex (RPc) which subsequently isomerizes into a catalytically competent RNA polymerase-promoter open complex (RPo) (24)(25)(26).As to promoters containing nonoptimal consensus elements (−35 element and −10 element), various transcription activators will cooperate with RNA polymerase to form transcription activation complexes (TACs), which then promote transcription initiation (27).So far, several classic bacterial transcription activators have been well studied, such as catabolite activator protein (CAP) and the direct transcriptional activator of the superoxide regulon (SoxS) in Escherichia coli, which have only one cognate binding box located at different promoter sites (27)(28)(29)(30)(31)(32)(33)(34).However, as to promoters containing two or three binding boxes as those of GlnR, the underlying transcription activation mechanisms still remain unexplored. In this report, we determined a crystal structure of GlnR_DBD in complex with its regulatory cis-element DNA and a cryo-EM structure of GlnR-dependent transcription activation complex (GlnR-TAC) comprising M. tuberculosis RNA polymerase (RNAP), M. tuberculosis GlnR (GlnR) protein, and a promoter containing four well-characterized conserved GlnR binding sites.These structures define the precise interactions between GlnR and its conserved cis-element DNA and reveal how GlnR protomers initiate transcription through specific interactions with promoter DNA and β flap domain, region 4 domain of σ A subunit (σ A R4), and N-terminal domain of RNAP alpha subunit (αNTD) from RNAP holoenzyme.In particular, four GlnR_DBDs synergistically bind around the upstream and downstream of -35
| 1,272 |
261296163
| 0 | 16 |
element in a head-to-tail manner, and four GlnR-RECs coordinately bridge domains of RNAP (αNTD and β flap) with GlnR_DBDs, stabilizing the GlnR-TAC complex.Additionally, one conserved C-terminal domain of RNAP alpha subunit (αCTD) engages both DNA and GlnR IV _DBD, promoting stabilization of the complex.In summary, our results offer a global molecular mechanism for actinobacteria-derived GlnR-dependent transcription activation and reveal a unique mode of bacterial transcription regulation. Co-crystal Structure of SaeGlnR_DBD Bound to Its Cis-Element DNA.To understand the structural basis of GlnR′s DNAbinding specificity to its conserved GlnR-binding cis element, we attempted to crystallize full-length GlnRs or GlnR-DBDs from several representative actinobacteria with high similarity (Fig. 1 A, Top and SI Appendix, Fig. S1).Fortunately, we finally determine a 2.95-Å resolution crystal structure of S. erythraea GlnR_DBD (SaeGlnR_DBD) bound to a designated conserved cis-element which includes two GlnR binding sites (a site and b site with a consensus sequence of "GTAAC") (Fig. 1 A, Bottom).The parameters of data collection and model refinement are summarized in supplemental SI Appendix, Table S1.The GlnR_DBD (residues 120 to 223) consists of three helices and two β-sheets made up of seven β strands and forming a typical winged-helix-turn-helix structural topology of the OmpR/PhoB subfamily proteins (35-39) (Fig. 1B).In in one asymmetric unit of the co-crystal structure, two GlnR_DBD molecules bind in tandem to one blunt-ended dsDNA molecule.The mFo-Fc electron density map shows unambiguous, sharp densities for all nucleotides and residues from DNA and proteins, respectively (SI Appendix, Fig. S2A).The α helices from the GlnR_DBD molecules contact the major grooves of the dsDNA, while
| 1,273 |
261296163
| 0 | 16 |
the C-terminal β-hairpins interact with the adjacent minor grooves in a head-to-tail orientation (Fig. 1B and SI Appendix, Fig. S2A).This head-to-tail arrangement for DNA recognition is similar to those for other OmpR/PhoB subfamily members like PhoB and PmrA (35,36).The specific DNA-protein recognition mainly includes direct hydrogen bond interactions, hydrophobic interactions, and Van der Waals forces (Fig. 1 B and C and SI Appendix, Fig. S2A).As to a site from the dsDNA, SaeGlnR I _DBD residues T149, K151, W177, H191, and T187 and residues R169, R193, R196, T209, R211, N212, and Y215 form direct hydrogen bonds with the backbone phosphates of nucleotides from the non-template-strand DNA (−2C, −3G, and −4T) and the template-strand DNA (−6T, −7G, −8T, and −9A), respectively.By analogy, the residues listed above from SaeGlnR II _DBD also form direct hydrogen bonds with the backbone phosphates of the nucleotides (−13G, −14G, −15T, −17T, −18G, −19T, and −20G) from the b site.Specifically, the interactions between SaeGlnR I _DBD and the a site are highly similar to those between SaeGlnR II _DBD and the b site (Fig. 1 B and C and SI Appendix, Fig. S2).Residue R194 is positioned to make a hydrogen bond with O6 of −3G (−14G of the b site) from the non-template-strand DNA.Residue V190 makes van der Waals interactions with the C5-methyl groups of −4T (−15T from the b site) from the non-template-strand DNA and −6T (−17T from the b site) from the template-strand DNA.The side chain atoms of R193 make wan der Waals interactions with the C5-methyl group of −5T (−16T from
| 1,274 |
261296163
| 0 | 16 |
the b site) from the non-template-strand DNA.Residue R186 is positioned to make hydrogen bonds with both O6 and N7 of −7G (−18G from the b site) from the non-template-strand DNA.These specific interactions between GlnR_DBD and the DNA bases afford the binding selectivity of GlnR to the GTAAC consensus sequence. Structural analysis also shows that SaeGlnR I _DBD clearly makes contact with SaeGlnR II _DBD (SI Appendix, Fig. S2 B and C).Residues R165, K216, and N212 from SaeGlnR I _DBD make hydrogen bonds with residues S135, T136, T138, and Y150 from SaeGlnR II _DBD.The hydrophobic interactions formed by residues V166, V210, and V213 from SaeGlnR I _DBD, along with the π-π stacking forces constituted by residues Y137 and Y150, also contribute to the stabilization of the SaeGlnR_DBD-DNA complex. Overall Structure of GlnR-TAC.To obtain the structure of GlnR-TAC, we constructed one DNA scaffold harboring GlnRdependent promoter amtB (from -64 to +13, positions numbered relative to the transcription start site) (Fig. 2A) (22,23).The promoter includes two GlnR binding boxes with typical a1, b1, a2, and b2 sites, a nonoptimal -35 element, a consensus -10 element, a 13-nucleotide transcription bubble, and an 11-bp downstream double-stranded DNA.The SDS-PAGE analysis of the purified complex showed that M. tuberculosis RNAP holoenzyme and M. tuberculosis GlnR were included as expected, indicative of a well-assembled GlnR-TAC (SI Appendix, Fig. S3 A and B).Meanwhile, we constructed three fragments of promoter DNA containing different GlnR-dependent promoters followed by one Mango III sequence encoding a fluorogenic aptamer (SI Appendix, Fig. S4).Upon addition of the purified M. tuberculosis GlnR,
| 1,275 |
261296163
| 0 | 16 |
M. tuberculosis RNAP showed obvious transcription activities on previously identified M. tuberculosis promoter narG DNA and the amtB DNA containing four GlnR binding sites (amtB4 DNA containing a1, b1, a2, and b2 sites) or six GlnR binding sites (amtB6 DNA containing a1, b1, a2, b2, a3, and b3 sites) (SI Appendix, Fig. S3C).This suggests that the purified M. tuberculosis GlnR promotes the formation of M. tuberculosis GlnR-TAC, making it functionally relevant. Then, the cryo-EM datasets were collected and processed, one intact GlnR-TAC structure at a nominal resolution of 3.70 Å was finally determined (Fig. 2 and SI Appendix, Figs.S5 and S6 and Table S2).The densities for the RNAP holoenzyme and the downstream DNA are unambiguous, and each component from the structure of M. tuberculosis RPo (PDB ID: 6VVY) could be well fitted into the cryo-EM map (Fig. 2C and SI Appendix, Fig. S7).The calculated local resolution is ~3.0 to 4.5 Å for the core RNAP and ~5.5 to 7.5 Å for the peripheral αCTD and GlnR, indicative of their flexibility (SI Appendix, Fig. S6C).Consistent with the previous biochemical and genetic experiments (21-23), we found that in the GlnR-TAC structure, four GlnR protomers simultaneously associate with the two GlnR binding boxes of the promoter DNA in a head-to-tail fashion (Fig. 2 B and C).Each GlnR_DBD uses its α8 helix and β-hairpin to make specific contacts with both the a and b sites, similar to those in the binary structure formed by SaeGlnR_DBD and its cis-element DNA (Fig. 3 and SI Appendix, Figs.S2 and S7 B-E).Strikingly, our results
| 1,276 |
261296163
| 0 | 16 |
differ from the previously reported classic transcription activation complexes (27-29, 31, 33, 34), as we found that GlnR I _ DBD binds to the b2 site located downstream of the −35 element and simultaneously interacts with β flap and σ A R4 of RNAP (Fig. 4 and SI Appendix, Fig. S7 F and G).Moreover, the density of αCTD shows that αCTD engages both the a1 site and GlnR IV _ DBD (Fig. 5 and SI Appendix, Fig. S7H).These observations suggest that the structural architecture of GlnR-TAC is established in a unique transcription activation mode distinct from the reported class I and class II complexes. Four GlnR Protomers Coordinately Engage the Promoter DNA in GlnR-TAC.Although previous studies have shown that the number of GlnR binding boxes varies from one to three in different target promoters, the classic two GlnR binding boxes are found in most GlnR target genes (21)(22)(23)40).To uncover the molecular mechanisms, the amtB promoter which contains two typical GlnR boxes (including a1, b1, a2, and b2 sites) was chosen to assemble GlnR-TAC.In GlnR-TAC, we found that four M. tuberculosis GlnR protomers engage their corresponding GlnR binding sites located at the DNA major grooves by using the α8 recognition helix from each GlnR_DBD (Fig. 3 A and B and SI Appendix, Fig. S7 B and C).This binding bends the linear dsDNA gently with a curvature of ~50°, which is larger than that previously observed in the PhoB_DBD-DNA complex (41).This enables GlnR to easily contact the core enzyme of RNAP and further stabilize the transcription initiation complex.The
| 1,277 |
261296163
| 0 | 16 |
binding mode of these M. tuberculosis GlnR_DBDs to DNA shares high similarity with that in the co-crystal structure formed by SaeGlnR_DBD and its conserved cis-element DNA (Figs. 2 and 3 A-D and SI Appendix, Figs.S2, S7 B and C, and S8A).A cluster of residues consisting of R188, T189, R195, R196, H193, R171, and R198 from the α8 helix, T151, K153, and W179 from the linker connecting the α6 and α7 helices, and Y217, T211, R213, and N214 from the C-terminal β hairpin of GlnR II _DBD possibly makes extensive contacts with the major and minor grooves of the a1 site by salt bridges, hydrogen bonds, and van der Waals forces (Fig. 3 B, Left and SI Appendix, Fig. S7B).Analogously, the same residues from GlnR I _DBD may also contribute to the stabilization of the interface between GlnR III _DBD and the b1 site DNA (Fig. 3 B, Right and SI Appendix, Fig. S7B).For GlnR III _DBD and GlnR IV _DBD, similar interactions are probably formed in the interfaces between GlnR_DBD and the a2/b2 sites (SI Appendix, Figs.S7C and S8A).Besides, there also exist multiple types of interactions between two adjacent GlnR_ DBDs.Similar to the interface between two SaeGlnR_DBDs, interactions between two adjacent GlnR_DBDs are possibly stabilized by hydrogen bonds, hydrophobic interactions, and π-π stacking forces (Fig. 3 C and D and SI Appendix, Figs.S7 D and E and S8 B−E).Consistently, substitutions of key residues (W179, R188, R196, R213, and Y217) involved in the GlnR_DBD-DNA interfaces displayed apparent defects in transcription activity, DNA binding, and GlnR-TAC formation.As identified
| 1,278 |
261296163
| 0 | 16 |
by our in vitro transcription assays and EMSA experiments, the mutation of residues (E136 and K218) involved in GlnR_DBD-GlnR_DBD interfaces also resulted in reduced GlnR-dependent transcription activity and ratio of GlnR-TAC formation (Fig. 3E and SI Appendix, Fig. S9 A and D).These findings not only provide evidence for the different interface interactions but also demonstrate their physiological importance in maintaining the special "head-totail" engagement of GlnR protomers. GlnR-TAC Is Stabilized by Complex Protein-Protein Interactions between GlnR and the Conserved Domains of RNAP.In addition to the GlnR-DNA interactions described above, GlnR-TAC is also stabilized by distinctive and complex protein-protein interactions between GlnR and the conserved domains (β flap, σ A R4, αCTD, and αNTD) of RNAP.We observed that in M. tuberculosis GlnR-TAC, GlnR I _DBD not only recognizes the b2 site located downstream of the -35 element but also interacts with the β flap domain from the RNAP holoenzyme (Fig. 4A and SI Appendix, Fig. S7F).Residues Y160, H164, R167, V168, F169, and Q173 from GlnR I _DBD as well as K763, L764, G765, E768, G810, E811, and E831 from the loop structure of β flap.This probably occurs through salt bridges, hydrogen bonds, and van der Waals forces, acting as a glue interface to stabilize the GlnR-TAC. Moreover, structural analysis also reveals additional interactions between GlnR II _DBD and σ A R4 of RNAP (Fig. 4B and SI Appendix, Fig. S7G).Residues R188, T187, G186, G185, F184, D182, A172, and H176 from GlnR II _DBD possibly make hydrogen bonds, salt bridges, and hydrophobic and van der Waals forces with
| 1,279 |
261296163
| 0 | 16 |
residues E501, Q505, S508, K509, and S512 from the conserved helix of σ A R4. Consistent with this finding, substitutions of the residues greatly reduced GlnR-dependent transcription activity and formation of GlnR-TAC, especially in H164 and R167 (SI Appendix, Figs.S9 B-D and S10C).In comparison with σ A R4 in M. tuberculosis RPo (PDB ID: 6VVY), the interface between GlnR II _DBD and σ A R4 partially occludes σ A R4 from binding to the -35 element (SI Appendix, Fig. S10 A and B).This coincides with the previous observations of the crystal structure of ternary complexes including PhoB_DBD, σR4, β-flap-tip-helix, and DNA (PDB ID: 3T72) (38), suggesting a common charge-based code between transcription factors and σ A R4. Although increasing evidence has established that αCTD plays an important role in bacterial transcription activation, αCTD is mostly invisible in these transcription activation complexes due to its high flexibility, especially when the upstream DNA is far away from the −35 element (29,32).Nevertheless, the αCTD of M. tuberculosis RNAP locates adjacent to the upstream promoter DNA of GlnR-TAC, contacting the promoter region that contains the a1 site (Fig. 5 A and B and SI Appendix, Fig. S7 H, Left).Residues T257, N262, R259, R266, R287, N288, and K292 from the conserved 265 determinant of αCTD may promote interactions with the DNA backbone phosphate.Residues D182, F183, F184, and G185 from GlnR IV _DBD are prone to make polar contacts with residues R266, K265, and N262 from αCTD (Fig. 5 C and D and SI Appendix, Fig. S7 H, Right).These data support the
| 1,280 |
261296163
| 0 | 16 |
notion that αCTD activates GlnR-dependent transcription by directly interacting with both the upstream DNA and GlnR IV _DBD.This distinguishes from the previous assertions that it only interacts with either the activator or the UP element DNA (27)(28)(29)34).The distinctive αCTD-DNA and αCTD-GlnR interfaces discovered in this study provide more evidence for the highly versatile activation modes of RNAP αCTD in bacterial transcription regulation.Four GlnR_RECs Synergistically Bridge GlnR_DBDs and the RNAP Core Enzyme.As mentioned above, four GlnR IV _DBDs are visualized in detail, but we had been unable to obtain a highresolution density map of the four GlnR_RECs in an intact GlnR-TAC structure.However, based on the previously reported crystal structures of GlnR_REC, KdpE-DNA, and PmrA-DNA (20,39,42), plus detailed information on the density orientations of the linkers connecting GlnR_DBD and GlnR_REC, we finally succeeded in docking and fitting four GlnR_RECs into the corresponding electron densities of the GlnR-TAC structure (Fig. 2 B and C and SI Appendix, Fig. S11).The overlapping density maps of different GlnR_RECs indicate that GlnR protomers are prone to tetramerize through possible salt bridges and hydrogen bonds formed by the polar residues from adjacent GlnR_RECs.These include residues like T99, E103, and R107 from GlnR IV _ REC and residues R58, R62, and D88 from GlnR I _REC.It is worth noting that GlnR I _REC and GlnR II _REC collaborate and partially contact αNTD and β flap, with the two buried surface areas of ∼80 Å 2 and ∼50 Å 2 (SI Appendix, Fig. S11).Several polar residues (R58, R62, and D88) from GlnR_REC might contribute to the
| 1,281 |
261296163
| 0 | 16 |
formation of this distinct interface, which serves as a bridge to connect the GlnR_DBDs to the RNAP core enzyme and stabilize GlnR_TAC.Consistently, this was later verified by the mutation experiments in which the GlnR mutants showed decreased activity in the formation of GlnR_TAC (SI Appendix, Fig. S9).This reveals that the four GlnR-RECs are required for stabilization of GlnR_TAC through synergistically bridging GlnR_DBDs to RNAP. Discussion The ability of bacteria to colonize, pervade, and abundantly reproduce in various changing environments depends on their finetuned adaptive responses.TCSs are regarded as one of the most elaborate and efficient ways for bacteria to transcriptionally adapt to environmental stressors (9,10,12).TCSs are ubiquitous in archaea, bacteria, lower eukaryotic organisms, and plants but absent in mammals, making bacterial TCSs promising targets for drug discovery and design (43).The OmpR/PhoB subfamily transcription factors in bacteria have been extensively studied and are known as typical RRs involved in primary and secondary metabolisms.However, most studies have either focused on the genetic characterization of the target promoter regulons or on the biochemical and crystal structural studies of the OmpR/PhoB subfamily RRs alone (DrrB, DrrD, RegX3, PrrA, MtrA, etc.) (44)(45)(46)(47)(48) or in complex with their cognate DNA (PhoB_DBD-DNA, fulllength KdpE and DNA, full-length PmrA and DNA, etc.) (35,36,42).Until now, the overall blueprint of how the OmpR/ PhoB subfamily RRs activate transcription.By combining our biochemical and structural data of GlnR-TAC, we have achieved an in-depth understanding of the above long-standing questions.On the one hand, we found that four GlnR protomers engage promoter DNA in GlnR-TAC predominantly via the interactions between
| 1,282 |
261296163
| 0 | 16 |
GlnR_DBDs and the corresponding cis elements.There are also many auxiliary interactions between two adjacent GlnR_ DBDs, as identified in the co-crystal structure of the saeGlnR_ DBD and DNA (Figs. 1 and 3 and SI Appendix, Figs.S7 and S8).These results are in good agreement with the high binding affinity between GlnR and DNA identified in the EMSA experiment (SI Appendix, Fig. S9).This distinct head-to-tail mode of DNA engagement not only expands our understanding of how transcription factors recognize promoters but also provides a general model for the OmpR/PhoB subfamily response regulators.On the other hand, complex protein-protein interactions between different GlnR molecules and the conserved domains of RNAP efficiently facilitate transcription activity and the formation of GlnR-TAC (Figs. 4 and 5 and SI Appendix, Figs.S7 and S9).Consistent with the previously reported transcription factors (38), GlnR I _DBD and GlnR II _DBD interact with the β flap and σ A R4. Interestingly, we also found that αCTD simultaneously contacts both the upstream DNA and GlnR IV _DBD, and four GlnR_ RECs collaborate and serve as a bridge to connect the GlnR_ DBDs to the αNTD, β flap domains of the RNAP core enzyme.These findings provide direct evidence for the versatility of the highly conserved domains of RNAP (especially the whole α subunit) in bacterial transcription initiation, showing how they act as specialized platforms to interact with various transcription factors.Notably, this unique architecture simultaneously covers the typical features of the reported canonical class I and class II activators-the upstream two GlnR protomers finely contact α subunit and promoter DNA like
| 1,283 |
261296163
| 0 | 16 |
class I activators (28,34), while the downstream two GlnR protomers deftly engages its cis-element DNA, and the conserved β flap, σ A R4 of RNAP like class II activators (27,29,31,33).Thus, these biochemical and structural data provide insights into the underlying transcription activation mechanisms for the OmpR/PhoB subfamily proteins and reveal a unique mode of bacterial transcription regulation. Although the GlnR cis-elements have been studied for many years, the GlnR-binding consensus sequences seem to be more complex and diverse than previously expected, especially concerning the number and importance of different GlnR binding sites.One typical GlnR-binding consensus element, which includes four GlnR binding sites (a1, b1, a2, and b2 sites) separated by six bases, was first characterized as GlnR cis-element based on a dozen of studies (21).Later, Wang et al. amended this, determining that only three of the four GlnR binding sites are essential for GlnR binding to DNA in vitro and gene transcription regulation in vivo, while the a2 site is unnecessary (22,23).In line with these, various types of interactions between GlnR-DBDs and the typical GlnR binding sites have been extensively verified in our structure of GlnR-TAC (Figs. 3-5 and SI Appendix, Figs.S7-S9).Based on our cryo-EM structure, we surmise that the dispensability of the a2 site is due to being not only engaged by GlnR II _DBD but also stabilized by σ A R4.When the a2 site is mutated, the σ A R4 domain may still recognize the nonoptimal −35 element DNA and strengthen the structure of GlnR-TAC in combination with the upstream GlnR-DNA interactions.However, the a2
| 1,284 |
261296163
| 0 | 16 |
site might also play a positive regulatory role in recruiting GlnR to DNA consensus sites ahead of the RNAP engagement. As to the kinetic activation mechanism of GlnR, a previous biochemical investigation of GlnR in A. mediterranei proposed that the protein might activate transcription via two complexes.Complex I is formed by GlnR binding to both the a1 and b1 sites of the GlnR-dependent operon, which subsequently promotes formation of a transcription-competent complex II by engaging the other GlnR binding sites (23).This regulation mode is quite similar to that of the LysR-type transcription regulators like LysR, CatR, and CbbR.Once a dimer binds to the regulatory binding site, these regulators first form complex I; then, the other dimer binds to the activation binding site and retains it in a tetramer state through N terminal dimerization domains and assembles into an active complex II to enhance transcription (23,49).Consistent with the above hypothesis, our cryo-EM map visualizes a structure of GlnR-TAC with four GlnR protomers included.Moreover, our single-molecule FRET assays observed four-step photobleaching events of the fluorescently labeled GlnR, indicating that the stoichiometry of GlnR involved in M. tuberculosis GlnR-TAC is ~4.0 (SI Appendix, Materials and Methods and Fig. S12).Both these and the transcription assays (SI Appendix, Fig. S3C) show that four GlnR protomers structurally and kinetically promote the formation of a more stable and competent GlnR-TAC, supporting a predominant active conformation for GlnR protomers.This is also in good agreement with our unsuccessful trials in obtaining another cryo-EM structure of GlnR-TAC using amtB promoter containing six GlnR binding sites (a3-b3,
| 1,285 |
261296163
| 0 | 16 |
a1-b1, and a2-b2 sites), which showed that the same four GlnR protomers bound to the downstream a1-b1 and a2-b2 sites as in our GlnR-TAC.This lower binding affinity of GlnR to the upstream a3-b3 sites could be explained by the competitive binding activities between GlnR and PhoP (a global regulator for phosphate metabolism) uncovered by Wang et al. (50).These relationships may play physiological roles in coordinating and responding to nitrogen and phosphate availability in bacteria.Taking these findings into consideration, this study reveals that GlnR-dependent transcription activation is established as a unique and an elaborate transcription activation mechanism.Upon stimuli, the GlnR protomers tetramerize and engage promoter DNA in a head-to-tail manner and then assemble into a competent GlnR-TAC (Fig. 5E), which may be the most optimal form for GlnR to fulfill its physiological function.The existence of an extra a3-b3 site may not be predominantly regulated by GlnR; it may instead act as an alternative transcription regulation strategy to promptly respond to environmental nitrogen availability and orchestrate cross-talk with the essential phosphate metabolism.However, this needs to be further explored both in vitro and in vivo. Purification of M. tuberculosis GlnR and S. erythraea GlnR-DBD. Plasmids of pET28a-his TEV-glnR or pET28a-his TEV-glnR derivative were first transformed into expression strain BL21(DE3) (Invitrogen, Inc.).Then, single colonies of the positive transformants were inoculated and amplified with 5 L LB broth supplemented with 50 μg/mL kanamycin at 37 °C with shaking.When OD 600 of the cultures reached about 0.8-1.0GlnR expression was induced by adding 0.5 mM IPTG, and cultures were incubated for another 15
| 1,286 |
261296163
| 0 | 16 |
h at 20 °C.After being harvested by centrifugation (5,500g; 15 min at 4 °C), the cell pellets were resuspended in 20 mL buffer A (20 mM Tris-HCl, pH 8.0, 0.2 M NaCl, and 5% glycerol) and lysed using an ATS AH-10013 cell disrupter (ATS, Inc.).After centrifugation at 13,000g for 30 min at 4 °C, the supernatant was loaded onto a 5-mL column of Ni-NTA agarose (Qiagen, Inc.) pre-equilibrated with buffer A. The column was washed with 25 mL buffer A containing 25 mM imidazole and eluted with 30 mL buffer A containing 200 mM imidazole.Subsequently, the elutes were concentrated and applied to a 120-mL HiLoad 16/600 Superdex 75 column (GE Healthcare, Inc.) equilibrated with buffer B (20 mM Tris-HCl, pH 8.0, 75 mM NaCl, and 5 mM MgCl 2 ), and the column was eluted with the same buffer for a column volume.The targeted fractions containing GlnR identified by SDS-PAGE were pooled and stored at -80 °C.Yield was ~3.0 mg/L, and purity was > 95%.GlnR derivatives or Saccharopolyspora erythraea GlnR-DBD (SaeGlnR-DBD) were prepared as described above.In order to remove the N-terminal his tag, the elutes of GlnR147 (his-TEV GlnR C60SL147C) from the Ni-NTA column were cleaved with recombinant tobacco-etch virus protease (Life Technologies) overnight at 4 °C, subsequently exchanged by buffer A, and applied onto a second Ni-NTA column to remove the residual his-tagged GlnR and protease.The flow-through sample from the second Ni-NTA column was finally concentrated and purified by HiLoad 16/600 Superdex 75 column as GlnR. Purification of M. tuberculosis RNAP.M. tuberculosis RNAP was prepared
| 1,287 |
261296163
| 0 | 16 |
from cultures of E. coli strain BL21(DE3) (Invitrogen, Inc.) cotransformed with plasmids of pACYC Duet-rpoA-rpoD, pCDF-rpoZ, and pET Duet-rpoB-rpoC and purified as described with some modifications (53).Single colonies of the resulting transformants were used to inoculate 100 mL LB broth containing 35 μg/mL chloramphenicol, 100 μg/mL ampicillin, and 50 μg/mL streptomycin, and cultures were incubated for 16 h at 37 °C with shaking.Subsequently, cultures were amplified by transferring every 10 mL aliquot into 1 L LB broth containing the same antibiotics and incubated at 37 °C with shaking.When OD 600 reached ~0.8 to 1.0, cultures were induced by the addition of 0.5 mM IPTG and incubated for 15 h at 20 °C.Then, cells were harvested by centrifugation (5,000g; 15 min at 4 °C), resuspended in 20 mL lysis buffer A, lysed using a JN-02C cell disrupter (JNBIO, Inc.), and centrifugated at 13,000g for 30 min under 4 °C condition.Then, the supernatants were precipitated by poly(ethyleneimine) to a ratio of 0.7% (m/v), washed three times by buffer C (10 mM Tris-HCl, pH 8.0, 0.5 M NaCl, 1 mM EDTA, and 5% glycerol), extracted by buffer D (10 mM Tris-HCl, pH 8.0, 1.0 M NaCl, 1 mM EDTA, and 5% glycerol), and subsequently by precipitated by ammonium sulfate to a ratio of 30.0%(m/v).The pellets were resuspended with buffer A and loaded onto a 10-mL column of Ni-NTA agarose (Qiagen, Inc.) equilibrated with buffer A. The column was washed with 50 mL buffer A containing 20 mM imidazole and eluted with 50 mL buffer A containing 0.3 M imidazole.The
| 1,288 |
261296163
| 0 | 16 |
eluate was diluted with buffer E (20 mM Tris-HCl, pH 7.9, 5 % glycerol, 1 mM EDTA, and 1 mM DTT) and loaded onto a Mono Q 10/100 GL column (GE Healthcare, Inc.) equilibrated in buffer E and eluted with a 160 mL linear gradient of 0.3 to 0.5 M NaCl in buffer E. Fractions containing M. tuberculosis RNAP were pooled and applied to a 120-mL HiLoad 16/600 Superdex 200 column (GE Healthcare, Inc.) equilibrated in buffer B, and the column was eluted with the same buffer.Fractions containing M. tuberculosis RNAP were pooled and stored at -80 °C.Yield was ~0.5 mg/L, and purity was >95%. Crystallization of SaeGlnR-DBD Complexed with Its Cis-Element DNA. To assemble the binary complex of SaeGlnR-DBD bound to its cis-element DNA, two strands of a designated conserved cis-element DNA (21 bp long, 5′-ACGTAACATCGCGGTAAC A-3′) consisting of two copies of GlnR binding sites (a site and b site shown in bold) (Fig. 1A) were synthesized and gel purified by Sangon Biotech, Inc.First, template-strand oligonucleotide and non-templatestrand oligonucleotide of this blunted DNA were dissolved into nuclease-free water to 1 mM and annealed at the ratio of 1:1 in the annealing buffer (10 mM Tris-HCl, pH 7.9, and 0.2 M NaCl).SaeGlnR_DBD and DNA were incubated in a molar ratio of 2: 1.1 at 4 °C for 1 h.Then, the sample was centrifugated and applied to robotic crystallization trials by using a Gryphon liquid handling system (Art Robbins Instruments), commercial screening solutions (Emerald Biosystems, Hampton Research, and Qiagen), and the sitting-drop vapor-diffusion technique (drop: 0.2 µL
| 1,289 |
261296163
| 0 | 16 |
sample mixed with 0.2 µL screening solution; reservoir: 60 µL screening solution; 22 °C).A total of 1,000 conditions were screened, and brickshaped crystals appeared within 1 wk.Through optimization by hanging-drop vapor-diffusion technique at 16 °C, high-quality crystals were obtained under the condition of 0.1 M Bis-Tris, pH 5.7, and 25% PEG3350 in 1 wk.Crystals were transferred to a reservoir solution containing 18% (v/v) (2R,3R) -(-)-2,3-butanediol (Sigma-Aldrich) and flash-cooled with liquid nitrogen. Crystal Data Collection and Structure Determination.By using the cryocooled crystals, diffraction data of SaeGlnR-DBD complexed with DNA were collected at Shanghai Synchrotron Radiation Facility beamline 17U, followed by data processing with HKL2000 (54).Finally, the structure was solved by molecular replacement with Molrep (55) using the structure of the PhoB effector domain in complex with pho box DNA (PDB 1GXP) (37) as the search model.The model of SaeGlnR-DBD complexed with DNA was built in Coot (56) and refined in Phenix (57).Structure data collection and refinement statistics are listed in SI Appendix, Table S1. Assembly of M. tuberculosis GlnR-TAC.DNA oligonucleotides used to assemble M. tuberculosis GlnR-TAC were synthesized and gel purified by Sangon Biotech, Inc. Template-strand DNA (amtB scaffold_T) and non-template-strand DNA (amtB scaffold_NT) of this amtB scaffold (Fig. 2A) were first dissolved into nuclease-free water to 1 mM and annealed at the ratio of 1:1 in the annealing buffer (10 mM Tris-HCl, pH 7.9, and 0.2 M NaCl).Then, M. tuberculosis GlnR-TAC was assembled by incubating M. tuberculosis RNAP, amtB scaffold, and M. tuberculosis GlnR in a molar ratio of 1:1:10 at 4 °C overnight.The sample was
| 1,290 |
261296163
| 0 | 16 |
then applied onto a 120-mL HiLoad 16/600 Superdex 200 column (GE Healthcare, Inc.) equilibrated in buffer B, and the column was eluted with the same buffer.The elutes were identified by SDS-PAGE and electrophoretic mobility shift assay (EMSA).Finally, fractions containing right assembled M. tuberculosis GlnR-TAC were concentrated to 40 mg/ mL using Amicon Ultra centrifugal filters (10 kDa MWCO, Merck Millipore, Inc.).Sequences of the above amtB oligonucleotides are shown in SI Appendix, Table S3. Cryo-EM Grid Preparation.C-flat grids (CF-1.2/1.3 400 mesh holey carbon grids; Protochips, Inc.) were glow discharged for 60 s at 15 mA.After being incubated with CHAPSO to a final concentration of 8 mM (Hampton Research Inc.), 3 μL of the purified M. tuberculosis GlnR-TAC was applied onto the grids, blotted with Vitrobot Mark IV (FEI), and immediately plunge frozen in liquid ethane with 95% chamber humidity at 10°C. Cryo-EM Data Collection and Processing.Cryo-EM data of M. tuberculosis GlnR-TAC were collected by using a 300-kV Titan Krios (FEI, Inc.) equipped with a K3 Summit direct electron detector.A total of 1,200 images were recorded with EPU in counting mode with a pixel size of 1.2 Å, a dose rate of 10 e/pixel/s, and an electron exposure dose of 50 e/Å 2 .Movies were recorded for 8.38 s, and defocus range varied between −1.4 μm and −2.2 μm.Subframes of individual movies were aligned using MotionCor2 (58), and contrast transfer function for each summed image was estimated using CTFFIND4.From the summed images, approximately 10,000 particles were manually picked and subjected to 2D classification in RELION 3.1 (59).The corresponding
| 1,291 |
261296163
| 0 | 16 |
distinct 2D classes were used as templates for particle autopicking.A total of 202,298 particles were autopicked, manually inspected, and subjected to further 2D classification by specifying 100 classes.By removing the poorly populated classes, 118,770 particles were subjected to 3D classification in RELION by using a map of M. tuberculosis RPo (PDB ID: 6VVY) (60) low-pass filtered to 40 Å resolution as a reference.First, three classes were obtained after autopicking, 2D classification, and 3D classification in Relion 4.0.Then, the particles from each class were reprocessed by 3D classification by dividing into two classes.From the density maps, we can see that two classes (41.8%, 58.2%) from the first resulted class 1 (13.8%)exhibit much shorter DNA contours than class 2 (62.6%) which possibly allow GlnR protomers to bind.Though Class 1 (58.7%)processed from the first resulted class 3 (23.6%)presented longer DNA, the density map of GlnR protomers is very weak.Meanwhile, the outline of the main body of class 2 (41.3%) processed from class 3 does not seem to be a correctly assembled RNAP holoenzyme.Though class 2 (26.8%) resulted from the first resulted class 2 (62.6%) showed longer DNA and extra density below the DNA, they are unable to fit full-length GlnR molecules.Therefore, the classes mentioned above were ignored, and only the class 1 (73.2%)processed from the first resulted class 2 (62.6%) was further sequentially processed by 3D autorefinement.The selected 54,432 particles were further reextracted, CTF refined, Bayesian polished, 3D autorefined, and postprocessed in RELION (SI Appendix, Fig. S5).The gold standard Fourier shell correlation analysis indicated a mean map resolution of
| 1,292 |
261296163
| 0 | 16 |
3.70 Å of M. tuberculosis GlnR-TAC (SI Appendix, Fig. S6). Model Building and Refinement.The model of M. tuberculosis RPo (PDB ID: 6VVY) (60) and the co-crystal structure of SaeGlnR_DBD bound to its cis-element DNA were manually fitted into the cryo-EM density map of GlnR-TAC by Chimera (61).Data were further calculated and validated by rigid body and real-space refinement in Coot and Phenix.Structures were analyzed with PyMOL (62) and Chimera. In Vitro Transcription Assay.In vitro transcription assays were performed in transcription buffer (40 mM Tris-HCl, pH 8.0, 50 mM NaCl, 10 mM MgCl 2 , and 5% glycerol) by using 96-well microplates (Corning incorporated, USA).Reaction mixtures (80 μL) contained: 0.1 μM M. tuberculosis RNAP, 30 nM mango-ended DNA, 0 or 8 or 16 μM M. tuberculosis GlnR or its derivatives, 0.1 mM NTP mix (ATP, UTP, GTP, and CTP), and 1 μM TO1-biotin.Once M. tuberculosis RNAP and DNA were incubated for 10 min at 37 °C, M. tuberculosis GlnR was added into the mixture and incubated for 20 min at 37 °C.Subsequently, NTP mix and TO1biotin were added, and the mixture was incubated for another 30 min at 37 °C.Finally, fluorescence emission intensities were measured using a multimode plate reader (EnVision, PerkinElmer Inc.; excitation wavelength = 510 nm; emission wavelength = 535 nm).Relative transcription activities of GlnR derivatives were calculated using according to the procedure of the manufacturer.The ratio of GlnR-TAC formation of different GlnR mutants was quantitatively analyzed by the software Image J. Qualification and Statistical Analysis.The Fourier shell correlation (FSC) cutoff criterion of 0.143 (63)
| 1,293 |
261296163
| 0 | 16 |
was used to calculate FSCs of GlnR-TAC (SI Appendix, Fig. S6 B and E).To assess the data of transcription assays (Figs.3G and 4D), the mean values and their corresponding SE from three independent measurements were analyzed and displayed.The local resolution of the cryo-EM maps (SI Appendix, Fig. S6 C and D) was estimated using blocres (64).PHENIX (57) was also used for quantification and statistical analyses during model refinement and validation of GlnR-TAC. Fig. 1 . Fig. 1.The co-crystal structure of SaeGlnR bound to the designed conserved consensus promoter DNA.(A) Domain architectures of M. tuberculosis GlnR (MtbGlnR), M. smegmatis GlnR (MsmGlnR), S. coelicolor GlnR (ScoGlnR), and S. erythraea GlnR (SaeGlnR) (Top); the sequences of designed conserved consensus promoter DNA for SaeGlnR protein with the a-and b-site sequences highlighted in orange and blue, respectively (Bottom).Domains of REC (N-terminal receiver domain) and DBD (C terminal DNA binding domain) are individually labeled above the architectures.(B) Co-crystal structure of SaeGlnR I _DBD, SaeGlnR II _DBD in complex with the designed conserved consensus promoter DNA.SaeGlnR I _DBD and SaeGlnR II _DBD are represented as orange and blue cartoon, respectively.NT, non-template-strand promoter DNA (in dark red cartoon); T, template-strand promoter DNA (in red cartoon).(C) Detailed interactions between SaeGlnR and the designed conserved consensus promoter DNA.Residues from SaeGlnR involved in interacting with the a-site or b-site of promoter DNA are shown in green sticks, DNA sequences are shown in red sticks.(D) Summary of the interactions between the promoter DNA and SaeGlnR.Solid black lines and dashed black lines indicate hydrogen bond interactions and van der
| 1,294 |
261296163
| 0 | 16 |
Waals interactions between the involved GlnR residues and the GlnR binding sites, respectively. Fig. 2 . Fig. 2. The overall structure of M. tuberculosis GlnR-TAC.(A) DNA scaffold used in structure determination of M. tuberculosis GlnR-TAC (Top).NT, non-templatestrand promoter DNA; T, template-strand promoter DNA.GlnR binding box is framed in yellow color, with the a1 site, b1 site, a2 site, and b2 site shaded in orange, cyan, light orange, and blue colors, respectively.The −35 element and −10 element are enclosed with pink and brown frames, respectively.(B and C) Two views of the cryo-EM density map (B) and structure model (C) of M. tuberculosis GlnR-TAC.The cryo-EM density maps and cartoon representations of M. tuberculosis GlnR-TAC are colored as indicated in the color key.DNA is colored as A. The main body of M. tuberculosis RNAP is colored in light gray except αCTD in magenta. Fig. 3 . Fig. 3. Four GlnR molecules engage promoter DNA in M. tuberculosis GlnR-TAC.(A) Relative locations of M. tuberculosis GlnR I _DBD, GlnR II _DBD, GlnR III _DBD, and GlnR IV _DBD located at the upstream double-stranded DNA.(B) Detailed interactions between M. tuberculosis GlnR II _DBD, GlnR I _DBD, and their corresponding GlnR binding sites.The key residues involved are shown as green spheres.(C and D) The relative locations and detailed interactions between GlnR II _DBD and GlnR I _DBD bound to the promoter DNA.The key residues involved in Gln-R II _DBD and GlnR I _DBD are shown as wheat and blue spheres, respectively.(E) Substitutions of GlnR residues involved in promoter engagement reduced in vitro transcription
| 1,295 |
261296163
| 0 | 16 |
activity.Colors are shown as in Fig. 2. Fig. 4 . Fig. 4. The critical protein-protein interactions between M. tuberculosis GlnR and domains of RNAP β flap, σ A R4. (A) Relative locations of M. tuberculosis GlnR I _DBD and RNAP β flap (Left and Middle).GlnR I _DBD and RNAP β flap are shown in surface style (Left).Residues involved between GlnR I _DBD and RNAP β flap are shown as slate (GlnR I _DBD) and cyan spheres (RNAP β flap) (Right).(B) Relative locations of M. tuberculosis GlnR II _DBD and RNAP σ A R4 (Left and Middle).Gln-R II _DBD and RNAP σ A R4 are shown in surface (Left).Residues involved in interactions between GlnR II _DBD and RNAP σ A R4 are shown as yellow (GlnR II _DBD) and wheat spheres (RNAP σ A R4) (Right).RNAP β flap is colored in cyan, and RNAP σ A R4 is colored in yellow.The other colors are shown as in Fig. 2. Fig. 5 . Fig. 5. RNAP αCTD simultaneously interacts with promoter DNA and GlnR IV _ DBD.(A) Relative locations of M. tuberculosis GlnR IV _DBD, DNA, and RNAP αCTD.RNAP αCTD is shown in surface with magenta color.(B) RNAP αCTD interacts with the a1 site of M.tuberculosis GlnR binding box (in stick).Key residues of RNAP αCTD involved in DNA interaction are shown as green spheres.(C) Relative locations of M. tuberculosis GlnR IV _DBD and RNAP αCTD shown in cartoon.(D) RNAP αCTD interacts with GlnR IV _DBD.Key residues of RNAP αCTD and GlnR IV _DBD involved are shown as magenta
| 1,296 |
261296163
| 0 | 16 |
and green spheres, respectively.Colors are shown as in Fig. 2. (E) Proposed working model for GlnR-dependent transcription activation.Four GlnR molecules structurally and kinetically activate the formation of a stable and competent GlnR-TAC.GlnR binding boxes are presented in dark blue. I WT − I 0 . where I WT and I are the fluorescence intensities in the presence of GlnR and GlnR derivatives; I 0 is the fluorescence intensity in the absence of GlnR.
| 1,297 |
261296163
| 0 | 16 |
3DVirtualReality ImplementationofTouristAttractionsBasedon the Deep Belief Neural Network In today’s society, information technology is widely used, and virtual reality technology, as one of the emerging frontier technologies, has entered a stage of rapid development. Virtual reality is the use of computer technology to simulate the real-life environment into a virtual simulation environment, with the help of special equipment to realize the natural interaction between users and technical environment, in which the tourism industry is the most widely used. In order to realize 3D virtual reality of tourist attractions and improve users’ immersive experience in the process of interaction, the deep belief neural network is introduced to realize the target recognition and reconstruction in virtual reality. *e results show that the algorithm has excellent performance in target recognition and target reconstruction, and deep belief networks improve the accuracy by 0.57% and 0.81% and the accuracy by 0.21% and 2.06%, respectively, compared with the current optimal algorithm in target recognition of 12 and 20 view regular projection images. Compared with the current optimal algorithm, deep belief networks are reduced by 0.2%, 3.7%, and 0.6%, respectively. *e accuracy index was increased by 2%, 0.1%, and 0.1%, respectively. *e above results show that the proposed algorithm based on the deep belief neural network can realize 3D virtual reality of complex scenes such as tourist attractions according to its excellent performance. Introduction Nowadays, the continuous breakthrough of information technology has brought great impact on people's production and life. e rapid development of virtual reality technology provides a new idea for the transformation
| 1,298 |
237582188
| 0 | 16 |
and development of traditional tourism form. Using this technology can give tourists an immersive and unique virtual reality experience. is is an emerging technology integrating intelligent pattern recognition, computer graphics, machine vision, and other disciplines [1]. Tourists can get an immersive experience interacting with the simulation environment without leaving home. However, in practical application, especially in the realization of 3D virtual reality of tourist attractions, there are three technical difficulties in complex scenes, including target recognition, target reconstruction, image display, and human-computer interaction [2]. Among them, target recognition and target reconstruction are the keys to the realization of virtual reality. When target recognition is carried out in tourist attractions, there will be many kinds of targets, easy occlusion, complex texture background, large gap in volume, and shape. Traditional recognition methods are easy to lose key information and reduce the accuracy of target detection. When reconstructing the target, the traditional structure from motion (SFM) [3] has poor robustness and adaptability, which will reduce the reconstruction accuracy of the target and directly affect the quality of user experience. ese also become the problems to be solved to improve the virtual reality implementation technology and user experience. In order to further understand the above problems, we refer to the research of the LSTM neural network algorithm based on particle swarm optimization in anomaly detection and the technology of sensor data and recursive network in 3D virtual reality. e conclusion is that the current deep confidence neural network algorithm can effectively solve this kind of problem. With the rapid development
| 1,299 |
237582188
| 0 | 16 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.